Method and computer system for calibrating plurality of state parameters of gearbox
By using a neural network model to automatically calibrate the gearbox state parameters, the problems of long time consumption and large subjective influence in traditional methods are solved, and more efficient and accurate calibration results are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-09
- Publication Date
- 2026-03-31
AI Technical Summary
Traditional automatic transmission calibration methods are time-consuming, labor-intensive, and the results are greatly affected by subjectivity, especially in multi-gear transmissions.
A neural network model is used to calibrate the state parameters of the gearbox. By inputting the first state parameter and the second state parameter, the output of the neural network model is used to predict the degree of deviation, and the first state parameter is automatically adjusted to achieve the calibration effect.
It improves the objectivity and accuracy of calibration results, saves manpower, material resources and time costs, and enhances calibration efficiency.
Smart Images
Figure CN121762234A_ABST
Abstract
Description
Technical Field
[0001] This application relates to automotive calibration, and more specifically to methods, computer systems, storage media, and computer program products for calibrating transmissions. Background Technology
[0002] The calibration of automatic transmissions has a crucial impact on various indicators of transmission shifting performance. Traditional automatic transmission calibration uses manual methods, which suffer from problems such as high workload, long processing time, and significant susceptibility to subjective judgment in the calibration results. With the continuous increase in the number of gears in modern transmissions, the workload of calibration has increased significantly, making these problems even more prominent. Summary of the Invention
[0003] To address or at least alleviate one or more of the above problems, the following technical solutions are provided.
[0004] According to one aspect of this application, a method for calibrating multiple state parameters of a transmission is provided, the method comprising:
[0005] A. Input the first state parameter and the second state parameter associated with the gearbox into the neural network model, wherein the first state parameter is a settable value and the second state parameter is a fixed value;
[0006] B. Determine whether the current set value of the first state parameter needs to be adjusted, at least based on the output of the neural network model, wherein the output of the neural network model is a predicted value of the deviation of the second state parameter before and after the gear shift operation; and
[0007] C. If adjustment is required, generate an adjustment value for the first state parameter, update the current setting value using the adjustment value, and return to step A. If no adjustment is required, determine the current setting value of the first state parameter as the calibration value.
[0008] According to an embodiment of this application, optionally, the first state parameter includes the clutch pressure of the transmission, and the second state parameter includes the turbine speed of the transmission. Further, the second state parameter also includes the torque of the transmission, and the predicted value of the deviation is a predicted value of the deviation of the turbine speed before and after the shift operation.
[0009] According to an embodiment of this application, optionally, in step B, it is determined whether the current setting value of the first state parameter needs to be adjusted in the following manner:
[0010] If the predicted deviation exceeds a set threshold, it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
[0011] According to an embodiment of this application, optionally, in step B, it is determined whether the current setting value of the first state parameter needs to be adjusted in the following manner:
[0012] If the predicted deviation exceeds a set threshold and the number of times the current setting value of the first state parameter is adjusted does not exceed a set number, then it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
[0013] According to one aspect of this application, a computer system is provided, the computer system comprising: at least one memory; at least one processor; and a computer program stored in the memory and executable on the processor, the execution of the computer program on the processor causing the following operations:
[0014] A. Input the first state parameter and the second state parameter associated with the gearbox into the neural network model, wherein the first state parameter is a settable value and the second state parameter is a fixed value;
[0015] B. Determine whether the current set value of the first state parameter needs to be adjusted, at least based on the output of the neural network model, wherein the output of the neural network model is a predicted value of the deviation of the second state parameter before and after the gear shift operation; and
[0016] C. If adjustment is required, generate an adjustment value for the first state parameter, update the current setting value using the adjustment value, and return to step A. If no adjustment is required, determine the current setting value of the first state parameter as the calibration value.
[0017] According to one aspect of this application, a computer-readable storage medium is provided that stores instructions which, when executed by a processor, implement the method described above.
[0018] According to one aspect of this application, a computer program product is provided, the computer program product comprising a computer program that, when executed by a processor, implements the method described above.
