An adaptive estimation method for working position points of a pneumatic clutch, a storage medium and a software product
By constructing a neural network model based on pressure and pressure change, the problem of electromagnetic displacement sensors being easily damaged in harsh environments was solved, and accurate estimation of the pneumatic clutch position was achieved, ensuring the stability and reliability of vehicle gear shifting.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHAANXI FAST GEAR CO LTD
- Filing Date
- 2025-10-20
- Publication Date
- 2026-08-04
AI Technical Summary
In the existing technology, electromagnetic displacement sensors are easily damaged in high temperature, strong electric field, and strong magnetic environment, which can cause the vehicle to be unable to shift gears normally and affect the shifting operation of mechanical automatic transmissions.
By constructing a position estimation neural network model, the working position point of the pneumatic clutch is estimated by using the pressure value and pressure change in the working chamber of the PCA, combined with the recursive least squares method and the feedforward neural network. In particular, when the magnetostrictive displacement sensor fails, the position is estimated by using the air pressure sensor.
When the magnetostrictive displacement sensor fails, it achieves accurate estimation of the pneumatic clutch position, ensuring the stability and reliability of vehicle gear shifting. The position estimation error is within 2mm, meeting the usage requirements.
Smart Images
Figure CN121630926B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to vehicle component control technology, specifically to an adaptive estimation method, storage medium, and software product for the working position point of a pneumatic clutch. Background Technology
[0002] When shifting gears, an Automated Mechanical Transmission (AMT) requires controlling the disengagement and engagement of the clutch. Throughout the entire process of controlling the disengagement and engagement of the clutch, the current working position of the clutch needs to be monitored in real time.
[0003] like Figure 1 As shown, electronically controlled pneumatic clutches are commonly used in commercial vehicles and engineering vehicles. The engagement and disengagement of this type of clutch are achieved by controlling the pneumatic clutch actuator (PCA). Therefore, the working position of the clutch can be indirectly obtained by monitoring the working position of the PCA.
[0004] Existing technology uses magnetostrictive displacement sensors to measure the working position of the PCA (Pressure Conductor Actuator). These sensors measure displacement by monitoring changes in the magnetic field caused by the movement of the PCA's moving parts and by using an electromagnetic conversion circuit. However, in practical applications, this method often results in the electromagnetic displacement sensor failing or even being damaged due to high temperatures, strong electric fields, or strong magnetic environments, leading to the vehicle being unable to engage gears and breaking down. Summary of the Invention
[0005] To address the shortcomings or deficiencies of existing technologies, this invention provides an adaptive estimation method for the working position point of a pneumatic clutch.
[0006] Therefore, the adaptive estimation method for the working position point of a pneumatic clutch provided by the present invention obtains the working position of the clutch by estimating the position of the clutch actuator. The method includes using the following formula to estimate the current working position point of the clutch actuator. Make an estimate:
[0007]
[0008] in:
[0009] This represents the current pressure value within the PCA working chamber.
[0010] This represents the change in pressure within the PCA working chamber relative to the previous moment.
[0011] The position of the clutch actuator is the position at the previous moment; the initial position of the clutch actuator is the position at the start of operation.
[0012] To utilize the clutch actuator position values collected during the clutch engagement process Pressure value within the PCA working chamber Seeking The corresponding parameters in the relation;
[0013] To utilize the clutch actuator position values collected during the clutch disengagement process Pressure value within the PCA working chamber Seeking The corresponding parameters in the relation; The output of the position estimation neural network model includes the following method for constructing the position estimation neural network model: training the feedforward neural network using the collected pressure inside the PCA working chamber, the pressure change inside the PCA working chamber, and the position change of the clutch actuator, wherein the pressure inside the PCA working chamber and the pressure change inside the PCA working chamber are the inputs, and the position change of the clutch actuator is the output.
[0014] This is the working position threshold of the clutch actuator.
[0015] An alternative approach is to use a method with an estimation period of 1–2 ms. The data acquisition period for parameter determination is 1 to 2 ms. The data acquisition period for parameter determination is 1 to 2 ms.
[0016] Another option is to use the recursive least squares method to obtain... The parameters in the relation are obtained using the recursive least squares method. The parameters in the relation.
[0017] The present invention also provides a method for determining the working position of a pneumatic clutch. The method is used to determine the working position of the pneumatic clutch in an AMT vehicle. The method includes: using a magnetostrictive displacement sensor to collect the position of the pneumatic clutch actuator; when the magnetostrictive displacement sensor fails, using the above method to estimate the position of the clutch actuator to ensure stable gear shifting of the vehicle.
[0018] A storage medium related to this invention stores a computer program / instructions, characterized in that, when executed by a processor, the computer program / instructions implement the steps of the above-described method. A software product related to this invention includes a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.
