METHOD AND SYSTEM FOR REAL-TIME ESTIMATION OF PAYLOAD CARRIED BY A VEHICLE

FIELD OF THE INVENTION

The present invention relates generally to motor vehicles. The present invention specifically relates to a method and system for real-time estimation of payload carried by the motor vehicles, in particular heavy duty vehicles.

BACKGROUND OF THE INVENTION

Vehicle's dynamic behavior and performance may be significantly affected by vehicle loading condition, i.e. payload and road grade parameters. In particular, the loadings due to the payload and the road grade can be significant for the heavy duty vehicles such as truck. Vehicle dynamic response to driver steering and braking inputs may also be significantly affected by the vehicle payload. Therefore, several chassis control systems benefit from information regarding vehicle payload and it is needed to estimate the vehicle payload, which aids in controlling of dynamic behavior and performance of the vehicle.

Conventionally, the vehicles are equipped with self-leveling suspensions and pressure sensors, so that the vehicle payload can be estimated based on pressure measurements in air springs of the self-leveling suspensions. Moreover, many modern vehicles are equipped with electronically controlled suspensions, acceleration sensors and suspension deflection sensors, which measure relative position of wheels with respect to the vehicle body. Such electronically controlled suspensions generally require estimation of body motions, which results in detection of payload conditions for vehicle performance improvements. Further, active safety systems are used in the vehicle, which require sensors for measuring the vehicle dynamic response. These conventional systems utilize reference models to estimate payload conditions in order to improve vehicle performance and/or reduce unnecessary activations in the vehicle.

In addition, some other conventional vehicles calculate gross driving force by deducting the running resistance and road slope resistance from total driving force of the engine. By calculating the acceleration, the payload carried by the vehicle is estimated based on Newton's second law of motion. This conventional technique of vehicle payload estimation increases the quality demands on the parameter estimation of road gradient that can be estimated by using GPS systems.

The estimation of vehicle payload under running condition is very important in heavy duty commercial vehicles, but the conventional approaches of estimation of the vehicle payload are not real-time process. With respect to conventional payload estimation approaches, many sensors such as relative position sensor, lateral acceleration sensor, roll rate sensor, etc, are used to determine the vehicle payload under running conditions. It is beneficial to determine vehicle payload condition using sensor information that becomes more widely available in modern vehicles. However, these techniques not only increase the overall cost of the system, but also increase the error in predicting payload due to errors in estimating running resistance, air drag resistance and rolling resistance coefficients. Therefore, it is desirable to provide a method and system for real-time estimation of payload carried by a vehicle, which is capable to address and overcome the above disadvantages of conventional systems.

SUMMARY OF THE INVENTION

An object of the present invention Is to provide a method for estimating payload carried by a vehicle, which facilitates accurate estimation of the vehicle payload in a real-time manner.

Another object of the present invention is to provide a method for estimating payload carried by a vehicle, which eliminates the need for calculation of rolling resistance losses, air drag resistance losses and road gradient losses.

A further object of the present invention is to provide a system for estimating payload carried by a vehicle, which eliminates the need for any additional sensors and reduces overall cost of the system.

According to one aspect, the present invention, which achieves the objectives, relates to a method for estimating payload carried by a vehicle, comprising acquiring a set of vehicle parameters at every specific time interval, where the vehicle parameters comprises vehicle speed, engine speed, throttle percentage and torque. Estimated driving force can be evaluated based on analysis of rate of change of the vehicle parameters and neural network models for predetermined vehicle loads (10 Tonne, 15 Tonne, 20 Tonne, 25 Tonne), after filtering the vehicle parameters with respect to desired threshold conditions, where the desired threshold conditions comprises vehicle speed greater than speed threshold, preferably in the range of 30-70 kmph, engine speed greater than rpm threshold, preferably in the range of 700-1500 rpm, throttle greater than or equal to throttle threshold, preferably in the range of 20%-50%, increase in percentage of throttle and positive acceleration. All these threshold values are different for different vehicle models. Driving force can be calculated based on torque, engine speed and vehicle speed. Ratio of estimated driving force to calculated driving force can be calculated at a specific vehicle load in accordance with estimated driving force predicted in neural network models for different loads. Vehicle payload can be estimated by linear interpolation of the calculated ratios of estimated driving forces obtained from neural network models for different loads to calculated driving force. Such method facilitates accurate estimation of the vehicle payload in a real-time manner, and also eliminates the need for calculation of rolling resistance losses, air drag resistance losses and road gradient losses.