[0019] The gearbox calibration scheme according to one or more embodiments of this application can make the calibration results more objective and accurate than manual calibration, improve efficiency, and save manpower, material resources and time costs. Attached Figure Description
[0020] The above and other features, aspects and advantages of this application will become better understood when the following detailed description is read with reference to the accompanying drawings, in which the same or similar elements are denoted by the same reference numerals, wherein:
[0021] Figure 1 This is a schematic diagram of a neural network model according to an embodiment of this application;
[0022] Figure 2 This is a flowchart illustrating the calibration of clutch pressure using a neural network model according to an embodiment of this application;
[0023] Figure 3 This is a schematic diagram of a computer system that uses a neural network model to calibrate clutch pressure according to an embodiment of this application. Detailed Implementation
[0024] The embodiments of this application are described below with reference to the accompanying drawings. It should be noted that the embodiments described in this application are merely exemplary and are not intended to limit the scope of the claimed subject matter. The technical solutions described in this application can be implemented in different forms without departing from the spirit and scope of this application. In the following detailed description of the embodiments, numerous specific details are set forth to provide a more thorough understanding of the disclosures of this application. However, in some embodiments, well-known features have not been described in detail to avoid unnecessarily complicating the description.
[0025] In this application, terms such as "comprising" indicate that, in addition to the units and steps that are directly and explicitly stated in the specification and claims, the technical solution described in this application does not exclude the presence of other units and steps that are not directly or explicitly stated.
[0026] The use of terms such as “first” or “second” is not intended to imply or establish any particular ordering of elements, nor is it intended to limit any element to a single element unless explicitly stated through the use of terms such as “before,” “after,” “single,” or other such terms. Rather, the use of such terms is to distinguish between elements. For example, the first element is different from the second element.
[0027] In this application, "deviation degree" refers to the degree to which the value of a state parameter when it approaches a steady state deviates from a set target value for that state parameter, such as, but not limited to, clutch pressure, turbine speed, and torque. For example, after a gearbox completes a shift operation, the turbine speed tends to stabilize. The turbine speed measured at this stable state may deviate from the set target turbine speed, and this deviation can be expressed in terms of absolute value or percentage. For instance, in one example, the measured turbine speed at stable state deviates from the set target turbine speed by 5 rpm or 0.5% (assuming the target turbine speed is 1000 rpm). The deviation degree may have a preset threshold, such as 10 rpm or 1%.
[0028] Where applicable, the order of the steps described herein may be altered, combined into compound steps, and / or divided into sub-steps to provide the features described herein. The versatility of the technical solutions described in this application is reflected in the calibration of various parameters of the transmission, including but not limited to clutch pressure. The following example, calibrating the clutch pressure during the speed control phase of an automatic transmission shift, will illustrate how to utilize neural networks to calibrate the transmission.
[0029] For automatic transmissions, during gear shifts, the turbocharger speed can be smoothly increased or decreased from one speed to another within a desired timeframe by controlling the clutch pressure. This process is called speed control. Taking downshifting as an example, the turbocharger speed increases during downshifting, and one of the purposes of calibrating the clutch pressure is to match the actual increase in turbocharger speed with the preset increase in turbocharger speed before the gear shift.
[0030] Calibrling clutch pressure using traditional manual methods is extremely time-consuming. In some embodiments of this application, efficiency and calibration accuracy can be greatly improved by utilizing algorithms such as machine learning (e.g., neural network models) to calibrate transmission parameters like clutch pressure.
[0031] In some specific implementations, the data used to train the neural network model for calibration includes, for example, one or more of the following: engine torque, engine speed, stable turbine speed at the end of the speed control phase, preset turbine speed, initial turbine speed at the start of the speed control phase, clutch torque, clutch pressure, etc. This data can be obtained by repeatedly performing a gear shift operation (e.g., downshifting from 6th to 5th gear) at different vehicle speeds and different accelerator pedal openings, and the data covers, for example, different turbine speed ranges and clutch torque ranges.
[0032] For example, the following description pertains only to the case where the transmission turbine speed (e.g., the stable turbine speed at the end of the speed control phase, the preset turbine speed, and the initial turbine speed at the beginning of the speed control phase), clutch torque, and pressure are used as training data. However, those skilled in the art will recognize that fewer or more types of data, or other types of data or one or more combinations of data, can be used as training data to better suit specific application scenarios.
[0033] In some specific examples (such as the calibration of clutch pressure), in addition to clutch pressure, the input to the neural network model can also be one of the following: i) the turbine speed of the transmission; ii) the clutch torque; and iii) both the turbine speed and torque.
[0034] In one specific example, N gear shifts are performed to obtain N sets of shift data for training a neural network model. Each set of shift data may include an initial turbo speed (the turbo speed at the start of the shift), a stable turbo speed (the turbo speed at the end of the shift), clutch torque, and clutch pressure. In another example, the stable turbo speed can be replaced by the difference between the initial turbo speed and the stable turbo speed.
