Technical Field
[0001]
The present invention relates to a method and device for inspecting a
crankshaft for use in automobile engines and others in the production process thereof.
In particular, the present invention relates to a method and device for inspecting a
crankshaft, which enable accurate detection of defects which occur partially in the
crankshaft, such as underfills and dent flaws, by discriminating these defects from
bending and torsion over an entire length of the crankshaft.
Background Art
[0002]
A crankshaft is produced by pressing a heated starting material with upper
and lower dies, and die-forging it to mold a forging including fins, thereafter
removing fins and applying shot-blasting. A crankshaft thus produced in such
production processes is subjected to machining so as to be properly assembled when
assembled into an automobile engine, etc.
[0003]
Figure 1 is a diagram to schematically show an example of crankshaft
(crankshaft for inline-four engine). Figure 1A is a front view of the crankshaft S
viewed from· its rotational center axis L, and Figure 1 B is a side view of the
crankshaft L viewed from a direction perpendicular to the rotational center axis L.
As shown in Figure 1, the crankshaft S includes: a plurality of pins S 1 for
attaching a connecting rod (not shown), which are provided at positions of
predetermined angles around the rotational center axis L; a plurality of journals S3;
2
and a plurality of arms S2 each linking a pin S 1 and a journal S3 which are adjacent
to each other. The arm S2 may include a counterweight for achieving rotational
balance. In the example shown in Figure 1, every arm S2 includes a counterweight.
The cross-sectional shape of the pin S 1 is a circle centering on a position displaced
from the rotational center axis L of the crankshaft S, and the cross-sectional shape of
the journal S3 is a circle centering on the rotational center axis L of the crankshaftS.
The cross-sectional shape of the arm S2 is a complicated shape which is bilaterally
symmetric or asymmetric.
[0004]
As described above, due to a complicated shape of a crankshaft, and due to
variations in the starting material size, unevenness of the starting material
temperature, and variations in forging operation, there may occur a defect called
underfill in which the starting material is not filled to each end of the die upon
forging, and may occur bending and torsion of the crankshaft over its entire length.
Moreover, there may occur dent flaws caused by contact with conveying equipment,
etc. during handling of the crankshaft. For this reason, in the production line of a
crankshaft, before subjecting it to machining, an actual shape of the crankshaft is
inspected by comparison with a criteria! shape to make pass/fail judgment.
[0005]
The criteria for pass/fail judgment of crankshaft includes:
(a) bending and torsion of the crankshaft is within a predetermined tolerance
range, and
(b) there is neither underfill nor dent flaw having a depth that disables to
ensure a sufficient machining stock.
3
The reason why the condition that bending and torsion of the crankshaft is
within a predetermined tolerance range is one of the pass/fail judgment criteria as
described in item (a) above is that if the bending of the crankshaft is large, or the
torsion thereof is large so that the placing position of the pin is significantly deviated
from the predetermined angle, it becomes difficult to achieve dimensional accuracy
and weight balance as a final product regardless of what kind of processing is applied
in subsequent processes.
Further, the reason why the absence of underfill and dent flaw, each having a
depth that disables to ensure a sufficient machining stock, is one of the pass/fail
judgement criteria, as described in item (b) above is that if the machining stock is too
small, there is little margin for machining in subsequent processes, and it becomes
difficult to achieve dimensional accuracy and weight balance as a final product.
[0006]
A conventional method for inspecting a crankshaft is performed in such a way
that each plate gauge which is formed to correspond to criteria! shapes of a pin and
an arm is put onto the pin and the arm of the crankshaft to be inspected, and a gap
between each plate gauge and the pin and arm is measured with a scale to judge that
the crankshaft has passed when a size of the gap (shape error) is within a tolerance
range. This method has a problem in that since the method is performed manually
by an operator by using a plate gauge which is formed to correspond to criteria!
shapes of the" pin and arm, not only personal difference occurs in inspection accuracy,
but also inspection requires a significant amount of time. For this reason, various
inspection methods of a crankshaft have been proposed to automatically perform
accurate inspection.
[0007]
4
Patent Literature 1 proposes a method of calculating a longitudinal direction
size of a predetermined portion of a crankshaft based on a detection result by a onedimensional
image sensor, in which the one-dimensional image sensor is disposed on
one side of the crankshaft such that the arrangement direction of its light receiving
elements corresponds to a direction perpendicular to the longitudinal direction of the
crankshaft, and a light source is disposed on the other side of the crankshaft, and in
which the one-dimensional image sensor is moved along the longitudinal direction of
the crankshaft.
Since the method according to Patent Literature 1 calculates a longitudinal
direction size of a predetermined portion of a crankshaft, bending may be calculable;
however, it is not possible to detect a partial defect such as an underfill, and torsion.
[0008]
Patent Literature 2 proposes a method of calculating angular positions of a pin
and a counterweight of a crankshaft by measuring a distance to the crankshaft surface
using a laser range meter while the crankshaft is rotated around the rotational center
axis with each end thereof being secured with a chuck.
Patent Literature 3 proposes a method of detecting an underfill of a
counterweight by measuring a distance to a counterweight of the crankshaft by a
two-dimensional laser range meter and comparing with a criteria! shape.
The method according to Patent Literature 2 can detect torsion of a crankshaft,
and the metli:od according to Patent Literature 3 can detect underfills. However,
since these methods use a one-dimensional laser range meter (Patent Literature 2)
and a two-dimensional laser range meter (Patent Literature 3), it takes a significant
amount of time to measure a distance along an entire length of a crankshaft.
Therefore, it is difficult to perforin inspection over the entire length of a crankshaft in
5
the production process of crankshaft, and it requires a sampling inspection, or an
inspection limited to a minimum number of areas of the crankshaft where inspection
Is necessary.
[0009]
Patent Literature 4 proposes a method for inspecting a crankshaft, in which a
surface shape of an entire crankshaft is measured by a three-dimensional shape
measurement device, whereby inspection is performed based on whether or not a
three-dimensional model for judgment satisfies a predetermined criterion, where the
three-dimensional model for judgement is obtained by supplementing portions where
measurement is impossible with a three-dimensional model for supplementation.
When judging whether or not the predetermined criterion is satisfied by the
method according to Patent Literature 4, it is conceivable, for example, to match
three-dimensional point cloud data (three-dimensional model for judgment) basically
obtained by a three-dimensional shape measurement device with a surface shape
model of the crankshaft created from CAD data, etc. based on the design
specification of the crankshaft, and evaluate a deviation therebetween. However, it
is difficult to accurately discriminate whether the deviation has occurred caused by a
partial defect such as an underfill, or caused by bending over an entire length of the
crankshaft.
Citation List
Patent Literature
[0010]
Patent Literature 1: JP 859-184814 A
Patent Literature 2: JP H6-265334 A
Patent Literature 3: JP H10-62144 A
6
Patent Literature 4: JP 2007-212357 A
Summary of Invention
Technical Problem
[0011]
The present invention has been made to solve the problems of conventional
art as described above, and has its objective to provide a method and device for
inspecting a crankshaft, which can accurately detect defects which occur partially in
a crankshaft, such as underfills and dent flaws, by discriminating these partial defects
from bending and torsion over an entire length of the crankshaft.
Solution to Problem
[0012]
In order to solve the problems as described above, the present inventors have
conducted diligent research. As a result, the present inventors have found that if
three-dimensional point cloud data (aggregate of three-dimensional coordinates) of a
crankshaft surface obtained by a three-dimensional shape measurement device is
divided to create a plurality of subregion three-dimensional point cloud data in which
each of the subregion three-dimensional point cloud data respectively corresponds to
each of a plurality of subregions of the crankshaft, and each of the three-dimensional
point cloud data is matched with a surface shape model created based on a design
specifications, then it becomes possible to discriminate defects that occur partially in
a crankshaft, , such as underfills and dent flaws, from bending and torsion over an
entire length of the crankshaft, thereby to accurately detect these partial defects.
