DESC:1
Field of the Invention:
The invention relates generally to the field of manufacturing, manufacturing systems, work-holding and tool design. More specifically, the invention relates to a workpiece-holding fixture enabled with hardware and integrated software which monitors various conditions of fixtures before and during machining based on readings collected from various sensors. The data collected before machining ensures that the fixture and component are rigidly held and suitable for the processes. The said sensors are mounted on the fixture and execute fixture-level analytics to help isolate and troubleshoot issues related to manufacturing and production.
Background of the Invention The automated and efficient management and troubleshooting of workholding fixtures in the process of manufacturing has always been a desirable requirement. Existing technologies in machine control only deal with checking the pressure values on components by means of a pressure gauge installed on the same. Even in such situations, the pressure reading on the pressure gauge has to be checked manually by the operator to track any changes in the values. This allows only for periodic inspections and does not provide a means to collect data in real time. If there is a sudden drop in pressure then the operator has to stop the machining operation to prevent any potential damage to components, fixture and machines. The process described herein is time-consuming and not effective. Additionally, since it is a manual process, man made errors are frequently encountered. Additionally, current solutions do not allow the pressure and/or other parameters of the components to be monitored when they are in the process of undergoing machining operations. Moreover, there are no solutions as of now for sensing vibration, component presence, oil level, quality of oil level etc. of the components on the fixture and the power pack. Accordingly, there is a need for an efficient and intelligent system that monitors premachining operations machines with higher accuracy as opposed to manual or mechanical control.
2
Checking the Sequence of Operations during pre-machining: During the clamping process, various sequences are involved such as end references, 3-point clamping, work supports, and clamp over work supports. The different sequences are controlled by sequence valves. In the event of sequence-valves malfunctioning, there is an error on the sequence which makes the clamping ineffective (If, for example, the sequence valve doesn’t work and the work supports go ahead of the main clamps, then the component will be lifted up and the clamping will not be effective.) In the subject invention, it can be identified if the pre fixed sequence for clamping is adhered to or not. If there is a malfunction, it can stop further processing.
The field of manufacturing has witnessed a transformation from conventional work-holding solutions to advanced solutions. Work-holding technology has been facing the challenge of effectively utilizing recent advancements in machine tool and cutting tool technology. There is a growing requirement in the field of work-holding to make fixtures smarter by making them Internet of things (IoT)- and Industry 4.0 (fourth industrial revolution) -enabled.
The subject invention addresses the limitations of the prior art by proposing an intelligent, precision work-holding fixture which works in a coordinated manner for comprehensive safety before and while machining components. It also uses machine learning and artificial intelligence as a part of the system, thereby digitizing the work-holding technology.
The subject invention provides a system and method consisting of a hardware and software component. The hardware can be attached onto any new or existing workpiece holding fixture. It automatically senses vibration in three axes, pressure and component presence on the fixture and temperature and oil level on the power pack during the machining process and transmits the analysed data in real time toa computer or other handheld computing device. The software module monitors the condition of machines on the floor based on readings collected from
3
sensors on the fixture. Multiple sensors are used for performance analysis and machine-learning and artificial intelligence is used to identify and predict failures, if any. The device also helps to constantly improve the performance of the component, by helping to isolate and troubleshoot the components to prevent any mishaps or accidents during the machining process. Since the analysis is done at the fixture level, the troubleshooting process is both faster and cost-effective. In fixture level analytics, sensors are placed on the fixtures and the data received from these sensors is analysed. As the fixture is closest to the component, any change during machining process or tooling problems will first be received by these sensors.
The subject invention ensures that the component is loaded correctly on the fixture and generates an alert if the component is not in place or if there happens to be a shift in its position during the machining process. This is an improvement over the prior art where the position of the components is only monitored by air seat sensors which are not efficient enough to indicate a shift in the position of the component. Additionally, in the existing solutions, clamping and de-clamping pressure details are checked manually and sequence of operations are not confirmed. The subject invention, on the other hand, enables continuous monitoring of the clamping and de-clamping cycles and also generates an alert in case of any inconsistency. Since the pressure is monitored during the machining and any unusual vibrations are detected on a real time basis, there are no component rejections because of an error in the component loading or pressure irregularity unlike the prior art.