[0035] It should be noted that clutch torque and clutch pressure can be determined in various ways. For example, clutch torque typically depends on the size and coefficient of friction of the friction plates, the pressure plate clamping force, and the number of friction pairs. In some examples, clutch torque can be simplified to the product of static friction force and the radius of the friction plates, where static friction force is determined by the clutch pressure and the friction plate coefficient, and pressure is generated by the pressure plate on the clutch. In other examples, the clutch pressure can be determined as the ratio of the clutch clamping force to the total friction area of the clutch driven plate. In summary, in practical applications, clutch torque and pressure can be determined based on specific application conditions, combined with theoretical calculations and experimental verification.
[0036] It should also be noted that the values of clutch torque and clutch pressure may vary during each gear shift. In some specific implementations, for each set of shift data, the average clutch torque during the corresponding shift operation can be used to characterize the clutch torque during that shift operation, and the average clutch pressure during the corresponding shift operation can be used to characterize the clutch pressure during that shift operation. Thus, a dataset {M} of N sets of shift data can be obtained. i e i n i p i}, where M i Let e be the clutch torque in the i-th shift data set (e.g., the average clutch torque mentioned above). iLet n be the difference between the stable turbo speed and the initial turbo speed in the i-th shift data set. i p represents the initial turbo speed in the i-th shift data set. i The clutch pressure in the i-th set of shift data (e.g., the average clutch pressure mentioned above).
[0037] Figure 1 This is a schematic diagram of a neural network model according to an embodiment of this application.
[0038] Figure 1 The neural network model 100 shown includes an input layer, intermediate hidden layers, and an output layer. The input layer may have multiple input nodes (e.g., 3), the output layer may have one or more output nodes (e.g., 1), and the intermediate hidden layers may contain one or more layers, with no limit on the number of nodes in each hidden layer. See also... Figure 1 The clutch torque M, the initial turbine speed n, and the clutch pressure p are respectively input to their respective input nodes, and the deviation or difference between the stable turbine speed and the initial turbine speed n is used as the output value of the output node.
[0039] Further description below Figure 1 The training method of the neural network model is shown. As mentioned above, N sets of shift data can be obtained by performing N shift operations. For the i-th set of shift data, its value M can be used. i n i and p i Construct vector X i =[M i ,n i ,p i For N sets of shift data, an N×3 matrix X can be obtained:
[0040]
[0041] Furthermore, for N sets of shift data, the deviation e between their respective stable turbo speeds and preset turbo speeds can be utilized. i Construct an N×1 column vector Y:
[0042]
[0043] Optionally, in some specific embodiments, matrices X and Y are further processed (e.g., normalized) to obtain matrices X' and Y':
[0044]
[0045]
[0046] Among them, X minLet X be a vector consisting of the minimum values of each column in matrix X. max Y is a vector consisting of the maximum values of each column in matrix X. min Let Y be the minimum value in the column vector Y. max This represents the maximum value in the column vector Y.
[0047] In some other specific implementations, matrices X (or X') and Y (or Y') can be used as the input and output of the neural network model 100 for training, respectively. Various training algorithms can be used, which will not be elaborated here.
[0048] It should be noted that, in Figure 1 In the neural network model shown, the clutch torque M, initial turbine speed n, and clutch pressure p are taken as inputs, and the deviation between the stable turbine speed and the preset turbine speed is taken as the output. Although the stable turbine speed could be used instead of the aforementioned deviation as the output of the neural network model, or the stable turbine speed could be taken as input and the clutch torque as the output, the inventors of this application, after research, discovered that, in comparison, Figure 1 The model shown has better performance in terms of both training speed (or convergence speed) and calibration accuracy.
[0049] Figure 2 This is a flowchart illustrating a method for calibrating clutch pressure using a neural network model according to an embodiment of this application. Exemplarily, Figure 2 The method shown takes the calibration of clutch pressure as an example, and employs... Figure 1 The calibration is performed using the trained neural network model shown.
[0050] It should be noted that different initial turbine speeds and clutch torques can form a series of combinations, through... Figure 2 The calibration process shown determines the corresponding clutch pressure for each of these combinations. For simplicity, Figure 2 The method shown only describes the clutch pressure calibration process for one combination of initial turbine speed and clutch torque. However, those skilled in the art will understand that by performing multiple iterations... Figure 2 The calibration process shown can be used to calibrate clutch pressure for multiple combinations.