Thus, the present inventors have completed the present invention.
[0013]
7
Thus, in order to solve the problems as described above, the present invention
provides a method for inspecting a crankshaft, which includes the following first to
fifth steps.
(1) First step: disposing an optical three-dimensional shape measurement
device to face a crankshaft in a direction perpendicular to a rotational center axis of
the crankshaft, in which the optical three-dimensional shape measurement device is
configured to measure a three-dimensional shape of a measurement object by
projecting and receiving light to and from the measurement object.
(2) Second step: measuring a surface shape of the crankshaft by the threedimensional
shape measurement device disposed in the first step to acquire threedimensional
point cloud data of a surface of the crankshaft over an entire length of a
measurement object region of the crankshaft.
(3) Third step: dividing the three-dimensional point cloud data acquired in the
second step to create a plurality of subregion three-dimensional point cloud data,
each of the subregion three-dimensional point cloud data respectively corresponding
to each of a plurality of subregions of the crankshaft along a direction parallel to the
rotational center axis of the crankshaft.
(4) Fourth step: translating and rotating each of the subregion threedimensional
point cloud data created in the third step to superpose each of the
subregion three-dimensional point cloud data on a surface shape model of the
crankshaft prepared in advance based on a design specification of the crankshaft,
such that a distance between each of the subregion three-dimensional point cloud
data and the surface shape model becomes minimum.
(5) Fifth step: calculating a distance between each of the subregion threedimensional
point cloud data after being superposed in the fourth step and the surface
8
shape model, and detecting a partial defect such as an underfill of the crankshaft
based on the distance calculated.
[0014]
According to the present invention, performing the first step and the second
step enables three-dimensional point cloud data of a crankshaft surface over an entire
length of a measurement object region to be obtained (for example, a region
spanning over arms located at each end of the crankshaft).
Then, performing the third step and the fourth step causes each of a plurality
of divided subregion three-dimensional point cloud data to be superposed on a
surface shape model individually (each of the subregion three-dimensional point
cloud data is translated and rotated such that a distance between each of the
subregion three-dimensional point cloud data and the surface shape model becomes
minimum). For this reason, compared with a case where the three-dimensional
point cloud data is superposed without being divided (undivided three-dimensional
point cloud data is translated and rotated such that the distance between the
undivided three-dimensional point cloud data and the surface shape model becomes
minimum), the subregion three-dimensional point cloud data can be superposed on
the surface shape model with influences of the bending and torsion over the entire
length of the crankshaft being reduced.
Thus, in the fifth step, by calculating the distance between each of the
subregion thiee-dimensional point cloud data after being superposed in the fourth
step and the surface shape model, based on the calculated distance, it is possible to
discriminate a partial defect such as an underfill of the crankshaft from bending and
torsion over the entire length of the crankshaft to accurately detect the partial defect.
[0015]
9
It is noted that "dividing three-dimensional point cloud data, each of the
subregion three-dimensional point cloud data respectively corresponding to each of a
plurality of subregions of the crankshaft" in the present invention includes both cases
where adjacent subregions of a crankshaft have and do not have an overlapped
portion along a direction parallel to the rotational center axis of the crankshaft. In
either case, if a plurality of divided subregions are combined, thus combined region
corresponds to the measurement object region of the crankshaft from which threedimensional
point cloud data have been acquired.
Moreover, "a distance between each of subregion three-dimensional point
cloud data and a surface shape model of a crankshaft becomes minimum" in the
present invention means that a total sum of distances between each data point
constituting each of the subregion three-dimensional point cloud data and the surface
shape model becomes minimum, or a total sum of squares of the distances becomes
minimum.
[0016]
A larger measurement field of v1ew of the three-dimensional shape
measurement device will result in a larger region of the crankshaft where the surface
shape can be measured at one time, and is therefore preferable in terms of reduction
of measurement time. However, in general, measurement resolution will decline as
the measurement field of view of the three-dimensional shape measurement device is
increased. 6n the other hand, installing a plurality of three-dimensional shape
measurement devices having a high measurement resolution (a small measurement
field of view) in a direction parallel to the rotational center axis of the crankshaft will
cause increase in cost and degradation of maintainability.
[0017]
10
Thus, preferably, in the present invention, in the second step, measurement of
the surface shape of the crankshaft by the three-dimensional shape measurement
device and relative movement of the three-dimensional shape measurement device in
a direction parallel to the rotational center axis of the crankshaft are alternately
repeated.
[0018]
As in the above-described preferred method, repeating the measurement by
the three-dimensional shape measurement device and the relative movement of the
three-dimensional shape measurement device will have advantages of improving
measurement resolution as well as mitigating increase in cost and decline of
maintainability.
[0019]
Preferably, the method according to the present invention further includes a
sixth step of evaluating bending and torsion of the crankshaft based on a translating
distance and a rotational angle when each of the subregion three-dimensional point
cloud data is translated and rotated in the fourth step.
[0020]
According to the above-described preferred method, it is possible not only to
detect a partial defect such as an underfill of a crankshaft, but also to evaluate
bending and torsion of a crankshaft. Specifically, it is possible to evaluate bending
by adding up translating distances of each of the subregion three-dimensional point
cloud data, and to evaluate torsion by adding up rotational angles of each of the
subregion three-dimensional point cloud data.
[0021]
11
Preferably, in the present invention, in the third step, when a defect detection
accuracy required in the fifth step is± L'le [mm], a size Lp [mm] ofthe subregion of
the crankshaft, which is to be divided, in a direction parallel to the rotational center
axis of the crankshaft is determined so as to satisfy the following Formula (1):
Lp ~ 2L • L'le I (L'lb +~a· R) ... (1)
where, in the above Formula (1), L [mm] refers to an entire length in a design
specification of the crankshaft; R [ mm] refers to a maximum radius in a design
specification of the crankshaft; ~a [rad] refers to an assumed torsion angle over an
entire length of the crankshaft; and ~b [ mm] refers to an assumed bending over an
entire length of the crankshaft.
[0022]
When an assumed bending over the entire length of the crankshaft is ~b [mm],
and the entire length in the design specification of the crankshaft is L [ mm], it can be
supposed that a bending per size Lp [mm] is ~b • Lp I L [mm]. Further, when an
assumed torsion angle over the entire length of the crankshaft is ~a [ rad], the entire
length in the design specification of the crankshaft is L [mm], and a maximum radius
in the design specification of the crankshaft is R [mm], it can be supposed· that
torsion per size Lp [mm] is ~a • R • Lp I L [mm]. For this reason, it is considered
that satisfying the following Formula (2) enables detection of a partial defect such as
an underfill at a detection accuracy of± L'le [mm].
~b • Lp I L + ~a • R • Lp I L ~ 2 • L'le ... (2)
Modifying Formula (2) will result in Formula (1) described above.
Therefore, determining the size Lp of a divided subregion (size in a direction parallel
to the rotational center axis of the crankshaft) so as to satisfy Formula (1) will make
12
it possible to detect a partial defect such as an underfill at a detection accuracy of±
L\e [mm].
[0023]
It is noted that in the above-described preferred method, it is possible to
determine an assumed torsion angle L\a and an assumed bending L\b based on past
production results of a crankshaft of the same kind of the measurement object, the
requirement specification of delivery destination of the production, and the like (for
example, by using maximum values and average values of the torsion angle and the
bending in past production results).