Problems with the Prior Art:
Chinese Patent Application No. CN203151795 (U) titled as “LED lighting fixture monitoring system” deals with an LED lighting fixture monitoring system by collecting and processing field data from the same. The components of the monitoring system include a central controller, a wireless device and a sub-controller. The central controller wirelessly
4
communicates with a plurality of sub-controllers via a wireless device, and each sub-controller passes through a control loop connected to a street light or a landscape light.
The aforementioned invention provides the user with an option to only monitor an LED lighting fixture by means of a wireless device. It does not give the user an option to monitor a component during pre-machining and machining or tool in motion. Additionally, there is no provision to provide the sensing, transmission and analysis of the data in real-time to the end user.
United States Patent Application No. US20150308856 titled as “Automatic fixture monitoring using mobile location and sensor data with smart meter data “deals with a method for monitoring resource information and user activity comprising, acquiring data streams from sensors, computing discrete events from each data stream, extracting a sequence of discrete sensor-meter event item sets based on the discrete events, discovering frequent sensor-meter event item sets that occur together, and determining a frequency of occurrence for each sensor-meter event itemset. The components of the invention include an accelerometer, a gyroscope, a microphone, a temperature sensor, a light sensor, a barometer, a magnetometer, and a compass in the system.
The aforementioned invention mentions fixture monitoring but does not gives the option of extracting data from a fixture while it is in the machining process. There is no reference to the sensors being attached to/on the fixture as in the subject invention. Additionally, it does not mention the features of the subject invention like real time data transmission and analysis, troubleshooting of components, provision of generating alerts, predictive analysis etc.
5
United States Patent Application No. US9971343 (B2) titled as “Portable Intelligent Controlling System for Machines” deals with a portable plug-and-play intelligent system for monitoring and controlling process variations of a workpiece of a machine. The components of the system include one or more sensors, a controller, a plug & play modular fixture with a motor, a database and a comparison unit. The sensors collect real-time data of the workpiece from the machine. The controller processes this data and stores it in a database. The comparison unit compares the real-time data of the workpiece with predefined specifications of the workpiece to check whether the real-time data is accurate or not. The control unit triggers the motor to allow the plug-and-play modular fixture to adjust the parameter variables of the machine to the predefined specifications of the workpiece when the real-time data is not accurate.
The aforementioned invention discusses how a machine can be controlled with real time data.
It does not deal with the determination of fixture performance and analytics on the component being held and monitored by the fixture, and the data analytics process and system that is able to monitor the pre-machining and machining parameters of the component while the machining is taking place and the fixture along with the component which is moving inside the machine.
In the subject invention, the vibration and pressure of the components on the fixture before and during machining is monitored. In the pre-machining sequence, the components are checked for the correct clamping and work support sequence of operations for the machining process to start.
Additionally, by means of vibrations sensors placed in the x,y and z directions, a tool breakage, casting defect or another error which leads to the vibration, pressure or another measured parameter exceeding the pre-set threshold value, activates an alarm and stops the machining operations.
6
Clamping pressures are also continuously measured in the subject invention and, through analysis of the data, it is possible to get weekly, monthly and yearly production analytics.
Objects of the Invention: • The main object of the invention is aimed at automatic monitoring of vibration along three axes, namely the X, Y and Z axis, pressure and component presence on a workpiece-holding fixture, temperature and oil level on the power pack unit during the machining process.
• A further object of the invention is to provide the sensing, transmission and analysis of the data relating to the component and the workpiece-holding fixture in real-time.
• A further object of the invention is to determine the tool life based on the analysed data.
• A further object of the invention is to calculate the downtime of the machine, i.e., the total time when the machining is not performing the machining operations.
• A further object of the invention is to prevent machining accidents when linked to the machine controls, by means of the provision of alert signal generation.
• A further object of the invention is to provide the data for the total number of clamp/de-clamp cycles on the fixture so the clamps can be changed or the oil seals changed at the appropriate time.
• A further object of the invention is to provide an indication of when the oil needs to be changed in the power pack.
• A further object of the invention is to generate alerts on the breach of a pre-set threshold specifications for various parameters.
• A further object of the invention is to provide comparative performance data for machines.
7
• A further object of the invention is to provide information on casting tolerances of the tool.
• A further object of the invention is to provide useful information to increase the efficiency and improve quality of work-holding technology and make smart manufacturing affordable.
• A further object of the invention is to increase safety and productivity in general and to optimize tool life.
• A further object of the invention is to optimize tool life
• A further object of the invention is to convert older fixtures to smart ones by making them IoT-enabled by mounting the present system on the existing fixtures running on these machines.