Furthermore, estimated driving forces are predicted for the given vehicle load using neural network models for different loads, based on analysis of rate of change of vehicle parameters. Several estimated payloads in a vehicle trip can be accumulated for statistical analysis using a standard Gauss fit process to derive payloads for an entire vehicle trip data. An iterative Gauss fit process can be implemented on the estimated payloads to display accurate payload carried by the vehicle by determining mean and deviation of Gauss fit curve applied on the estimated payloads.

In case of instantaneous vehicle payload estimation, middle value of the estimated vehicle payloads is configured and displayed as instantaneous vehicle payload, after sorting all the estimated payloads in ascending order.

According to another aspect, the present invention, which achieves the objectives, relates to a system for estimating payload carried by a vehicle, comprising an electronic control unit connected to a vehicle controller area network (CAN) bus for acquiring a set of vehicle parameters at every specific time interval, where the vehicle parameters comprises vehicle speed, engine speed, throttle percentage and torque. This electronic control unit is also used to transmit the estimated payload to other ECUs which are connected on the same CAN bus. A storage unit is accumulated with a set of neural network models for different vehicle loads trained with an estimated driving force predicted in relation to rate of change of vehicle parameters sampled at a specific vehicle load. A processing unit is in communication with the electronic control unit for calculating driving force from vehicle parameters, after filtering the vehicle parameters in accordance with desired threshold conditions, where the desired threshold conditions comprises vehicle speed greater than speed threshold, preferably in the range of 30-70 kmph, engine speed greater than rpm threshold, preferably in the range of 700-1500 rpm, throttle greater than or equal to throttle threshold, preferably in the range of 20%-50%, increase in percentage of throttle and positive acceleration.

The processing unit calculates ratio of estimated driving force to calculated driving force in accordance with the estimated driving force predicted in each neural network model for different loads, and estimating vehicle payload by linearly interpolating these calculated ratios.

In addition, the processing unit is in communication with a display unit to display the estimated payload of the vehicle. The system also comprises a power supply unit that is electrically connected to a vehicle battery for providing power supply to all components of the system. The vehicle payload estimation system is configured as a stand alone unit or part of vehicle cluster or part of vehicle body controller unit. The vehicle payload estimation system uses combination of neural network models for different loads with statistical analysis which eliminates the need for any additional sensors and reduces overall cost of the system.

The foregoing and other features and advantages of the Invention will become further apparent from the following detailed description of the presently preferred embodiments, read in conjunction with the accompanying drawings. The detailed description and drawings are merely illustrative of the invention, rather than limiting the scope of the invention being defined by the appended claims and equivalents thereof.

BRIEF DESCRIPTION OF THE DRAWINGS

Referring now to the drawings wherein the showings are for the purpose of Illustrating a preferred embodiment of the invention only, and not for the purpose of limiting the same.

FIG. 1 shows a hardware architecture diagram of a system for estimating payload carried by a vehicle, in accordance with an exemplary embodiment of the present invention;

FIG. 2 illustrates a schematic view of a neural network for the system used for estimating payload carried by a vehicle, in accordance with an exemplary embodiment of the present invention;

FIG. 3 illustrates a flow diagram of a method for estimating payload carried by a vehicle for a trip, in accordance with an exemplary embodiment of the present invention; and