[0051] Figure 2 The method shown begins at step 210. In this step, the apparatus used to perform calibration (e.g., Figure 3 The computer system shown determines the undetermined calibration value p of the clutch pressure corresponding to the combination of the initial turbine speed n (e.g., 1000 rpm) and the clutch torque M (e.g., 10 Nm).
[0052] For the first execution of step 210, an initial value (e.g., 15 kPa) can be assigned to the calibration value p to be determined, thereby obtaining the vector X = [n, M, p] as the input to the neural network model 100. Optionally, the vector X can be normalized using the above equation (3) to obtain the normalized vector X' = [n', M', p'] as the input to the neural network model 100. In step 210, any value within the range of possible clutch pressure values can be selected as the initial value. In some specific embodiments, the corresponding initial value can be selected based on an existing clutch pressure calibration curve (which is obtained based on historical data), for example, the clutch pressure value corresponding to the specific combination mentioned above on the existing clutch pressure calibration curve can be determined as the initial value.
[0053] If step 210 has already been performed, when performing step 210 now, the undetermined calibration value p of the clutch pressure can be updated to the adjusted value of the clutch pressure determined in step 250.
[0054] After step 210, Figure 2 The process shown proceeds to step 220. In step 220, the device for calibration inputs the vector X or X' constructed in step 210 into the neural network model 100 to obtain the output, namely, the predicted value e of the deviation or difference between the stable turbine speed and the initial turbine speed n. Optionally, the obtained predicted value e can be normalized using the above equation (4) to obtain the normalized predicted value e'.
[0055] Next, in step 230, the calibration device compares the predicted value e or e' with a set threshold TH (e.g., 10 rpm) to determine the output of the step neural network model 100. In some specific embodiments, if the predicted value e or e' does not exceed the threshold TH, it indicates that the undetermined calibration value p of the clutch pressure determined in step 210 meets the calibration requirements, and therefore proceeds to step 240; otherwise, it proceeds to step 250.
[0056] In step 240, the device used to perform the calibration determines the calibration value p to be determined as the final clutch pressure calibration value.
[0057] In step 250, the calibration apparatus adjusts the calibration value p to be calibrated. Those skilled in the art will understand that various adjustment algorithms or strategies can be employed to adjust the calibration value p. For example, in some embodiments, the calibration value p can be adjusted to p' by superimposing a fixed or varying step size on the calibration value p along the addition / subtraction direction. As described above, this adjustment value will be used in step 210 to update the current calibration value p.
[0058] After completing step 250, Figure 2 The process shown will then proceed to step 210.
[0059] exist Figure 2 In a variation of the process shown, step 230 can be modified as follows:
[0060] If the predicted value e or e' exceeds the threshold TH and Figure 2 If the number of iterations in the method flow shown does not exceed the set number TH', then proceed to step 250; otherwise, proceed to step 240.
[0061] Figure 3 This is a schematic diagram of a computer system that uses a neural network model to calibrate clutch pressure according to an embodiment of this application. Figure 3 As shown, the computer system 300 includes at least one memory 310 (such as a non-volatile memory like flash memory, ROM, hard disk drive, magnetic disk, optical disk, etc.), at least one processor 320, and a computer program 330 stored on the memory 310 and executable on the processor 320. By running the computer program stored on one or more memories on one or more processors (e.g., in a multi-processor cooperative manner or in a single-processor individual manner), the processor 320, when executing the computer program 330, implements one or more steps or operations of a method for calibrating clutch pressure using a neural network model according to an embodiment of this application.
[0062] Additionally, the technical solution of this application can also be implemented as a computer-readable storage medium, in which instructions, when executed by, for example, a processor, implement a method for calibrating clutch pressure using a neural network model according to an embodiment of this application. The computer-readable storage medium described herein may include RAM, ROM, EPROM, E2PROM, registers, hard disk, removable disk, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other temporary or non-temporary storage medium capable of carrying or storing program code units in the form of instructions or data structures and accessible by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Combinations of the above should also be included within the scope of protection of computer-readable storage media. Exemplarily, the storage medium is coupled to a processor so that the processor can read / write information from / to the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside, for example, in an ASIC. The ASIC may reside, for example, in a user terminal. In an alternative, the processor and storage medium may reside as discrete devices, for example, in a user terminal.