• A further object of the invention is to use the data from the machine such as speeds, feeds, temperature and hydraulics etc. and analyse the same to obtain signature parameters for good output and quality.
Statement and Summary of the Invention:
According to the invention there is, therefore, provided a device, a system and a method to automatically monitor vibration along three axes, namely the X, Y and Z axis, pressure and component presence on a work holding fixture, temperature and oil level on the power pack during the machining process and present it to the end user in a useful format. The said system also checks the sequences during pre-machining determining whether the clamps and work supports are operating in the right sequence. The analyzed data is transmitted in real time to a computer or a hand-held device. If any of these parameters are above the set threshold levels, then an alert/alarm is sent to a computer or handheld device and the machining can be stopped to prevent any errors or rejections of components.
8
The subject invention consists of:
(a) A primary mechanism or means for capturing the data from the component before and during the machining process by means of sensors on the fixture and the power pack.
(b) A primary mechanism or means for storage and management of captured data in a database to be used for analysis. The data is collected from the sensors on the fixture and the powerpack.
(c) A primary mechanism or means of visualization of data in real time is through the Dashboard.
(d) A primary mechanism or means of enabling a prediction analysis based on pre-set information provided by the user with the vibration and pressure data that is collected from the sensors on the fixture, uploaded to the cloud server and then analysed with the help of the database and displayed on the GUI.
(e) An auxiliary mechanism or means of generating alerts on the breach of a pre-set threshold values for various parameters to reduce and prevent rejections during machining. The alerts are generated with the data from the sensors which is analysed and if the it exceeds the threshold level, the alerts are generated.
Description:
The description of the preferred embodiment is meant to demonstrate the broad working principles of the invention without limitation as to possible adaptations, extensions, applications etc., which would be obvious to a person skilled in the art. This invention is illustrated in the accompanying drawings, throughout which, like reference numerals indicate corresponding parts in the various figures. In the interest of brevity and for the purposes of exemplary explanation, references have been made to a system, depicted in figures1 to 8 herein without limitation, to describe the invention which is essentially directed towards providing the user with option to automatically monitor three-dimensional vibration, pressure and
9
component presence on a workpiece holding fixture, temperature and oil level on the power transmitting unit during the machining process.
Figure 1gives a schematic view of the functional units that comprise the system. and comprises a plurality of components such as a Fixture (1), Component/workpiece (2), Proximity Sensor (3), Pressure Sensor (4), Vibration Sensor X Axis (5), Vibration Sensor Y Axis (6) Vibration Sensor Z axis (7), Power Transmitting Unit (8), Inductive power transmitter (9), Hydraulic Power transmitting unit (10), Temperature Sensor (11), Oil level sensor (12) and Dashboard (13).
Figure 2 describes the step by step working of the subject invention in the flow chart provided.
Figure 3 describes the vibration profiles of different machining operations.
Figure 4 illustrates the general steps of sensing, transmission and analysis of the captured data.
Figure 5 gives a visual representation of the machine/fixture layer, sensor layer, cloud analytics and AI layer and the graphical user interface (GUI) layer.
Figure 6describes the vibration sensor data collection and analysis.
Figure 7illustrates the proximity and pressure sensor data collection and analysis.
Figure 8describes how machine learning model works on test and train data.
Figure 9 illustrates the block diagram of the smart work holding fixture system.
Figure 10illustrates the steps of the Machine Operation Classification analysis.
Components of the Invention
The subject invention comprises the following:
1. Fixture- Workholding device that holds the component in place when the machining is being done on the component. When the same type of components is reloaded on the fixture, the position of the components always remains the same in reference to X,Y and Z Axis
10
2. Proximity Sensor- Senses that the component is present on the fixture
3. Pressure Sensor- To monitor the clamping and de-clamping pressure of the hydraulic oil in the clamping circuit in the fixture
4. Vibration Sensors- Monitors the vibration in three-dimensions, namely the X, Y and Z axis, during machining of the component using different cutting tools.
5. Power transmitting Unit- Consists of power supply batteries, battery charger, Wi-Fi signal transmitter unit and a microcontroller. The battery provides the power supply to the transmitter.
6. Inductive power transmitting unit- Used to transfer power to the battery charger
7. Hydraulic Power transmitting unit- A unit that transmits power to the fixture using hydraulic fluid for clamping, de-clamping, activating work supports etc.