The term "torsion angle" in the above-described preferred method means an
angle by which an arm provided on one end of a crankshaft is rotated with respect to
CAD data of an arm based on a design specification when the arm provided on the
other end is aligned to the CAD data of the arm.
The term "bending" in the above-described preferred method means a
distance by which one end of a crankshaft is displaced in a direction perpendicular to
a rotational center axis with respect to one end of CAD data of a crankshaft based on
a design specification when the other end of the crankshaft is aligned to the other end
of the CAD data of the crankshaft.
Further, "maximum radius in a design specification of a crankshaft" means a
maximum distance from a rotational center axis of a crankshaft to a crankshaft
surface (spedfically, an arm surface) in CAD data based on a design specification.
[0024]
Here, bending and torsion of a crankshaft predominantly takes place at
journals or pins. Thus, preferably, in the third step, the subregion of the crankshaft
is determined such that both ends of the subregion, which is to be divided, in a
13
direction parallel to the rotational center axis of the crankshaft are located at adjacent
journals of the crankshaft, adjacent pins of the crankshaft, or a journal and a pin
adjacent to each other of the crankshaft.
It is noted that when the size Lp of the subregion, which is to be divided, is
determined so as to satisfy Formula (1) described above, if Lp is made too small, an
accuracy of superposition deteriorates when superposing the subregion threedimensional
point cloud data on the surface shape model. For this reason, it is not
true that Lp is preferably as small as possible. It is preferable to determine Lp such
that Lp is equal to or higher than a minimum value among a distance between
adjacent journals, a distance between adjacent pins, and a distance between a journal
and a pin which are adjacent to each other.
[0025]
According to the above-described preferred method, smce dividing into
subregions is performed at positions where bending and torsion are likely to take
place, it is possible to superpose the subregion three-dimensional point cloud data on
the surface shape model in a state where influences of bending and torsion are
effectively reduced. For this reason, it is possible to discriminate partial defects
such as underfills of a crankshaft from bending and torsion over the entire length of
the crankshaft, thereby to accurately detect such partial defects.
[0026]
In the fifth step of the present invention, calculating distances between each of
the subregion three-dimensional point cloud data after being superposed in the fourth
step and the surface shape model is performed. In other words, a large number of
numerical information which consists of three-dimensional coordinates of each data
point constituting the subregion three-dimensional point cloud data, and distances
14
associated with each data point. It is not easy to detect a partial defect such as an
underfill of the crankshaft by using these large number of numerical information as
they are.
[0027]
Thus, preferably, in the present invention, in the fifth step, a two-dimensional
gradation image is created, in which a pixel constituting the two-dimensional
gradation image has a density corresponding to a distance between each of the
subregion three-dimensional point cloud data after being superposed in the fourth
step and the surface shape model, wherein the two-dimensional gradation image is a
projection of each of the subregion three-dimensional point cloud data after being
superposed in the fourth step to a plane perpendicular to a facing direction between
the three-dimensional shape measurement device and the crankshaft, and a partial
defect of the crankshaft is detected based on a feature quantity obtained by subjecting
the two-dimensional gradation image to predetermined image processing.
[0028]
According to the above-described preferred method, since the large number of
numerical information is transformed into a two-dimensional gradation image having
a pixel density according to a distance between the subregion three-dimensional
point cloud data after being superposed in the fourth step and the surface shape
model, it is possible to subject the two-dimensional gradation image to image
processing similar to that for a conventional general inspection method. It is
possible to automatically and easily detect a partial defect such as an underfill, for
example, by extracting a pixel region having a density higher than a predetermined
threshold (a pixel region having a large distance between superposed subregion
15
three-dimensional point cloud data and the surface shape model), and using a feature
quantity calculated from an area, a density and the like of the extracted pixel region.
[0029]
Preferably, the method according to the present invention further includes a
seventh step of: creating a two-dimensional image in which a pixel constituting the
two-dimensional image has a density or a color corresponding to a distance between
each of the subregion three-dimensional point cloud data after being superposed in
the fourth step and the surface shape model, wherein the two-dimensional image is a
projection of each of the subregion three-dimensional point cloud data after being
superposed in the fourth step to a plane perpendicular to a facing direction between
the three-dimensional shape measurement device and the crankshaft; and displaying
the two-dimensional image.
[0030]
According to the above-described preferred method, a two-dimensional image
is created and displayed, in which a pixel region corresponding to a larger distance
between each of the superposed subregion three-dimensional point cloud data and the
surface shape model has a density or a color different from that of surrounding pixel
regions. As a result, it is possible to easily detect a pixel region corresponding to a
large distance between each of the superposed subregion three-dimensional point
cloud data and the surface shape model, that is, a partial defect such as an underfill,
by an operator visually recognizing the two-dimensional image.
[0031]
In order to solve the above-described problems, the present invention also
provides a device for inspecting a crankshaft, which includes: an optical threedimensional
shape measurement device which is disposed to face a crankshaft in a
16
direction perpendicular to a rotational center axis of the crankshaft, and which is
configured to measure a three-dimensional shape of a measurement object by
projecting and receiving light to and from the measurement object; and a control
computing device configured to control an operation of the three-dimensional shape
measurement device and executing predetermined computing on a result measured
by the three-dimensional shape measurement device, in which: a surface shape
model of the crankshaft prepared based on a design specification of the crankshaft is
stored in advance in the control computing device; three-dimensional point cloud
data of a surface of the crankshaft over the entire length of a measurement object
region of the crankshaft acquired by the three-dimensional shape measurement
device measuring the surface shape of the crankshaft is input to the control
computing device; and the control computing device is configured to execute steps
of: dividing the three-dimensional point cloud data which is input, to create a
plurality of subregion three-dimensional point cloud data, each of the subregion
three-dimensional point cloud data respectively corresponding to each of a plurality
of subregions of the crankshaft along a direction parallel to the rotational center axis
of the crankshaft; translating and rotating each of the subregion three-dimensional
point cloud data to superpose each of the subregion three-dimensional point cloud
data on the surface shape model such that a distance between each of the subregion
three-dimensional point cloud data created and the surface shape model stored
becomes miril.mum; and calculating a distance between each of the subregion threedimensional
point cloud data after being superposed and the surface shape model,
and detecting a partial defect such as an underfill of the crankshaft based on the
distance calculated.
[0032]
17
Preferably, the device for inspecting a crankshaft according to the present
invention further includes a moving mechanism whose operation is controlled by the
control computing device and which is configured to relatively move the threedimensional
shape measurement device in a direction parallel to the rotational center
axis of the crankshaft, in which the control computing device is configured to control
the operation of the three-dimensional shape measurement device and the moving
mechanism such that measurement of the surface shape of the crankshaft by the
three-dimensional shape measurement device, and relative movement of the threedimensional
shape measurement device in a direction parallel to the rotational center
axis of the crankshaft by the moving mechanism are alternately repeated.
[0033]
As the three-dimensional shape measurement device, a three-dimensional
shape measurement device of a pattern projection type can be used.
[0034]
As a three-dimensional shape measurement device of pattern projection type,
there is known a device which can measure a surface shape of several hundreds mm
square in a time period of about one to two seconds, at a measurement resolution of
about 0.1 mm. Since a length of crankshaft is about 350 mm to 600 mm, as in the
above-described preferred device, performing measurement by relatively moving the
three-dimensional shape measurement device of pattern projection type in a direction
parallel to tne rotational center axis of the crankshaft will make it possible to
accurately measure the entire length by repeating measurement two to three times,
and provide an advantage that time required for measurement may be as short as 1 0
seconds or less.