[0019] This invention uses pressure data measured by a pressure sensor installed within the PCA and position data measured by a magnetostrictive displacement sensor to identify the parameters of the constructed position estimation equations for each stage, while simultaneously training the feedforward neural network online. This invention is particularly suitable for estimating the clutch operating position point based on the pressure value obtained from the air pressure sensor, the latest equation parameters obtained before the failure, and the trained network when the magnetostrictive displacement sensor malfunctions. Attached Figure Description
[0020] Figure 1 This is a schematic diagram of the structure of a pneumatic clutch and clutch actuator.
[0021] Figure 2 The diagram showing the relationship between PCA air pressure and clutch position obtained in the example is shown.
[0022] Figure 3 The diagram shows the air pressure position relationship during the clutch engagement process analyzed in the example.
[0023] Figure 4 The diagram shows the air pressure position relationship during the clutch disengagement process, as analyzed in the example.
[0024] Figure 5 To illustrate the estimation effect of the estimation function of the present invention during the clutch engagement process in the embodiment, the blue line (data1) in the figure is the relationship curve between the position value and pressure measured by the magnetostrictive displacement sensor during the vehicle operation process, and the red line (recursive) is the relationship curve between the position value and pressure obtained based on the method of the present invention.
[0025] Figure 6 To illustrate the estimation effect of the estimation function of the present invention during the clutch disengagement process in the embodiment, the blue line (data1) in the figure represents the relationship curve between the position value and pressure measured by the magnetostrictive displacement sensor during the vehicle operation process, and the red line (cubic) represents the relationship curve between the position value and pressure obtained based on the method of the present invention.
[0026] Figure 7 This is a schematic diagram of the neural network structure used in the embodiment.
[0027] Figure 8 The example shows the effect of the neural network on the position estimation during the clutch position fine-tuning process (the actual value is the value collected during vehicle movement, and the predicted value is the value estimated by the method of this invention). Detailed Implementation
[0028] Unless otherwise specified, the scientific and technical terms used in this article are intended for understanding by those skilled in the art.
[0029] The following detailed description of an adaptive estimation method for the working position point of a pneumatic clutch according to the present invention, with reference to specific embodiments and accompanying drawings, will provide a further detailed description.
[0030] Example:
[0031] This embodiment takes an electro-pneumatic central push clutch as an example, whose actuator is as follows: Figure 1 As shown, when the clutch is engaged, the intake valve opens and the exhaust valve closes, causing the clutch actuator to push the clutch to disengage. When the intake valve opens again and the exhaust valve closes, the clutch actuator retracts, and the clutch returns to the engaged state under the action of the return spring. During the process of the clutch actuator pushing the clutch, the distance pushed represents the clutch disengagement distance, and this distance parameter is an important reference for clutch control. The clutch actuator is equipped with a displacement measurement sensor and a pressure monitoring sensor.
[0032] During clutch operation, the gas pressure value inside the PCA and the clutch actuator position value obtained by the displacement sensor at the corresponding time are collected (collection period is 1ms). A graph showing the relationship between gas pressure and position inside the PCA working chamber is then plotted. Figure 2 As shown in the figure, it is clear that the relationship between air pressure and position is not monotonic; a monotonic mapping relationship cannot be obtained by directly using air pressure and position data. Further, the position data is divided into two processes: position rise and position fall, i.e., clutch disengagement and clutch engagement. The pressure-position relationship diagrams for these two processes are shown below. Figure 3 and Figure 4 As shown in the diagram, the relationship between air pressure and position is still non-monotonic, so the working process of the clutch needs to be further divided.
[0033] Based on the above analysis, this invention divides the clutch operation process into three stages according to the position data obtained from the PCA position sensor: clutch disengagement, clutch position fine-tuning, and clutch engagement. Based on vehicle driving experiments, data on normal clutch operation is collected and analyzed to obtain the clutch actuator's working position threshold L. When the position value continuously increases but is less than L, the obtained position and pressure data belong to the clutch disengagement process; when the position value continuously decreases but is less than L, the obtained position and pressure data belong to the clutch engagement process; and when the position value is greater than or equal to L or the pressure value remains unchanged, the obtained data belongs to the clutch position fine-tuning process.
[0034] Among them, the design of the pressure position estimation function equation and parameter fitting for the clutch engagement and disengagement process are as follows: There is a clear linear relationship between the position value of the clutch actuator and the pressure value in the PCA working chamber during the clutch engagement process. Based on the data characteristics, the functional relationship between the position value Posn and the air pressure value Pcca is designed as follows: Formula 1:
[0035]
[0036] During the clutch disengagement process, there is a significant nonlinear relationship between the position value and the pressure value inside the PCA working chamber. Based on experience, the functional relationship between the position value Posn and the air pressure value Pcca is as follows: (2)
[0037]
[0038] Using the position values of the clutch actuator and the pressure values inside the PCA working chamber collected during 75 engagement and disengagement processes (with a collection period of 1ms), the parameters of equations (1) and (2) are calculated using the recursive least squares method. In this embodiment... , , .