8. Temperature Sensor- Senses the oil temperature and generates an alert if the temperature is below or above a pre-set threshold.
9. Oil level sensor- Used in the power pack to check the level of the hydraulic oil and to indicate the drop-in level over a period of time
10. Load cell: Senses the force applied on the component and generates alert if the captured force value is above or below a pre-set threshold.
11. Strain Gauge: Senses the strain on the fixture and generates alert if the captured strain is above or below a pre-set threshold.
12. Power Sensor: Reads the power value during operation and generates alert when exceeds thresholds
13. Noise Sensor: Reads the noise level and analyzes surrounding ambient sound in the audible frequency
14. RPM sensor: Reads the rate of revolutions per minute
15. Dashboard- Custom-built for the end user which resembles their factory floor, showing all the machines in action as well as a summary on the health of each machine,
11
fixture and powerpack, along with the provision for pre alerts and alerts if any value is out of sync.
16. Camera- To take images of the component and processing it to analyse and identify the correct component among a group of components, to check missed machining operations on the component.
The Proximity Sensor, Pressure Sensor, Vibration Sensors, Temperature Sensor, Oil level sensor, Power Sensor, Noise Sensor, RPM sensor and Camera or any subset of the said components may collectively be referred to as the Sensors.
Smart Work-piece holding fixture:
The work-piece holding fixture with the component affixed on it is placed inside the machine and machining operations are done on the component. The fixture along with the component is in motion inside the machine.
The data collection process for the system begins at the sensors mounted on the fixture. Sensors to monitor vibration, pressure, and component sensing are mounted on the fixture, and sensors to monitor oil levels and temperature are mounted on the power transmitting unit. The hydraulic power transmitting unit is connected to the fixture with quick coupling hoses. The component clamping sequences take place when the power transmitting unit is switched on. After the clamping has taken place, the quick coupler hoses are removed and the fixture and the component go into the machine for all the operations to be performed.
System to calculate downtime of the machine and time analysis and Operation Labelling using a Smart Work-piece holding fixture The Smart work-piece holding fixture system comprises of the hardware and software modules. The hardware module of the system consists of the components as covered above. The system
12
also consists of a microcontroller which is responsible for converting the sensor data to a format that can be processed by the software. Additionally, a specially designed IoT Gateway plays the role of the local server on the system environment. The IOT gateway is the link between our network and the internet. The IOT Electronics consist of Sensors to acquire Vibration, Pressure, Proximity, Load cells, Strain Gauges, Accelerometer data from the fixtures. These data, which may be analog or digital, are collected and transmitted to cloud using TCP IP by microcontroller. The fixture may sometimes be a moving component, resulting in the possibility that power cabling and wiring may interfere with the operation of the system. As a result, there is a provision for the electronics to be optionally powered by a Battery or Inductive Power coupling/Transmitting Unit to avoid said possibility.
The smart work holding fixture system automatically senses vibration in three dimensions, pressure and component presence on the fixture and temperature and oil level on the power pack during the machining process and transmits the analysed data in real time to a computer or other handheld computing device. The Software module is enabled with machine learning and artificial intelligence and monitors the condition of machines on the floor based on readings collected from sensors on the fixture. Multiple sensors viz. Vibration, Pressure, Proximity, Temperature, Battery etc. record data which is analyzed for unique patterns for idle/working state, different operations, tool state and every deviated behavior from the normal operating behavior.
Based on these patterns, models are built to indicate any abnormality in the operation. This platform is used in predictive maintenance of the machine and fixture and prevents any kind of accident or mishap.
13
This platform also gives overall production details without any manual intervention and helps organization monitor/maintain optimum production rate.