[0035]
18
Preferably, the device for inspecting a crankshaft according to the present
invention includes four three-dimensional shape measurement devices disposed at a
pitch of 90° around the rotational center axis of the crankshaft, in which: the moving
mechanism can separately move the four three-dimensional shape measurement
devices in a direction parallel to the rotational center axis of the crankshaft; and the
control computing device can separately control timing of measuring the surface
shape of the crankshaft by the four three-dimensional shape measurement devices
and timing of moving the four three-dimensional shape measurement devices by the
moving mechanism.
[0036]
According to the above-described preferred device, since four threedimensional
shape measurement devices are disposed at a pitch of 90° around the
rotational center axis of the crankshaft, it is possible to measure the shape of the
entire length and the entire circumference of the measurement object region of the
crankshaft without relatively rotating the crankshaft in the circumferential direction,
and thus to reduce the measurement time. Moreover, according to the abovedescribed
preferred device, a moving mechanism can move the four threedimensional
shape measurement devices separately in a direction parallel to the
rotational center axis of the crankshaft, and a control computing device can
separately control the timing of measuring the surface shape of the crankshaft by the
four three-dimensional shape measurement devices and the timing of moving the
four three-dimensional shape measurement devices by the moving mechanism. For
this reason, it is possible to perform positional control of the three-dimensional shape
measurement device and timing control of measurement and movement of the threedimensional
shape measurement device, which can avoid a situation in which
19
projected light of any of the three-dimensional shape measurement devices comes
into the measurement field of view of another three-dimensional shape measurement
device different from the concerned three-dimensional shape measurement device
during measurement, thereby disabling the measurement by the concerned another
three-dimensional shape measurement device.
[0037]
More specifically and preferably, the control computing device is configured
to control the operation of the moving mechanism and the four three-dimensional
shape measurement devices such that while any one pair of three-dimensional shape
measurement devices disposed in a direction opposite to each other among the four
three-dimensional shape measurement devices are measuring the surface shape of the
crankshaft, any other pair of three-dimensional shape measurement devices disposed
in a direction opposite to each other are moved without performing measurement,
and such that projected light of one three-dimensional shape measurement device
constituting the pair of three-dimensional shape measurement devices measuring the
surface shape of the crankshaft does not enter a measurement field of view of the
other three-dimensional shape measurement device.
[0038]
According to the above-described preferred device, while any one pair of
three-dimensional shape measurement devices disposed in a direction opposite to
each other measures the surface shape of the crankshaft, any other pair of threedimensional
shape measurement devices disposed in a direction opposite to each
other will move without performing measurement. For this reason, projected light
from one pair of the three-dimensional shape measurement devices which are
measuring the surface shape of the crankshaft will not influence the other pair of the
20
three-dimensional shape measurement devices which are moving without performing
measurement. Moreover, according to the above-described preferred device, since
control is performed such that projected light of one three-dimensional shape
measurement device constituting one pair of three-dimensional shape measurement
devices which are measuring the surface shape will not enter the measurement field
of view of the other pair of three-dimensional shape measurement devices, it is
possible to avoid a situation in which the other pair of three-dimensional shape
measurement devices becomes unable to perform measurement.
Advantageous Effects of Invention
[0039]
According to the present invention, it is possible to discriminate defects that
partially occur in a crankshaft such as underfills and dent flaws from bending and
torsion over the entire length of the crankshaft and to accurately detect the defects
that partially occur in a crankshaft.
Brief Description of Drawings
[0040]
[Figure 1] Figure 1 is a diagram schematically showing one example of crankshaft (a
crankshaft for an inline-four engine).
[Figure 2] Figure 2 is a diagram showing a schematic configuration of an inspection
device of a crankshaft according to a first embodiment of the present invention.
[Figure 3] Figure 3 is a schematic diagram to explain a situation in which threedimensional
point cloud data is superposed on a surface shape model without being
divided, for a crankshaft having neither bending nor torsion.
21
[Figure 4] Figure 4 is a schematic diagram to explain a situation in which threedimensional
point cloud data is superposed on a surface shape model without being
divided, for a crankshaft having bending.
[Figure 5] Figure 5 is a schematic diagram to explain a situation in which divided
subregion three-dimensional point cloud data is superposed on a surface shape model
by using an inspection method according to the present invention for a crankshaft
having bending.
[Figure 6] Figure 6 is an explanatory diagram to illustrate one example of the method
of determining a subregion in the third step of an inspection method according to the
present invention.
[Figure 7] Figure 7 is a diagram to show an example of two-dimensional image
obtained when three-dimensional point cloud data is superposed, without being
divided, on a surface shape model for a crankshaft having underfill and torsion.
[Figure 8] Figure 8 is a diagram to show one example of two-dimensional image
obtained when divided subregion three-dimensional point cloud data is superposed
on a surface shape model by using an inspection method according to the present
invention, for a crankshaft having underfill and torsion.
[Figure 9] Figure 9 is an explanatory diagram to illustrate an example of translating
distance and rotational angle when subregion three-dimensional point cloud data is
translated and rotated by using an inspection method according to the present
invention, for a crankshaft having underfill and torsion.
[Figure 1 0] Figure 10 is a diagram showing a schematic configuration of an
inspection device of a crankshaft according to a second embodiment of the present
invention.
22
[Figure 11] Figure 11 is a diagram showing an example of results of evaluation of
surface-shape measurement time of a crankshaft by an inspection device according to
the second embodiment of the present invention.
Description of Embodiments
[0041]
Hereinafter, embodiments of the present invention will be described
appropriately referring to the appended drawings.
[0042]

Figure 2 is a diagram showing a schematic configuration of an inspection
device of a crankshaft (hereinafter, referred to simply as an "inspection device")
according to a first embodiment of the present invention. Figure 2A is a front view
of a crankshaft S viewed from a direction of a rotational center axis L of the
crankshaft S. Figure 2B is a side view seen from a direction perpendicular to the
rotational center axis L. In Figure 2A, the crankshaft S is shown in a transparent
view, and a control computing device 2 is omitted from illustration.
As shown in Figure 2, an inspection device 100 according to the present
embodiment includes an optical three-dimensional shape measurement device 1, a
control computing device 2, a moving mechanism 3, and a rotating device 4.
[0043]
The three-dimensional shape measurement device 1 is a device which
measures a three-dimensional shape of a measurement object (a crankshaft S in the
present invention) by projecting and receiving light to and from the measurement
object. Specifically, the three-dimensional shape measurement device 1 projects
light to the crankshaft S and receives light reflected at the surface of the crankshaft S,
23
to measure a surface shape of the crankshaft S. The three-dimensional shape
measurement device 1 is disposed to face the crankshaft S in a direction
perpendicular to a rotational center axis L of the crankshaft S (a vertical direction
which is the Z direction shown in Figure 2).
As the three-dimensional shape measurement device 1 of the present
embodiment, a three-dimensional shape measurement device of pattern projection
type is used. A three-dimensional shape measurement device of patter projection
type generally includes a pattern projector of liquid crystal type or DMD (digital
mirror device) type, and an imaging device. The three-dimensional shape
measurement device of patter projection type measures a surface shape of a
measurement object by using a principle of triangulation in which a fringe pattern is
projected from the pattern projector to a measurement object, and an image of the
measurement object to which the fringe pattern has been projected is taken by the
imaging device to analyze deformation of the fringe pattern. Preferably used is a
three-dimensional shape measurement device (for example, SD-3K supplied by
ShapeDrive GmbH) utilizing a spatial coding method, whereby a fringe pattern in
which a bright part and a dark part are alternately arranged at an arbitrary spacing is
projected to binary-encode a space.