FIG. 4 Illustrates a graphical representation depicting linear interpolation of vehicle payload from sample ratios, in accordance with an exemplary embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Referring to FIG. 1, a hardware architecture diagram of a system for estimating payload carried by a vehicle is illustrated, in accordance with an exemplary embodiment of the present invention. The present invention generally relates to a vehicle payload estimation system utilized for a heavy duty commercial vehicle, for example truck. Vehicle payload is one of important parameters considered for effective loading of the vehicle in fields, shifting gear in automatic transmission, operation of anti lock brake system, fuel injection map selection in engine control unit, so the estimation of vehicle payload under running condition is very important in heavy duty commercial vehicles. The payload estimating system utilizes a set of vehicle parameters, in particular vehicle speed, engine speed, throttle and torque data, as input parameters to estimate the vehicle payload. The components of the system process the input parameters and carries out the required computations to fully meet requirements and specifications of the system.

The system is arranged with an electronic control unit (10), a processing unit (20) and a power supply unit (50). The electronic control unit (10) is optionally Implemented inside the payload estimation system or the payload estimation system utilizes inbuilt electronic control units (ECU) or CAN controller (10) of the vehicle. Hereafter, it is referred as electronic control unit utilized inside the estimation system only for the purpose of explanation. The electronic control unit (10) is connected directly to a controller area network (CAN) backbone of the vehicle to receive vehicle speed, engine speed, torque and throttle data via CAN backbone. The sampled data from CAN backbone can be recorded at a specific time interval, for example, 20 millisecond. The engine speed, vehicle speed, throttles and torque data are processed and fed to the processing unit (20), preferably microprocessor, for performing the required computations. All these input parameters required for the estimation system are assumed to be available on the vehicle's CAN data bus. The power supply unit (50) is electrically connected to a vehicle battery (60) to provide power supply to the components of the estimation system.

The processing unit (20) is in communication with the electronic control unit (10), such that the vehicle parameters, i.e. vehicle speed, engine speed, torque and throttle data of the vehicle can be filtered in accordance with desired threshold conditions, in particular velocity should be greater than speed threshold value, preferably in the range of 30-70 kmph, throttle should be greater than or equal to throttle threshold value, preferably in the range of 20%-50%, throttle should be strictly monotonically increasing, engine speed greater than rpm threshold value, preferably in the range of 700-1500 rpm, and acceleration should always be positive. The processing unit (20) calculates acceleration for less time duration for example 500 milliseconds, demands very high precision of speed. Since high precision of speed is not supported by CAN, vehicle acceleration can be calculated using Taylor's series as follows.

It minimizes the effect of noise in velocity and increases the accuracy of vehicle acceleration.

The processing unit (20) is configured to calculate acceleration, rate of change of throttle, rate of change of torque and average force for continuous sampled data over certain time period. In particular, the processing unit (20) calculates driving force from the filtered vehicle parameters using an equation as follows,

where F Is vehicle driving force generated by vehicle engine, N is engine speed in RPM, T is torque generated by vehicle engine and v is vehicle velocity. The engine speed, engine torque and vehicle velocity are continuously available from CAN bus in the vehicle.

Vehicle trails can be conducted with different loads or tonnages to record the CAN data for training a neural network (200). For each load, vehicle speed, engine speed, engine torque and throttle are recorded from the vehicle CAN bus on certain conditions such as minimum throttle, minimum vehicle speed and continuous increase of throttle.

The condition on minimum throttle ensures the proper clutch engagement and no application of brake, whereas the minimum vehicle speed ensures higher gear engagement. Similarly, the continuous Increase of throttle ensures the continuous increasing demand of engine torque.

Testing should be performed alternatively while training the neural network (200) to avoid overtraining for each vehicle load separately and create different models for different loads. The rate of change of torque, rate of change of throttle, total driving force and rate of change of speed (acceleration) can be calculated for certain time duration, preferably 500 milliseconds. All these calculated parameters are used to train the neural network (200). The neural network (200) models for different loads are used to estimate the payload with statistical approach. In the payload estimation system of the present invention, the neural network (200) is trained with rate of change of speed (acceleration), rate of change of torque, rate of change of throttle as inputs and vehicle driving force as output, as shown in FIG. 2, which illustrates a schematic view of a neural network (200) for the system for estimating payload carried by a vehicle, in accordance with an exemplary embodiment of the present invention. Thereby, weights of the neural network model for different loads are calculated and computed to predict the vehicle driving force corresponding to unknown payload input parameters. These weights of the neural network model for different loads can be stored in a storage unit (30), preferably EEPROM (Electronically Erasable Programmable Read-Only Memory). The processing unit (20) is associated with the storage unit (30) to access the predicted vehicle driving force recorded in each weight of the neural network model.