The sensors are used for performance analysis and machine learning and artificial intelligence is used to identify and predict failures, if any. The system also helps to constantly improve the performance of the component, by helping to isolate and troubleshoot the components to prevent any mishaps or accidents during the machining process. This is achieved by the integration of available preset machine information with the sensor data collected from the different sensors used in the system, which helps build a machine learning model. A machine learning model is built by collecting continuous data and categorizing the data into different stages like good machining condition, bad machining condition, different types of failures, pre-failure condition etc. By training the model continuously on these categories, the model is able to intuitively predict events before actual occurrence and thus forecast the health of the machine, thereby providing predictive maintenance as a feature. Categories like good machining condition, bad machining condition etc. are created by observing the quality of Sensors data, denoted by factors like frequency of threshold breaches per cycle, per hour, per day etc, rate of production and other anomalous behavior recorded by sensors data. A machine learning algorithm is provided with data to learn from in multiple iterations which is used to train the machine learning model. The model is programmed in a way that it gives predictions based on the feature it extracted during training on the sensors data. Depending on the problem statement a wide range of machine learning models are used like linear regression, logistic regression, k means clustering, time series analysis etc. to name a few. The user can click on each machine to get a more detailed understanding of the analysed data per fixture. Data from each sensor is visualized graphically with the provision to go back and check the historical performance of the machine, the fixture and the powerpack. Detecting tool wear, predicting tool breakage, predicting probable production efficiency, anomaly detection etc are done by
14
applying the above-mentioned machine learning algorithms on the sensor data. For example, an analysis of data collected from vibration could reveal a wide variety of information like casting tolerances, vibration level comparison between tools, idle time of the machine, number of cycles run, number of components produced and vibration comparison across machines in the plant. If the vibration crosses a set upper threshold limit, the system can be designed to shut the machine off remotely through the emergency switch of the machine. An analysis of the vibration data would also give an accurate feedback on how many cycles the machine has run, the idle time of the machine, the reasons for machine downtime and the number of components produced. When no machining is going on, the vibration will be less compared to when actual machining is going on. The downtime of the machine can be calculated from this set of data. Machine maintenance schedules could also be linked to this information. Sensors to monitor pressure generates alerts to the end-user and which can again be linked to the machine’s emergency switch if required to ensure that in case of a sudden pressure drop the machine is switched off. Immediate alerts for any change in the sequencing of primary clamps, work supports and secondary clamps during the loading and unloading cycles would also help in giving immediate feedback to the machine to ensure that quality related issues due to wrong sequencing are eliminated. Since the analysis is done at the fixture level, the troubleshooting process is both faster and cost-effective. Fixture level analysis implies that the data is collected from sensors mounted on the fixtures, which gives real time physical conditions of the fixture e.g. Vibration, Pressure, Proximity, Temperature, Battery etc. The subject invention uses graphical data visualizations to predict the type of machining which is taking place along with forecasting the machine and tool life ensuring proactive and effective maintenance. Machine learning/artificial intelligence model is made for drawing insights in primarily 7 different phases. 1) Data collection
15
2) Exploratory data analysis 3) Choosing the model suitable to problem statement 4) Training the model 5) Evaluation 6) Parameter tuning 7) Prediction or Inference 8) Monitoring (additional) In an embodiment of the invention, 3 Vibration sensors and 1 pressure sensor are mounted on the system. The machine learning model was executed in the following steps: 1) Data Collection: Data was collected for an approximate time of 3 months for observation and model preparation. During this period, real events like tool breakage, tool change, machining parameter change, were registered and tagged against the data collected. In this way, numerous occurrences of same events are registered. 2) Exploratory Data Analysis: After significant amount of data is collected in the collection phase, the data is cleaned of any unwanted or lesser important variables. For example,:(i) Only operation/clamped cycle, i.e data when pressure is above 15 bar while machining the components, was used to train a model for tool wear, tool breakage etc (ii) To detect tool wear, only maximum Vibration of the clamped cycle was processed, to track the trend of the increasing vibration (provided all other machining parameters are constant) 3) Choosing model suitable for problem statement: (i) In order to predict tool breakage, the k-means cluster algorithm was chosen which is a clustering algorithm to create clusters of data that depict tool
16
breakage. In the test case, when both axial and vertical vibration exceeded a peak value during a clamp cycle at the same instance then tool breakage was indicated. (ii) For tool wear prediction, Support Vector Machine and time series analysis was used to classify different stages of the tool life with the increasing trend of Vibration data which indicated tool wear (provided all other machining parameters are constant) 4) Train the model: (i) K-means clustering algorithm/model was trained on tool breakage events historical data, creating clusters of tool breakage event and non events (80% data) and tested (ii) Support Vector Machine model was trained on data comprising of maximum vibrations of each clamped cycle collected over several tools life to categorize the data into 3 different stages of tool life. Time series model was trained on time index data of maximum vibration of clamped cycle. 5) Evaluation: (i) Evaluation of the k-means clustering model was done on the preceding tool breakage events giving an accuracy of above 85% which is a good accuracy rate as tool breakage is there are many variables in the event of tool breakage (ii) Evaluation of SVM model gave an accuracy of 80% which is a fine accuracy as there are multiple variables that accelerate or decelerate tool wear Evaluation of time series model gave an accuracy of above 70%. 6) Parameter tuning: To give better accuracy the calculative parameters of the model are tuned with more iterations or simulations. The parameters that give best accuracy was chosen for the final model.