As the three-dimensional shape measurement device 1 of the present
embodiment, a three-dimensional shape measurement device of pattern projection
type is used in which a measurement field of view is 200 mm (in a direction parallel
to the rotational center axis L of the crankshaft S, which is the X direction shown in
Figure 2) x 100 mm (in a direction perpendicular to the X direction and the Z
direction, which is the Y direction shown in Figure 2) x 80 mm (in the Z direction
shown in Figure 2) when the distance to the measurement object is 400 mm. The
24
measurement resolution in the X direction and the Y direction is 0.1 mm, and the
measurement resolution in the Z direction is 0.02 mm. The measurement time is
within 2 seconds.
[0044]
The control computing device 2 controls an operation of the threedimensional
shape measurement device 1, the moving mechanism 3 and the rotating
device 4, and executes predetermined computation on measurement results by the
three-dimensional shape measurement device 1. The control computing device 2
includes, for example, a personal computer in which a program or an application is
installed, the program and the application executing the above-described control and
computation.
Further, in the control computing device 2, a surface shape model of the
crankshaft S prepared based on a design specification of the crankshaft S is stored in
advance. Specifically, three-dimensional CAD data based on the design
specification is input to the control computing device 2, and the control computing
device 2 transforms the input CAD data into a surface shape model composed of a
triangular mesh. Then, the surface shape model is stored in the control computing
device 2. Since the surface shape model can be prepared and stored for each kind
of the crankshaft S, it is not necessary to prepare the surface shape model for each
inspection when successively inspecting the same kind of crankshafts S.
[0045]
The movmg mechanism 3 relatively moves the three-dimensional shape
measurement device 1 in a direction (the X direction shown in Figure 2) parallel to
the rotational center axis L of the crankshaft S. As the moving mechanism 3, for
example, a single-axis stage can be used. The single-axis stage to be used as the
25
moving mechanism 3 is preferably capable of positioning or indicating its location at
a resolution of 0.1 mm or less. It is noted that although the moving mechanism 3 of
the present embodiment is a mechanism to move the three-dimensional shape
measurement device 1, but is not limited thereto, the moving mechanism 3 may be a
mechanism configured to move the crankshaft S in the X direction.
[0046]
The rotating device 4 secures each end of the crankshaft S by chucking, and
rotates, thereby causing the crankshaft S to rotate around the rotational center axis L.
The rotating device 4 is preferably capable of rotationally positioning or indicating
rotational positions at a pitch of 0.1° or less, such as one which is rotated by a
stepping motor or one which is provided with a rotary encoder at the rotational center,
such that the rotational angle of the crankshaft S can be indicated.
[0047]
Hereinafter, an inspection method of a crankshaft S by using the inspection
device 100 having the above-described configuration will be described.
An inspection method according to the present embodiment is characterized
by including first to fifth steps. Hereinafter, each step will be successively
described.
[0048]
(1) First step
In the first step, the three-dimensional shape measurement device 1 is
disposed to face the crankshaft S in a direction (Z direction) perpendicular to the
rotational center axis L of the crankshaft S. Specifically, the three-dimensional
shape measurement device 1 is disposed to face the crankshaft S in the Z direction by
26
securing the crankshaft S to the rotating device 4 such that the rotational center axis
L thereof is horizontal.
[0049]
(2) Second step
In the second step, three-dimensional point cloud data of the surface of the
crankshaft S over the entire length of a measurement object region of the crankshaft
S is acquired by measuring the surface shape of the crankshaft S with the threedimensional
shape measurement device 1. Specifically, the control computing
device 2 controls an operation of the three-dimensional shape measurement device 1
and the moving mechanism 3 such that measurement of the surface shape of the
crankshaft S by the three-dimensional shape measurement device 1 and movement of
the three-dimensional shape measurement device 1 in the X direction by the moving
mechanism 3 are alternately repeated. That is, when measurement of the surface
shape of the crankshaft S in the above-described one measurement field of view is
finished by the three-dimensional shape measurement device 1, the threedimensional
shape measurement device 1 is moved in the X direction by the moving
mechanism 3, and measures the surface shape of the crankshaft S in the next
measurement field of view. A combined region of these plurality of measurement
fields of view extends over the entire length of the measurement object region ofthe
crankshaft S. Since the length of the crankshaft S is about 350 to 600 mm for a 3-
to 6-cylinder 'engine, and the field of view of the three-dimensional shape
measurement device 1 in the X direction is 200 mm, it is possible to acquire threedimensional
point cloud data of the surface of the crankshaft S over the entire length
of the measurement object region of the crankshaft S by repeating measurement two
to three times.
27
The three-dimensional point cloud data of the surface of the crankshaft S over
the entire length of the measurement object region of the crankshaft S, which has
been acquired as described above, is input into the control computing device 2 via
Ethernet (registered trademark) or the like and is stored.
[0050]
(3) Third step
In the third step, the control computing device 2 divides the three-dimensional
point cloud data which has been input and stored as described above, to create a
plurality of subregion three-dimensional point cloud data, in which each of the
plurality of subregion three-dimensional point cloud data corresponds to each of a
plurality of subregions of the crankshaft S along a direction (X direction) parallel to
the rotational center axis L. The method for determining the subregion will be
described later.
It is noted that the stored three-dimensional point cloud data is subjected as
needed to removal of isolated data points for reducing noise, and to thinning of data
points into a predetermined pitch (for example, thinning of the X direction and Y
direction into 0.5 mm pitch) for increasing processing speed. Furthermore, as
needed, the three-dimensional point cloud data after thinning is subjected to
smoothing processing for reducing noise. As needed, after application of these
signal processing, a plurality of subregion three-dimensional point cloud data is
created.
[0051]
(4) Fourth step
In the fourth step, the control computing device 2 causes, for each of created
subregion three-dimensional point cloud data, each of the subregion three-
28
dimensional point cloud data to be translated and rotated respectively so as to be
superposed on the surface shape model such that a distance between each of the
subregion three-dimensional point cloud data and the surface shape model stored as
described above becomes minimum (a total sum of distances between each data point
constituting the subregion three-dimensional point cloud data and the surface shape
model becomes minimum, or a total sum of squares of the distances becomes
minimum). In this occasion, as with the subregion three-dimensional point cloud
data, the surface shape model is also divided to create each of divided surface shape
model respectively corresponding to each of a plurality of subregions in the X
direction as well. Then, each of the subregion three-dimensional point cloud data is
superposed on each of the divided surface shape model of a region corresponding to
each subregion three-dimensional point cloud data.
[0052]
(5) Fifth step
In the fifth step, the control computing device 2 calculates the distance
between each of the subregion three-dimensional point cloud data after being
superposed as described above and the surface shape model, and detects a partial
defect such as an underfill of the crankshaft S based on the calculated distance.
Specific detection method will be described later.
[0053]
It is noted that the computation in the third step to the fifth step of the control
computing device 2 as described so far can be executed by using commercial threedimensional
analysis software (for example, HALCON12 manufactured by MVTec
Software GmbH).
[0054]
29
By executing the first step to the fifth step described above, inspection on a
predetermined portion in the circumferential direction of the crankshaft S is finished.
Next, the rotating device 4 is driven by the control computing device 2, and the
crankshaft S is rotated (for example, rotated by 90°) around the rotational center axis
L by the rotating device 4 and stopped. Then, by executing the above-described
second to fifth steps on other portions in the circumferential direction of the
crankshaft S, the inspection on another portion is finished. By repeating the abovedescribed
operations, inspection of the entire length and the entire circumference of
measurement object region of the crankshaftS is performed.
It is noted that in the second step, by rotating the crankshaft S with the
rotating device 4, it is also possible to acquire in advance the three-dimensional point
cloud data of the surface of the crankshaft S over the entire length and the entire
circumference of the measurement object region of the crankshaft S and thereafter to
successively execute the third to the fifth steps.