The processing unit (20) is coupled with a display unit (40), preferably liquid crystal display (LCD) for displaying the estimated vehicle payload to the vehicle user. The processing unit (20) is interfaced with a random access memory (RAM) (21), which stores the vehicle data to be readily accessible by the processing unit. The vehicle parameters are collected continuously and stored in the RAM (21) with a fixed sampling rate, preferably 20 milliseconds which are used for ail calculations. The payload estimation system is also arranged with a communication interface (70) and a programming interface (80), through which the processing unit (20) and the storage unit (30) are interfaced with vehicle components and external components for data communication. Such system achieves accurate and real-time estimation of the vehicle payload based on neural network models for different loads without using any additional sensors and without estimating air drag and running resistance coefficients.

The payload estimation can be carried by the moving vehicle without using any additional sensors. The payload estimation system utilizes real time engine data acquisition and performs a simple algorithm to compute the vehicle payload. The estimation system requires only minimal Inputs such as vehicle speed, engine speed, torque and throttle percentage to compute the vehicle payload, where these inputs can be obtained from the vehicle CAN bus. The estimation system can be designed as a stand alone unit or part of vehicle cluster or part of vehicle body controller unit which can interface with other electronic control units such as ABS, EMS and AMT. The present invention implements initial vehicle testing with different loads to record CAN data which is used to train the neural network (200) models for different loads. These models are used for estimating the total driving force of the vehicle for unknown load. Thus, it avoids the need of any additional sensors for payload estimation, which results in reduction in overall cost of the system.

Referring to FIG. 3, a flow diagram of a method for estimating payload carried by a vehicle for a trip is illustrated, in accordance with an exemplary embodiment of the present invention. The method aims at an algorithm which can estimate the payload carried by the vehicle without using any additional sensor and without being affected by road gradient. The vehicle parameters sampled for predefined time duration, preferably 500 milliseconds, are adequate enough for calculating the vehicle payload. The road gradient is negligible since the calculations are carried out with data captured for the predefined time duration. So, CAN data, i.e. vehicle parameters are collected continuously for some time in a buffer with a fixed sampling rate, preferably of 20 milliseconds, which is used for all calculations.

As specified in step 310, real-time vehicle parameters such as vehicle speed, engine speed in rpm, throttle percentage and vehicle torque, can be acquired at every specific time interval. As illustrated in step 320, the vehicle parameters can be filtered with respect to desired threshold conditions, i.e. where vehicle speed should be greater than speed threshold, preferably in the range of 30-70 kmph, engine speed should be greater than rpm threshold, preferably in the range of 700-1500 rpm, throttle should be greater than or equal to throttle threshold, preferably in the range of 20%-50%, and percentage of throttle should be increasing. As mentioned in step 330, estimated driving force can be evaluated based on analysis of rate of change of the vehicle parameters, once the vehicle parameters are filtered with respect to the desired threshold conditions.

The neural network models for different loads are stored in memory. When the rate of change of vehicle parameters of unknown payload are passed through the neural network model which is trained with vehicle parameters of the same load, then the ratio of estimated driving force to calculated driving force is close to 1. Similarly, when the rate of change of vehicle parameters of unknown vehicle payload are passed through the neural network model which is trained with vehicle parameters of higher payload, statistically the estimated driving force is more than the calculated driving force and vice versa. An example for ratios of estimated driving forces to calculated driving forces when rate of change of vehicle parameters of different payloads, are passed through neural network models for different loads are given in Table 1.