17
7) Prediction or Inference: Once the accuracy scores of our models were satisfactory enough, they were integrated with the software to provide tool breakage prediction and tool wear detection. 8) Monitoring: The predictions were monitored closely and on finding any discrepancy in the output, the data was inspected to check if the nature of data is still same on which the models were built. If and when the nature of data changes, the entire process of building model is repeated from scratch i.e step 1).
The data acquired from the sensors also helps in the detection of any tool wear or tool breakage in the device. This is done by a continuous monitoring of the tool vibration pattern, a change in which indicates a tooling dysfunction. An anomaly in the tool condition can thus be predicted in advance, which in turn can help us indicate when the tool will wear out and needs to be changed and before the component gets rejected. This situation generally happens when a blunt tool or a broken tool is used to machine the component. The vibration patterns can also be analysed individually for determining minute tooling dysfunction. This analysis is primarily based on the fact that each and every machining operation for e.g. milling, drilling, pocket milling, face milling etc. is associated with a unique and specific vibration pattern, and the identification of the respective pattern for each operation coupled with the analysis of the received data leads the user to the required information.
Sensors data produces unique patterns for each state of machine, each operation, new tool, old tool, broken tool. These patterns are labelled with each condition. From the patterns we build machine learning algorithms using which we can predict when a condition commences. This is how predictive maintenance works. This platform is used in predictive maintenance of machine and fixture and prevents any kind of accident or mishap. This platform also gives
18
overall production details without any manual intervention and helps organization monitor/maintain optimum production rate. Additionally, as part of trouble-shooting process, there is a provision of generation of an alert signal in case a pre-set threshold value is crossed for a measured parameter. The thresholds are set by the user beforehand on the Graphical User Interface (GUI). The generated alert can be of any form i.e.visual alert on the GUI or visual LED alert on the system hardware. The thresholds for different sensor parameters are set by the user based on the manufacturing process being used by the system. Alerts are triggered on crossing the threshold limits of parameters, which helps operators to stop the machine and prevent the machining component from being damaged or distorted, and save valuable production hours. This feature enables easy interpretation of data for a suitable solution or for productivity improvements. E.g., in the process of temperature monitoring, the system will indicate the need for a heater or cooler when the temperature goes outside the limits. If the vibration crosses a set upper threshold limit, the system can be designed to shut the machine off remotely through the emergency switch of the machine. Sensors to monitor pressure would alert the operator via their phone or on any smart device and this can again be linked to the machine’s emergency switch if required to ensure that in case of a sudden pressure drop the machine is switched off. Immediate alerts for any change in the sequencing of primary clamps, work supports and secondary clamps during the loading and unloading cycles would also help in giving immediate feedback to the machine to ensure that quality related issues due to wrong sequencing are eliminated. By monitoring the clamp/de-clamp cycles, alerts for maintenance of fixtures, ordering of spares and seal kits would also be automatically generated by the system and alerts sent via smartphone to the concerned people. An analysis of vibration would also give an accurate feedback to the management on how many cycles the machine has run, what the idle time per machine is, the reasons for machine downtime and the number of components produced. Machine maintenance schedules could also be linked to this information.
19
Since the subject invention is enabled with Artificial Intelligence and Machine Learning, it can also help the user be more proactive on machine, fixture and tool performance as well as how machines are performing compared to each other. There is a provision of analysis of the historical data of the machine as well, and a comparison of various parameters can be made across all machines in the plant giving the user useful information about cost reduction, increase of efficiency and quality improvement. The comparison unit of the system compares the real-time data of the workpiece with predefined specifications of the workpiece to check whether the real-time data is accurate or not. Additionally, the system uses artificial intelligence to assess the quality of the components and identify the machined components within spec from the machined components which are out of spec or out of tolerance. To accomplish this, the camera captures the images of the components in their correct setting. Thereafter, the system software compares the trained images to the subsequent ones taken in the manufacturing process to see if there is a variation. Collecting images of good and bad components is a pre-requisite process of training the AI/ML model. Once AI/ML model is trained, it is tested with similar data and trained model is capable of giving scores to the image based on the features extracted by the model during training. Basically artificial intelligence model is used to train on images of good component and bad component and based on this AI model assesses the quality of the component.