[0055]
According to the inspection method according to the present embodiment, the
three-dimensional point cloud data of the surface of the crankshaft S over the entire
length of the measurement object region of the crankshaftS is acquired by executing
the first and second steps.
Then, by executing the third and fourth steps, a plurality of divided subregion
three-dimensional point cloud data is individually superposed on the surface shape
model (each of the subregion three-dimensional point cloud data is translated and
rotated such that a distance between each of the subregion three-dimensional point
cloud data and the surface shape model becomes minimum). For this reason,
compared with a case in which ·the three-dimensional point cloud data is directly
30
superposed without being divided (the three-dimensional point cloud data is
translated and rotated such that the distance to the surface shape model becomes
minimum), the three-dimensional point cloud data will be superposed on the surface
shape model with influences of bending and torsion over the entire length of the
crankshaft S being reduced.
Thus, in the fifth step, it is possible to calculate the distance between the
subregion three-dimensional point cloud data after being superposed in the fourth
step and the surface shape model, and to accurately detect a partial defect such as an
underfill of the crankshaft S based on the calculated distance, discriminating such
partial defects from bending and torsion over the entire length of the crankshaft.
[0056]
Hereinafter, the above-described contents will be described more specifically
referring to the drawings.
Figure 3 is a schematic diagram to explain a situation in which threedimensional
point cloud data without being divided is superposed on a surface shape
model, for a crankshaft S having neither bending nor torsion. Figure 4 is a
schematic diagram to explain a situation in which three-dimensional point cloud- data
without being divided is superposed on a surface shape model, for a crankshaft S
having bending. Figure 5 is a schematic diagram to explain a situation in which
divided subregion three-dimensional point cloud data is superposed on a surface
shape model by using an inspection method according to the present embodiment, for
a crankshaftS having bending. Figures 3A, 4A and SA each show a surface shape
model. Figures 3B, 4B and 5B each show three-dimensional point cloud data.
Figures 3C, 4C and 5C each show a result of superposing by translating and rotating
31
three-dimensional point cloud data such that a distance between the threedimensional
point cloud data and the surface shape model becomes minimum.
[0057]
As shown in Figure 3, if the crankshaft S has neither bending nor torsion, it is
possible to detect a part (encircled part F in Figure 3C) in which a partial defect such
as an underfill and a dent flaw has occurred, even if the three-dimensional point
cloud data without being divided is superposed on the surface shape model.
However, as shown in Figure 4, if the crankshaft S has bending, deviation
between the rotational center axis L 1 of the surface shape model and the center axis
L2 of the three-dimensional point cloud data is large. Thus, even if the threedimensional
point cloud data is translated and rotated to be superposed such that the
distance between the three-dimensional point cloud data and the surface shape model
becomes minimum, a non-superposed part (for example, encircled parts F', F" in
Figure 4C) will occur in addition to the part of a partial defect such as an underfill
and a dent flaw (an encircled part F in Figure 4C). For this reason, even if it is
possible to detect that there is a certain type of shape failure, it is not possible to
discriminate whether a defect which occurs partially in the crankshaft S such as an
underfill and a dent flaw, or, bending or torsion over the entire length of the
crankshaft. Further, it is not possible to quantify a size of the defect.
[0058]
In contrast, as shown in Figure 5, even if the crankshaft S has bending,
according to an inspection method of the present embodiment, the three-dimensional
point cloud data is divided to create a plurality of subregion three-dimensional point
cloud data each of which respectively corresponding to each of a plurality of
subregions of the crankshaft (four subregions Al to A4 in an example shown in
32
Figure 5), each of the subregion three-dimensional data group is translated and
rotated to be superposed on the surface shape model. Thus, deviation decreases
between the rotational center axis Ll of the surface shape model and the rotational
center axis L2 consisting of each rotational center axis of the subregion threedimensional
data group, and influences of bending will be reduced. For this reason,
it is possible to accurately detect a part (encircled part F in Figure 5C) in which a
partial defect such as an underfill and a dent flaw has occurred.
[0059]
Hereinafter, a method of determining a subregion in the third step of the
inspection method according to the present embodiment will be described.
Figure 6 is an explanatory diagram to illustrate an example of the method of
determining a subregion. As shown in Figure 6, an entire length in a design
specification of a crankshaft S (the entire length with no bending) is L [mm], a
maximum radius in the design specification of the crankshaft S is R [mm] (not
shown), an assumed torsion angle over the entire length of the crankshaft S is ~a
[rad], and an assumed bending over the entire length of the crankshaftS is ~b [mm].
Here, the maximum radius R in the design specification of the crankshaft S
means a distance from the rotational center axis of the crankshaft S to the surface of
the crankshaft S (specifically, the surface of the arm S2) in which the distance
becomes maximum in CAD data based on the design specification. Moreover, the
torsion angle' ~a means an angle by which an arm S22 provided on one end side of
the crankshaft S is rotated with respect to the arm S22 in CAD data when the arm
S21 provided on the other end side of the crankshaft S is aligned to the arm S21 of
the CAD data based on the design specification. Further, the bending ~b means a
distance by which one end of the crankshaft s is displaced in a direction
33
perpendicular to the rotational center axis with respect to one end of the CAD data
when the other end of the crankshaft S is aligned to the other end of CAD data based
on the design specification.
[0060]
In the above-described case, bending per size Lp [ mm] can be assumed to be
L'lb • Lp I L [mm]. Moreover, torsion per size Lp [mm] can be assumed to be L'la • R
• Lp I L [mm]. For this reason, it is considered that satisfying the following
Formula (2) enables detection of a partial defect such as an underfill at a detection
accuracy of ±L'le [ mm].
L'lb • Lp I L + L'la • R • Lp I L ::::; 2 • L'le ... (2)
Modifying Formula (2) described above will result in Formula (1) described
below.
Lp ::::; 2L • L'le I (L'lb + L'la • R) .. ~(1)
Determining a size Lp (size in a direction parallel to the rotational center axis
of the crankshaftS) of a subregion A to be divided so as to satisfy Formula (1) makes
it possible to detect a partial defect such as an underfill at a detection accuracy of
±L'le [mm].
[0061]
For example, in case where the crankshaftS is one for an inline-four engine,
when the entire length L = 450 mm in the design specification of the crankshaft S, an
assumed bending L'lb [ mm] = 1 mm over the entire length of the crankshaft S, and an
assumed torsion angle L'la = 0 [rad] over the entire length of the crankshaftS, in order
to detect an underfill and a dent flaw at a detection accuracy L'le = 0.2 mm, it is
required that the size Lp of the subregion is to be 180 mm or less since the right hand
side of Formula (1) will be 180 illm (=2 x 450 x 0.2 I 1).
34
For example, if the spacing between adjacent journals is 100 mm, determining
a subregion such that both ends of the subregion are located at adjacent journals will
satisfy Formula (1) described above, and can reduce the influences of bending and
torsion. Besides, it is considered that the subregion is determined such that both
ends of the subregion are located at adjacent pins, or a journal and a pin which are
adjacent to each other.
[0062]
Hereinafter, a defect detecting method in the fifth step of an inspection
method according to the present embodiment will be specifically described.
In the fifth step, the control computing device 2 creates a two-dimensional
gradation image in which a pixel constituting the two-dimensional gradation image
has a density corresponding to a distance between each of the subregion threedimensional
point cloud data after being superposed in the fourth step and the surface
shape model. The two-dimensional gradation image is a projection ofthe subregion
three-dimensional point cloud data after being superposed in the fourth step to a
plane (XY plane) perpendicular to a facing direction (Z direction) between the threedimensional
shape measurement device 1 and the crankshaft S. Then, the control
computing device 2 detects a partial defect such as an underfill of the crankshaft S
based on a feature quantity obtained by subjecting the created two-dimensional
gradation image to predetermined image processing.