Table 1: Model ratios of estimated force to calculated force table

In order to estimate the vehicle payload, rate of change of corresponding vehicle parameters are passed to neural network models for different loads and then every model predicts the vehicle driving force. The driving force predicted In each neural network model is configured as estimated driving force, which is determined using neural network (200) based on analysis of rate of change of the vehicle parameters. As depicted in step 340, ratios of estimated driving forces to calculated driving force can be calculated for a specific vehicle load in accordance with estimated driving forces predicted for neural network models for different loads, i.e. ratios of estimated force to calculated force are calculated based for every neural network model's output. As shown in step 350, the vehicle payload can be estimated by linear interpolation of the calculated ratios of estimated driving force to calculated driving force.

In particular, unknown vehicle payload can be interpolated from the ratios of estimated driving force to calculated driving force. The practical example for ratios of estimated driving force to calculated driving force where estimated driving forces are obtained from neural network models for different loads when a set of 21 Tonne vehicle trails parameters is given as input are shown in Table 2.

Table 2: Ratios calculated with 21 Tonne vehicle data

The estimation system computes slope and offset of linear fit applied on these ratios with least squares sense. In this case, slope is 39.03 and offset is -17.396. The payload corresponding to ratio 1.00 on this line is the actual payload, and then the vehicle payload can be determined using an equation as follows,

Payload = slope x Ratio + offset

As the payload corresponding to ratio = 1.00, then the actual payload = 39.03 X 1.00 + (-17.396) = 21.634 Tonne. In this way, computations of all these sample ratios are plotted with respect to payload by linear interpolation to find out the vehicle payload corresponding to ratio 1.00, as shown in FIG. 4, which illustrates a graphical representation depicting linear Interpolation of vehicle payload from sample ratios, in accordance with an exemplary embodiment of the present invention.

In case of vehicle payload calculation for a vehicle trip, the estimation system collects CAN data continuously for small time duration, preferably of 500 milliseconds. This data set is used to compute the vehicle payload after filtration as described earlier. This computation repeats during whole trip. So at the end of trip, every filtered input dataset is provided with corresponding vehicle payload. All these estimated payloads forms a Gaussian distribution with the actual payload is centered on the mean of the distribution. So, at the end of the trip, the estimation system further analyzes the vehicle payloads statistically by using a standard Gauss fit process. In particular, as illustrated in step 360, several vehicle payloads in a vehicle trip can be accumulated for statistical analysis using the standard Gauss fit process to derive an entire vehicle trip data.

Thereafter, as depicted in step 370, an iterative Gauss fit process is implemented on the estimated payloads to display final trip payload carried by the vehicle. In every iteration, the estimation system determines mean and deviation of Gauss fit curve applied on the estimated payloads with least squares sense. The estimated payloads which are not in the range of sigma around mean of Gauss fit can be neglected, whereas the remaining estimated payloads in the bell shaped curve are used for next iteration with reduced intervals of payload for Gaussian distribution fit. The iterations are repeated until the number of remaining estimated pay load points are sufficient enough to form histogram. So, the vehicle payload in the trip can be recognized statistically from the mean of bell shaped or Gaussian fit curve which derives real close to the actual value.

Using this payload estimation method, the value of unknown payload can be obtained with the combination of neural network (200) models for different loads and statistical approach, where full trip vehicle data (8-10 hours) is needed for statistical approach. As this method utilizes neural network (200) to predict the gross driving force on the vehicle, there is no need of individually calculating the rolling resistance losses, air drag resistance losses and road gradient losses. Such combination of neural network (200) models for different loads with statistical models achieves more accurate estimation of vehicle payload.

Since the statistical approach cannot be applied for less number of data points, estimation of instantaneous vehicle payload can be derived from this method by means of separate technique which requires only five estimated payload values. Initially, first five estimated payload values are taken in a buffer and these estimated payload values are sorted in ascending order. After sorting in ascending order, middle value is taken as actual instantaneous payload to display the vehicle payload. Then, the estimation system neglects minimum and maximum values from the five point buffer and wait for the next two more estimated payload values to fill the buffer. By combining these two new payload values with the previous points, once again device sorts all five payload values in the buffer. After sorting in ascending order, device takes the middle value again as the final payload value and so on. The mass estimating device communicates instantaneous payload to other ECUs which are connected on CAN bus via CAN controller.