Method of Conducting Idle/Run time analysis and Operation Labelling using a Smart Work-piece holding fixture
Idle time is the time when the fixture is not clamped and performing no operation and is not part of a clamp cycle. Run time is the time when the fixture is clamped and performing an operation on the component and is a part of the clamp cycle. The sensors capture the analog signal variations from the component on the fixture based on the machining operation being
20
done on it and send those values to the microcontroller. The microcontroller reads the sensor values after being digitized through an analog to voltage convertor circuit. This parsed data is then sent to the specially designed IoT (Internet of things) gateway. IoT gateway (hub) is a device which interfaces with the system on one side and the external network on the other side. Any and all data from the system to the external network/internet /server will pass through the IoT gateway. The IoT gateway can collect data from more than one system module. The gateway is compatible to communicate with different types of networks in order to collect sensor and machine data as well as potentially interact with the machines. The gateway is also the point where data has the option to leave the customer’s network and transfer to the cloud platform which is where the heavy computation analytics takes place as the cloud can leverage scalable computational performance based on the volume of data coming in.
Data from the microcontroller is sent to the IoT gateway using the MQTT (Message Queue Telemetry Transport) protocol. The MQTT protocol is a publish-subscribe (Pub/Sub) network protocol that allows messaging between machines. It is used in a situation where wireless connectivity latency keeps varying, playing the role of a message broker to ensure that the data is not lost in the process.
After the data makes it to the specially designed IoT gateway through the MQTT protocol, it is ready to be pushed to the cloud, where the system database is hosted. For this process, another queue (RabbitMQ) is used to stage the data before storing as it helps ensure that data is not lost in this process. The RabbitMQ waits for a sizeable payload of data before it is pushed into the Cloud Database. The database chosen for this purpose is MongoDB.
21
Apart from data being stored in MongoDB, the latest data coming in is stored in a Caching Database named Redis in order to provide easy retrieval of data while the User Interface makes a call to display data.
Along with the data captured, the software code stores a unique identifier for each sensor which helps determine the fixture it is associated with as well as the parameter type it is capturing. The data stored in the database is then used for analysis. Server-side software is written to query this database to serve the data to the end user. Apart from handling the actual data, the business layer also enforces authentication on who can access this data. The analysed data is then transmitted in real time to a computer or handheld device. A presentation layer displays the sensor data to the end user. For this, front end software is written so that it has the ability to interact with the business layer and also present the data in the form of graphical representations which also ensures a continuous data capture.
The next step in the process is Machine learning prediction analysis for the acquired data. A portion of the data is correlated to the pre-set information provided by the customer regarding the machining operations to be performed in that time period, and the data is trained into a certain category to inform the user about type of machining operation taking place. Once the classification training is done, the model is tested with the remaining data to determine accuracy. This step is repeated till an acceptable level of accuracy is attained, after which the work is deployed into the live server. ,CLAIMS:We claim: 1.A system to automatically monitor vibration, pressure and component presence on a work-piece holding fixture during the machining process within a specific period comprising: a) sensor inputs for the capture of operational data from the component before and during the machining process. b) A data collecting unit for the collection of operational data from the sensor inputs. c) A data transmission unit for transmission of such collected operational data. d) A server for the collection of the transmitted data from the data transmission unit. e) A data storage unit located on the server for the storage and management of captured data in a database. f) An analysis unit located on the server for the determination of parameters based on the transmitted data. g) A historical data repository unit located on the server for the storage of the transmitted data and corresponding operational data parameters including historical operational data. h) An evaluation unit located on the server for comparison of transmitted data pertaining to operational data against corresponding historical data in the historical data repository i)A logic unit located on the server for the determination of deviations in operational data and the corresponding historical data relating to a particular parameter j) A power transmitting unit housing the power supply batteries, battery charger, Wi-Fi signal transmitter unit and a microcontroller. k) A hydraulic power transmitting unit to transmit power to the fixture using hydraulic fluid. l) A means of forecasting the system’s health by predicting events before actual occurrence based on the comparison of transmitted data pertaining to operational data against corresponding historical data in the historical data repository and deviations therein. m) A means of generating alerts on encountering deviations in operational data and the corresponding historical data relating to a particular parameter.