According to the above-described defect detection method, for example, it is
possible to extract a pixel region (pixel region corresponding to a large distance to
the surface shape model) having a density higher than a predetermined threshold, and
it is possible to automatically and easily detect a partial defect such as an underfill by
35
using a feature quantity calculated from an area and a density of the extracted pixel
region.
[0063]
It is noted that the control computing device 2 has a function of creating a
two-dimensional image in which a pixel constituting the two-dimensional image has
a density or a color corresponding to the distance between the subregion threedimensional
point cloud data after being superposed in the fourth step and the surface
shape model, in which the two-dimensional image is a projection of the subregion
three-dimensional point cloud data after being superposed in the fourth step to a
plane (XY plane) perpendicular to the facing direction (Z direction) between the
three-dimensional shape measurement device 1 and the crankshaft S. The control
computing device 2 also has a function of displaying the two-dimensional image on a
monitor.
It is possible to easily detect a pixel region corresponding to a large distance
to the surface shape model, that is, a partial defect such as an underfill, by an
operator visually recognizing the two-dimensional image.
[0064]
Figure 7 is a diagram to show an example of a two-dimensional image (twodimensional
gradation image having a density corresponding to a distance) obtained
by superposing three-dimensional point cloud data without being divided on a
surface shape model for a crankshaft S for an inline-four engine having underfills
and torsion. Figure 8 is a diagram to show an example of a two-dimensional image
(a two-dimensional gradation image having a density corresponding to a distance)
obtained by superposing divided subregion three-dimensional point cloud data on a
surface shape model by using an inspection method according to the present
36
invention, for a crankshaft S having underfills and torsion same as the one of Figure
7.
As shown in Figure 7, it can be seen that in a two-dimensional image (Figure
7B) obtained when three-dimensional point cloud data obtained on the measurement
object region (Figure 7 A) is not divided and is superposed on the surface shape
model, a distance of 0.5 mm or more has occurred over a wide range due to
influences of torsion even in a portion without underfills.
In contrast, as shown in Figure 8, it is possible to visually and clearly
recognize three underfills (encircled parts F1 to F3) having a distance of 1 mm or
more in a two-dimensional image (Figure 8B) obtained when three-dimensional
point cloud data obtained on a measurement object region of 350 mm excluding both
ends of 50 mm of the entire length 450 mm of the crankshaft S is divided and
superposed on the surface shape model such that each of subregions (A1 to A4) of a
size Lp = 100 mm corresponds to each of the subregion point cloud data and such
that both ends ofthe subregion are located at adjacent journals (divided such that the
adjacent subregions have an overlapped portion). One part (encircled part F1)
among three parts F1 to F3 cannot be visually and clearly recognized in the twodimensional
image shown in Figure 7.
In Figures 7 and 8, although description has been made taking a twodimensional
image to be subjected to visual recognition by an operator as an example,
it is also possible to automatically and easily detect a partial defect such as an
underfill by subjecting this two-dimensional image to image processing.
[0065]
It is noted that if a translating distance and a rotational angle when the
subregion three-dimensional point cloud data is translated and rotated in the
37
.,
superposition in the fourth step are stored in the control computing device 2, it is
possible to evaluate bending and torsion of the crankshaft S based on the translating
distance and the rotational angle.
Figure 9 shows a translating distance and a rotational angle when the
subregion three-dimensional point cloud data are translated and rotated, for a
crankshaft S having underfills and torsion same as those in Figures 7 and 8, in which
each of the divided subregion three-dimensional point cloud data corresponds to each
of subregions (AI to A4).
As shown in Figure 9, since the translating distance in theY and Z directions,
that is, the translating distance in a direction perpendicular to the rotational center
axis L of the crankshaft S is about 0 rnrn for any subregion, it can be evaluated that
the crankshaft S has no bending. On the other hand, since the rotational angle
around the X direction varies in a range of- 0.3° to 0.7°, it can be evaluated that the
crankshaftS has torsion for each subregion (for each cylinder).
[0066]

Figure 10 is a diagram showing a schematic configuration of an inspection
device according to a second embodiment of the present invention. Figure 1 OA is a
front view of the crankshaft S viewed from a direction of the rotational center axis L.
Figure 1 OB is a cross sectional view taken along NN line of Figure 1 OA. In Figure
1 OA, the control computing device 2 is omitted from illustration. In Figure 1 OB, a
support mechanism 5 is omitted from illustration.
As shown in Figure 10, an inspection device 1 OOA according to the present
embodiment, as well as the inspection device 100 according to the first embodiment,
includes an optical three-dimensional shape measurement device 1, a control
38
computing device 2, and a moving mechanism 3. In the case of performing
inspection by using the inspection device 1 OOA according to the present embodiment
as well, it is similar to the first embodiment in that the first to fifth steps are
performed.
However, the inspection device 1 OOA according to the present embodiment is
different from the first embodiment in that the inspection device 1 OOA does not
include a rotating device 4, and instead includes a supporting mechanism 5 that
chucks and fixes ends of the crankshaft S (the supporting mechanism 5 has no
rotating function). Moreover, the inspection device 1 OOA according to the present
embodiment is different from that of the first embodiment in that four threedimensional
shape measurement devices 1 ( 1 a to 1 d) are disposed at a pitch of 90°
around the rotational center axis L of the crankshaft S. Further, it is also different
from the first embodiment in that the moving mechanism 3 includes four single-axis
stages so as to be able to separately move the four three-dimensional shape
measurement devices 1 in a direction (X direction) parallel to the rotational center
axis L of the crankshaftS.
[0067]
The control computing device 2 of the inspection device 1 OOA according to
the present embodiment can independently control timing of measuring a surface
shape of a crankshaft S by the four three-dimensional shape measurement devices 1,
and timing of moving the four three-dimensional shape measurement devices 1 by
the moving mechanism 3.
[0068]
As described so far, the inspection device lOOA according to the present
embodiment is capable of measuring the shape of the entire length and the entire
39
circumference of a measurement object region of the crankshaft S without relatively
rotating the crankshaft S in circumferential direction (hence, the rotating device 4 is
unnecessary as described above) and thus can reduce measurement time, since the
four three-dimensional shape measurement devices 1 are disposed at a pitch of 90°
around the rotational center axis L of the crankshaftS.
Further, the moving mechanism 3 can separately move the four threedimensional
shape measurement devices 1 in the X direction, and the control
computing device 2 can separately control the timing of measuring the surface shape
of the crankshaft S by the four three-dimensional shape measurement devices 1, and
the timing of moving the four three-dimensional shape measurement devices 1 by the
moving mechanism 3. For this reason, it is possible to perform positional control of
the three-dimensional shape measurement devices 1 and the timing control of
measurement and movement of the three-dimensional shape measurement devices 1 ,
which can avoid a situation in which projected light of any of the three-dimensional
shape measurement devices 1 enters into a measurement field of view of another
three-dimensional shape measurement devices 1 different from the concerned threedimensional
shape measurement device 1, disabling the another three-dimensional
shape measurement devices 1 to perform measurement.