[0039] Training of the position estimation neural network for clutch position fine-tuning: The relationship between PCA pressure and clutch position during clutch position fine-tuning is constructed using a feedforward neural network. The network structure used in this embodiment is as follows: Figure 7 As shown, the network input is the pressure inside the PCA working chamber. and the change in PCA pressure The neural network output is the change in clutch position (measured using a pressure sensor to measure the pressure within the clutch actuator's working chamber, in kPa, and a magnetostrictive displacement sensor to measure the real-time position of the clutch actuator during operation, in mm). The dataset in this embodiment consists of the aforementioned data collected during vehicle operation, including at least 20 clutch engagement and disengagement processes. The data acquisition frequency is 1 ms, and the training to testing data ratio is 3:2. A 3-layer network structure is used, with j=5 nodes in the hidden layers. The network weights are trained using a backpropagation (BP) neural network algorithm. W ij It is the connection weight from the input layer to the hidden layer ( W i1j1 (This represents the connection weights between the first input layer and the first hidden layer). In this embodiment, the number of input layers i=2. W jk It is the connection weight from the hidden layer to the output layer ( W j2k1(This represents the connection weights between the second hidden layer and the first output layer). In this embodiment, the number of output layers k=1. The network uses the hyperbolic tangent activation function for the hidden layers and a linear function for the output layers. Whenever a new PCA pressure signal and position signal are obtained, the pressure change is calculated based on the current pressure signal and the pressure signal at the previous moment (the initial pressure change is set to 0 kPa). Then, the pressure and pressure change are input into the trained neural network to obtain the network-estimated position value Posn1.
[0040] To verify the reliability of the present invention, the pressure data within the working chamber collected during vehicle operation is input into the estimation formula or method for the corresponding stage to obtain the estimated position of the clutch actuator at that stage. The estimated position is then compared with the position of the clutch actuator collected by sensors during vehicle operation. Specifically, during the clutch engagement stage, the effect diagram of the method is shown below. Figure 5 As shown, the deviation between the clutch position value obtained by this method and the actual clutch value can be controlled within 2mm, which fully meets the usage requirements; the effect during clutch disengagement is shown in the diagram. Figure 6 As shown, the deviation between the position value obtained by this method and the actual value can be controlled within 2mm, which fully meets the usage requirements; the effect diagram of the estimation during the adjustment stage is shown in the figure. Figure 8 The estimated value of the clutch at this stage follows the actual value very well, with the maximum position deviation within 2mm, which fully meets the usage requirements.
Claims
1. An adaptive estimation method for the working position point of a pneumatic clutch, characterized in that, The method obtains the working position of the clutch by estimating the position of the clutch actuator. The method includes using the following formula to calculate the current working position of the clutch actuator. Make an estimate: in: This represents the current pressure value within the PCA working chamber. This represents the change in pressure within the PCA working chamber relative to the previous moment. The position of the clutch actuator is the position at the previous moment; the initial position of the clutch actuator is the position at the start of operation. To utilize the clutch actuator position values collected during the clutch engagement process Pressure value within the PCA working chamber Seeking The corresponding parameters in the relation; To utilize the clutch actuator position values collected during the clutch disengagement process Pressure value within the PCA working chamber Seeking The corresponding parameters in the relation; The output of the position estimation neural network model includes the following method for constructing the position estimation neural network model: training the feedforward neural network using the collected pressure inside the PCA working chamber, the pressure change inside the PCA working chamber, and the position change of the clutch actuator, wherein the pressure inside the PCA working chamber and the pressure change inside the PCA working chamber are the inputs, and the position change of the clutch actuator is the output. This is the working position threshold of the clutch actuator.
2. The adaptive estimation method for the working position point of the pneumatic clutch according to claim 1, characterized in that, The estimated period is 1 to 2 ms.
3. The adaptive estimation method for the working position point of the pneumatic clutch according to claim 1, characterized in that, The data acquisition period for parameter determination is 1 to 2 ms.
4. The adaptive estimation method for the working position point of the pneumatic clutch according to claim 1, characterized in that, The data acquisition period for parameter determination is 1 to 2 ms.
5. The adaptive estimation method for the working position point of a pneumatic clutch according to claim 1, characterized in that, The recursive least squares method is used to obtain The parameters in the relation.
6. The adaptive estimation method for the working position point of a pneumatic clutch according to claim 1, characterized in that, The recursive least squares method is used to obtain The parameters in the relation.
7. A method for determining the working position point of a pneumatic clutch, the method being used to determine the working position point of the pneumatic clutch in an AMT vehicle, characterized in that, The method includes: using a magnetostrictive displacement sensor to collect the position of the pneumatic clutch actuator; when the magnetostrictive displacement sensor malfunctions, using the method described in any one of claims 1-6 to estimate the position of the clutch actuator, thereby ensuring stable gear shifting of the vehicle.
8. A storage medium, characterized in that, It stores a computer program / instruction thereon, characterized in that when the computer program / instruction is executed by a processor, it implements the steps of the method described in any one of claims 1 to 6.
9. A software product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the method described in any one of claims 1 to 6.