[0063] Furthermore, the technical solution of this application can also be implemented as a computer program product, which includes a computer program that, when executed by, for example, a processor, implements a method for calibrating clutch pressure using a neural network model according to an embodiment of this application. The computer program can be stored on one or more computer-readable storage media. It is also contemplated that the above-described computer program can be executed using one or more networked and / or otherwise general-purpose or special-purpose computers and / or computer systems.
[0064] Where applicable, the embodiments provided in this application may be implemented using hardware, software, or a combination of hardware and software. Furthermore, where applicable and without departing from the scope of this application, the various hardware and / or software components described herein may be combined into composite components comprising software, hardware, and / or both. Where applicable and without departing from the scope of this application, the various hardware and / or software components described herein may be divided into sub-components comprising software, hardware, or both. Additionally, where applicable, it is contemplated that software components may be implemented as hardware components, and vice versa.
[0065] The embodiments and examples presented herein are provided above to best illustrate embodiments according to the present technology and its particular applications, thereby enabling those skilled in the art to implement and use the present application. Those skilled in the art will understand that the above description and examples are provided for ease of illustration and example only. The descriptions presented are not intended to cover all aspects of the present application or to limit the present application to the precise forms disclosed.
Claims
1. A method for calibrating multiple state parameters of a transmission, the method comprising: A. Input the first state parameter and the second state parameter associated with the gearbox into the neural network model, wherein the first state parameter is a settable value and the second state parameter is a fixed value; B. Determine whether the current set value of the first state parameter needs to be adjusted, at least based on the output of the neural network model, wherein the output of the neural network model is a predicted value of the deviation of the second state parameter before and after the gear shift operation; and C. If adjustment is required, generate an adjustment value for the first state parameter, update the current setting value using the adjustment value, and return to step A. If no adjustment is required, determine the current setting value of the first state parameter as the calibration value.
2. The method as described in claim 1, wherein, The first state parameter includes the clutch pressure of the transmission, and the second state parameter includes the turbine speed of the transmission.
3. The method as described in claim 2, wherein, The second state parameter also includes the torque of the transmission, and the predicted value of the deviation is the predicted value of the deviation of the turbine speed before and after the shift operation.
4. The method as described in any one of claims 1-3, wherein, In step B, it is determined whether the current setting value of the first state parameter needs to be adjusted in the following manner: If the predicted deviation exceeds a set threshold, it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
5. The method as described in any one of claims 1-3, wherein, In step B, it is determined whether the current setting value of the first state parameter needs to be adjusted in the following manner: If the predicted deviation exceeds a set threshold and the number of times the current setting value of the first state parameter is adjusted does not exceed a set number, then it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
6. A computer system, comprising: At least one memory; At least one processor; as well as A computer program stored in the memory and executable on the processor, the execution of which causes the following operations: A. Input the first state parameter and the second state parameter associated with the gearbox into the neural network model, wherein the first state parameter is a settable value and the second state parameter is a fixed value; B. Determine whether the current set value of the first state parameter needs to be adjusted, at least based on the output of the neural network model, wherein the output of the neural network model is a predicted value of the deviation of the second state parameter before and after the gear shift operation; and C. If adjustment is required, generate an adjustment value for the first state parameter, update the current setting value using the adjustment value, and return to operation A. If no adjustment is required, determine the current setting value of the first state parameter as the calibration value.
7. The computer system as claimed in claim 6, wherein, The first state parameter includes the clutch pressure of the transmission, and the second state parameter includes the turbine speed of the transmission.
8. The computer system as claimed in claim 7, wherein, The second state parameter also includes the torque of the transmission, and the predicted value of the deviation is the predicted value of the deviation of the turbine speed before and after the shift operation.
9. The computer system according to any one of claims 6-8, wherein, In operation B, determine whether the current setting of the first state parameter needs to be adjusted in the following manner: If the predicted deviation exceeds a set threshold, it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
10. The computer system according to any one of claims 6-8, wherein, In operation B, determine whether the current setting of the first state parameter needs to be adjusted in the following manner: If the predicted deviation exceeds a set threshold and the number of times the current setting value of the first state parameter is adjusted does not exceed a set number, then it is determined that the current setting value of the first state parameter needs to be adjusted; otherwise, it is determined that the current setting value of the first state parameter does not need to be adjusted.
11. A computer-readable storage medium storing instructions that, when executed by a processor, implement the method of any one of claims 1-5.
12. A computer program product comprising a computer program that, when executed by a processor, implements the method of any one of claims 1-5.