We claim:
[Claim 1]
A method for inspecting a crankshaft, comprising:
a first step of disposing an optical three-dimensional shape measurement
device to face a crankshaft in a direction perpendicular to a rotational center axis of
the crankshaft, wherein the three-dimensional shape measurement device is
configured to measure a three-dimensional shape of a measurement object by
projecting and receiving light to and from the measurement object;
a second step of measuring a surface shape of the crankshaft by the threedimensional
shape measurement device disposed in the first step to acquire threedimensional
point cloud data of a surface of the crankshaft over an entire length of a
measurement object region of the crankshaft;
a third step of dividing the three-dimensional point cloud data acquired in the
second step to create a plurality of subregion three-dimensional point cloud data,
each of the subregion three-dimensional point cloud data respectively corresponding
to each of a plurality of subregions of the crankshaft along a direction parallel to the
rotational center axis of the crankshaft;
a fourth step of translating and rotating each of the subregion threedimensional
point cloud data created in the third step to superpose each of the
subregion three-dimensional point cloud data on a surface shape model of the
crankshaft p;epared in advance based on a design specification of the crankshaft,
such that a distance between each of the subregion three-dimensional point cloud
data and the surface shape model becomes minimum; and
a fifth step of calculating a distance between each of the subregion threedimensional
point cloud data after being superposed in the fourth step and the surface
46
shape model, and detecting a partial defect such as an underfill of the crankshaft
based on the distance calculated.
[Claim 2]
The method for inspecting a crankshaft according to claim 1, wherein
in the second step, measurement of the surface shape of the crankshaft by the
three-dimensional shape measurement device and relative movement of the threedimensional
shape measurement device in a direction parallel to the rotational center
axis of the crankshaft are alternately repeated.
[Claim 3]
The method for inspecting a crankshaft according to claim 1 or 2, further
comprising
a sixth step of evaluating bending and torsion of the crankshaft based on a
translating distance and a rotational angle when each of the subregion threedimensional
point cloud data is translated and rotated in the fourth step.
[Claim 4]
The method for inspecting a crankshaft according to any of claims 1 to 3,
wherein
in the third step, when a defect detection accuracy required in the fifth step is
± ~e [mm], a size Lp [mm] of the subregion of the crankshaft, which is to be divided,
in a direction parallel to the rotational center axis of the crankshaft is determined so
as to satisfy the following Formula (1):
Lp ~ 2L·~e I (~b +~a· R) ... (1)
where, in the above Formula (1), L [mm] refers to an entire length in the design
specification of the crankshaft; R [mm] refers to a maximum radius in the design
specification of the crankshaft; ~a [rad] refers to an assumed torsion angle over the
47
entire length of the crankshaft; and ilb [mm] refers to an assumed bending over the
entire length of the crankshaft.
[Claim 5]
The method for inspecting a crankshaft according to any of claims 1 to 4,
wherein
in the third step, the subregion of the crankshaft is determined such that both
ends of the subregion, which is to be divided, in a direction parallel to the rotational
center axis of the crankshaft are located at adjacent journals of the crankshaft,
adjacent pins of the crankshaft, or a journal and a pin adjacent to each other of the
crankshaft.
[Claim 6]
The method for inspecting a crankshaft according to any of claims 1 to 5,
wherein
in the fifth step,
a two-dimensional gradation image is created, in which a pixel constituting
the two-dimensional gradation image has a density corresponding to a distance
between each of the subregion three-dimensional point cloud data after being
superposed in the fourth step and the surface shape model, wherein the twodimensional
gradation image is a projection of each of the subregion threedimensional
point cloud data after being superposed in the fourth step to a plane
perpendicular to a facing direction between the three-dimensional shape
measurement device and the crankshaft, and
a partial defect of the crankshaft is detected based on a feature quantity
obtained by subjecting the two-dimensional gradation image to predetermined image
processing.
48
:)
[Claim 7]
The method for inspecting a crankshaft according to any of claims 1 to 6,
further comprising
a seventh step of:
creating a two-dimensional image in which a pixel constituting the twodimensional
image has a density or a color corresponding to a distance between each
of the subregion three-dimensional point cloud data after being superposed in the
fourth step and the surface shape model, wherein the two-dimensional image is a
projection of each of the subregion three-dimensional point cloud data after being
superposed in the fourth step to a plane perpendicular to a facing direction between
the three-dimensional shape measurement device and the crankshaft; and
displaying the two-dimensional image.
[Claim 8]
A device for inspecting a crankshaft, comprising:
an optical three-dimensional shape measurement device which is disposed to
face a crankshaft in a direction perpendicular to a rotational center axis of the
crankshaft, and which is configured to measure a three-dimensional shape· ·of a
measurement object by projecting and receiving light to and from the measurement
object; and
a control computing device configured to control an operation of the threedimensional
shape measurement device and executing predetermined computing on a
result measured by the three-dimensional shape measurement device, wherein:
a surface shape model of the crankshaft prepared based on a design
specification of the crankshaft is stored in advance in the control computing device;
49
three-dimensional point cloud data of a surface of the crankshaft over the
entire length of a measurement object region of the crankshaft acquired by the threedimensional
shape measurement device measuring the surface shape of the
crankshaft is input to the control computing device; and
the control computing device is configured to execute steps of:
dividing the three-dimensional point cloud data which is input, to create a
plurality of subregion three-dimensional point cloud data, each of the subregion
three-dimensional point cloud data respectively corresponding to each of a plurality
of subregions of the crankshaft along a direction parallel to the rotational center axis
of the crankshaft;
translating and rotating each of the subregion three-dimensional point cloud
data to superpose each of the subregion three-dimensional point cloud data on the
surface shape model such that a distance between each of the subregion threedimensional
point cloud data created and the surface shape model stored becomes
minimum; and
calculating a distance between each of the subregion three-dimensional point
:-:
cloud data after being superposed and the surface shape model, and detecting a
partial defect such as an underfill of the crankshaft based on the distance calculated.
[Claim 9]
The device for inspecting a crankshaft according to claim 8, further
comprising
a moving mechanism whose operation is controlled by the control computing
device and which is configured to relatively move the three-dimensional shape
measurement device in a direction parallel to the rotational center axis of the
crankshaft, wherein
50
the control computing device is configured to control the operation of the
three-dimensional shape measurement device and the moving mechanism such that
measurement of the surface shape of the crankshaft by the three-dimensional shape
measurement device, and relative movement of the three-dimensional shape
measurement device in a direction parallel to the rotational center axis of the
crankshaft by the moving mechanism are alternately repeated.
[Claim 10]
The device for inspecting a crankshaft according to claim 8 or 9, wherein
the three-dimensional shape measurement device is a three-dimensional shape
measurement device of a pattern projection type.
[Claim 11]
The device for inspecting a crankshaft according to any of claims 8 to 10,
compnsmg
four three-dimensional shape measurement devices disposed at a pitch of 90°
around the rotational center axis of the crankshaft, wherein:
the moving mechanism can separately move the four three-dimensional shape
measurement devices in a direction parallel to the rotational center axis of the
crankshaft; and
the control computing device can separately control timing of measuring the
surface shape of the crankshaft by the four three-dimensional shape measurement
devices and timing of moving the four three-dimensional shape measurement devices
by the moving mechanism.
[Claim 12]
The device for inspecting a crankshaft according to claim 11, wherein
51
the control computing device is configured to control the operation of the
moving mechanism and the four three-dimensional shape measurement devices such
that while any one pair of three-dimensional shape measurement devices disposed in
a direction opposite to each other among the four three-dimensional shape
measurement devices are measuring the surface shape of the crankshaft, any other
pair of three-dimensional shape measurement devices disposed in a direction
opposite to each other are moved without performing measurement, and such that
projected light of one three-dimensional shape measurement device constituting the
pair of three-dimensional shape measurement devices measuring the surface shape of
the crankshaft does not enter a measurement field of view of the other threedimensional
shape measurement device.