The foregoing description is a specific embodiment of the present invention. It should be appreciated that this embodiment is described for purpose of illustration only, and that numerous alterations and modifications may be practiced by those skilled in the art without departing from the spirit and scope of the invention. It is intended that all such modifications and alterations be included insofar as they come within the scope of the invention as claimed or the equivalents thereof.

WE CLAIM:

1. A method for estimating payload carried by a vehicle, comprising:

acquiring a plurality of vehicle parameters at every specific time interval;

evaluating estimated driving force using the neural network models for different loads based on analysis of rate of change of the vehicle parameters, after filtering the vehicle parameters with respect to desired threshold conditions;

calculating ratios of estimated driving force to calculated driving force in accordance with an estimated driving force predicted at a specific vehicle load in each neural network model for different loads; and

estimating vehicle payload by linear interpolation of the calculated ratios of estimated driving force to calculated driving force at specific vehicle load.

2. The method as claimed in claim 1, wherein each estimated driving force is predicted using the neural network models for different loads which are trained based on analysis of rate of change of vehicle parameters sampled for different vehicle loads.

3. The method as claimed in claim 1, further comprising:
accumulating several estimated payloads in a vehicle trip for statistical analysis using a standard Gauss fit process to derive an entire vehicle trip data; and

implementing an iterative Gauss fit process on the estimated payloads to display final trip payload carried by the vehicle by determining mean and deviation of Gauss fit curve applied on the estimated payloads.

4. The method as claimed in claim 1, wherein in case of instantaneous vehicle payload estimation, middle value of the estimated vehicle payloads is configured and displayed as instantaneous vehicle payload, after sorting a value of all the estimated payloads in ascending order.

5. The method as claimed in claims 1 and 2, wherein said plurality of vehicle parameters comprises vehicle speed, engine speed, throttle percentage and torque.

6. The method as claimed in claims 1 and 5, wherein the desired threshold conditions comprises vehicle speed greater than speed threshold, preferably in the range of 30-70 kmph, engine speed greater than rpm threshold, preferably in the range of 700-1500 rpm, throttle greater than or equal to throttle threshold, preferably in the range of 20%-50%, increase in percentage of throttle and positive acceleration.

7. A system for estimating payload carried by a vehicle, comprising:

an electronic control unit connected to a vehicle network bus for acquiring a plurality of vehicle parameters at every specific time interval;

a storage unit accumulated with a neural network having a set of neural network models for different vehicle loads trained with an estimated driving force predicted in relation to rate of change of vehicle parameters sampled at a specific vehicle load; and

a processing unit in communication with said electronic control unit for evaluating calculated driving force based on vehicle parameters, after filtering the vehicle parameters in accordance with desired threshold conditions,

wherein said processing unit is interfaced with said storage unit for calculating ratios of estimated driving force to calculated driving force in accordance with the estimated driving force predicted in each neural network model, and estimating vehicle payload by linearly interpolating the calculated ratios of estimated driving force to calculated driving force at specific vehicle load.

8. The system as claimed in claim 7, wherein said processing unit is in communication with a display unit to display the estimated payload of the vehicle.

9. The system as claimed in claim 7, further comprising a power supply unit electrically connected to a vehicle battery for providing power supply to all system components.

10. The system as claimed in claim 7, wherein said plurality of vehicle parameters comprises vehicle speed, engine speed, throttle percentage and torque.

11. The method as claimed in claim 7 and 10, wherein the desired threshold conditions comprises vehicle speed greater than speed threshold, preferably in the range of 30-70 kmph, engine speed greater than rpm threshold, preferably in the range of 700-1500 rpm, throttle greater than or equal to throttle threshold, preferably in the range of 20%-50%, increase in percentage of throttle and positive acceleration.

12. The system as claimed in claim 7, wherein the system is configured as a stand alone unit or part of vehicle cluster or part of vehicle body controller unit.