n) Camera to capture images of the component and processes it to analyse and identify the correct component among a group of components, to check missed machining operations on the component o) A display unit for visualization of data. 2. The system as claimed in claim 1 wherein the sensor inputs may include all or any of the following: Proximity Sensor, Pressure Sensor, Vibration Sensors, Temperature Sensor, Oil level sensor, Load cell, Strain Gauge, Power Sensor, Noise Sensor, RPM sensor. 3. The system as claimed in claim 1 wherein the operational data collected may include data relating to all or any of the following operational parameters: presence of the component on the fixture, clamping and de-clamping pressure of the hydraulic oil in the clamping circuit in the fixture, vibration in three-dimensions, namely the X, Y and Z axis during machining of the component, oil temperature, level of the hydraulic oil, force applied, power during operation, noise level and surrounding ambient sound in the audible frequency, rate of revolutions per minute, alarms, messages and other notifications. 4. The system as claimed in claim 1 wherein the server is a remote server located at a different location than the fixture system. 5. The system as claimed in claim 1 wherein the visualisation of data and the generation of alerts are available to the operator or any person in real time during the course of execution of the machining process.
6. The system as claimed in claim 1 wherein the work-piece holding fixture with the component affixed on it is placed inside the machine and machining operations are done on the component.

7. The system as claimed in claim 1 wherein the fixture along with the component is in motion inside the machine.
8. The system as claimed in claim 1 wherein the power transmitting unit is an inductive power transmitting unit.
9. The system as claimed in claim 1 wherein the Sensors to monitor vibration, pressure, and component sensing are mounted on the fixture, and sensors to monitor oil levels and temperature are mounted on the power transmitting unit.
10. The system as claimed in claim 1 wherein the collection of operational data begins at the sensors mounted on the fixture.
11. The system as claimed in claim 1 additionally comprising a microcontroller which is responsible for converting the sensor data to a format that can be processed by the software.
12. A method to automatically monitor vibration, pressure and component presence on a work-piece holding fixture during the machining process within a specific period comprising: a) Collecting, by means of data collecting unit, of operational data from the fixture sensor inputs. b) Transmitting, by means of data transmission unit, of such collected data, to the server. Storing such transmitted data on the data storage unit. c) Analysing such transmitted data, by means of the analysis unit, for the determination of operational parameters based on the transmitted data

d) Comparing, by means of the evaluation unit, the transmitted data pertaining to operational data and the operator input data against corresponding historical data in a historical data repository e) Determining, by means of logic unit, of deviations in the operational data from the reference data with respect to a particular parameter f) enabling a prediction analysis with the captured data based on pre-set information provided by the user, and g) generating alerts on the breach of a pre-set threshold values for various parameters. 13. The method as claimed in claim 12 wherein Collecting, by means of data collecting unit, of operational data from the fixture sensor inputs is conducted in real time during the course of the machining operation. 14. The method as claimed in claim 12 wherein the server is a remote server located at a different location then the fixture system. 15. The method as claimed in claim 12 wherein the visualisation of data and the generation of alerts are available to the operator or any person in real time during the course of execution of the machining process.
16.A method of conducting Idle/Run time analysis and labelling of machining operations using a smart work-piece holding fixture comprising the following steps:
a) Analog signal variations are captured by the sensors from the component on the fixture based on the machining operation being done on it.
b) Captured data is sent to the microcontroller and digitized through an analog to voltage convertor circuit.
26
c) Data from the microcontroller is then sent to an IoT (Internet of things) gateway device and gets stored in the system database.
d) The data stored in the database is then analysed to give insights on any or all of the below:
i) track tool wear
ii) detect the type of machining operation
iii) detect tool breakage
iv) fault detection
v) give production efficiency
e) The analysed data is then transmitted in real time to a computer or handheld device and displayed to the end user. f) Machine operation classification analysis is then performed for the acquired data, which comprises the following steps: - A portion of the data is correlated to the pre-set information provided by the customer regarding the machining operations to be performed in that time period, - the data is trained into a certain category to inform the user about the type of machining operation taking place. - Once the classification training is done, the model is tested with the remaining data to determine accuracy. g) This step is repeated till an acceptable level of accuracy is attained, after which the work is deployed into the live server. 17. The method as claimed in claim 16 wherein the Internet of Things (IoT) Gateway device plays the role of the local server on the system environment and is the link between the system and the external network for the data transfer.

18. The method as claimed in claim 16 wherein along with the data captured, the software code stores a unique identifier for each sensor which helps determine the fixture it is associated with as well as the parameter type it is capturing.