AMT pneumatic clutch actuator starting control method

Abstract

The purpose of this invention is to address the problem that existing technologies suffer from poor real-time performance and robustness in control algorithms when facing complex and harsh working conditions, making it difficult to guarantee smooth start-up. This invention discloses a start-up control method for an AMT pneumatic clutch actuator. The steps include: reading and verifying the online status of sensors, engagement valves, and disengagement valves, and loading control parameters; collecting real-time sensor data and synchronizing it with time, then preprocessing the data and calculating the slope angle, vehicle mass, adhesion coefficient, and driver demand intensity, and then calculating the comprehensive working condition difficulty coefficient K before setting the start-up; determining whether the clutch has reached the torque transmission critical point, entering the adaptive slip control stage, executing three-layer closed-loop control until the speed difference between the engine speed and the transmission input shaft speed meets the preset conditions, then controlling the clutch to lock; after locking, performing a quality score and determining whether to update and save the control parameters based on the quality score result.

Description

Technical Field

[0001] This invention relates to a starting control method for a clutch actuator, specifically a starting control method for an AMT pneumatic clutch actuator. Background Technology

[0002] For engineering vehicles starting under complex and harsh working conditions, the clutch plays a crucial role in smoothly transmitting engine power and achieving a smooth transition from a standstill to motion. In traditional manual transmissions, the driver directly operates the clutch pedal in conjunction with the accelerator. However, in automatic mechanical transmissions (AMT), the engagement and disengagement of the clutch are completely automated by the controller and actuator. Therefore, in AMT, the control precision, response speed, and safety strategies of the actuator directly determine the starting quality, driving comfort, and service life of the transmission system.

[0003] Currently, there has been considerable research both domestically and internationally on clutch start-up control in automatic manual transmissions (AMTs). Chinese patent CN115352445A discloses a start-up control method for heavy-duty trucks using an AMT. This method sets target speeds for the engine and clutch, calculates the base torque and proportional-integral (PI) torque based on the speed difference between the two, sums them to obtain the clutch target torque, and then controls clutch engagement to ensure smooth start-up. Chinese patent CN117803670A provides an AMT clutch start-up control method. This method uses throttle opening and starting gear as the main parameters, and combines oil temperature, load, and other factors to optimize proportional-integral-derivative (PID) control parameters. It drives the solenoid valve based on the difference between the clutch target and actual positions, aiming to achieve precise clutch control and reduce wear.

[0004] However, the aforementioned existing technologies are mainly geared towards highway vehicles. When applied to complex and harsh working conditions such as mines and construction sites, engineering vehicles often face various highly nonlinear, strongly coupled, and highly disturbed conditions during the starting process, including steep slopes, approaching rated loads, sudden changes in road surface adhesion coefficients, and a sharp increase in resistance due to large steering angles. Furthermore, these conditions change rapidly, placing more stringent demands on the real-time performance and robustness of the control algorithm. Therefore, when the aforementioned starting control methods are applied to the starting control of engineering vehicles in complex and harsh working conditions such as mines and construction sites, their real-time performance and robustness are insufficient to guarantee the smoothness of the starting process. Summary of the Invention

[0005] The purpose of this invention is to solve the technical problem that the existing technology is difficult to guarantee the smoothness of starting of engineering vehicles when applied to the starting control of engineering vehicles in complex and harsh working conditions such as mines and construction sites, and to provide a starting control method for AMT pneumatic clutch actuator.

[0006] To achieve the above objectives, the technical solution provided by this invention is as follows:

[0007] A starting control method for an AMT pneumatic clutch actuator, wherein the clutch actuator includes a fast engagement valve, a slow engagement valve, a fast disengagement valve, and a slow disengagement valve; its distinctive feature is that it includes the following steps:

[0008] S1. Read and verify the online status of multi-source sensors, fast engagement valve, slow engagement valve, fast disengagement valve and slow disengagement valve, and establish a unified time reference, while loading the factory or saved control parameters; the control parameters include clutch estimation model parameters, pre-engagement point database, pre-engagement point estimation bias, K fusion weight, thermal model parameters and safety override parameters;

[0009] S2. Collect real-time sensing data from multiple source sensors and synchronize the real-time sensing data in time.

[0010] S3. The real-time sensing data after time synchronization is preprocessed, and the slope angle, vehicle mass, adhesion coefficient and driver demand intensity are calculated. Then, the comprehensive difficulty coefficient K1 of the working condition is calculated by using the slope angle, vehicle mass, adhesion coefficient, driver demand intensity, and steering wheel angle in the real-time sensing data and the K fusion weight.

[0011] S4. Based on the comprehensive difficulty coefficient K1, slope angle, vehicle weight, adhesion coefficient, and driver's required intensity, control the fast engagement valve and slow engagement valve to preset the start-up, so that the actual clutch pressure P act The engine output torque reaches a preset coordinated state at the moment of start-up;

[0012] S5. Based on real-time sensor data and actual clutch pressure P act The speed difference between the two sides of the clutch pressure plate and driven plate and the estimated torque of the clutch are calculated to determine whether the clutch has reached the critical point of torque transmission. If the critical point of torque transmission is reached, the adaptive slip control stage is entered.

[0013] S6. In the adaptive slip control phase, a three-layer closed-loop control is executed based on acceleration, torque, inner layer pressure, and slip. During this three-layer closed-loop control, safety override based on a risk index and graded thermal protection based on a thermal model are executed in parallel. This continues until the speed difference between the engine speed and the transmission input shaft speed in the real-time sensor data is less than a preset synchronization threshold and the vehicle's longitudinal acceleration change rate is less than a preset impact threshold. At this point, the fast engagement valve, slow engagement valve, fast disengagement valve, and slow disengagement valve are activated to adjust the actual clutch pressure P. act Increase the pressure to full engagement and control clutch lock-up;

[0014] S7. After the clutch locks up, i.e. the start-up is completed, the quality of the start-up process is scored. Based on the quality score, it is decided whether to update the control parameters and save them, thus completing the start-up control of the AMT pneumatic clutch actuator.

[0015] Furthermore, the process of synchronizing the real-time sensing data in step S2 includes:

[0016] Define a unique reference clock to time-stamp the real-time sensing data from multiple sensors;

[0017] The real-time sensor data from different sampling periods is resampled to the main control period using an interpolation / hold algorithm.

[0018] For real-time sensor data with fixed communication delays, a model is created in the end-to-end delay model for time compensation in steps S3-S6.

[0019] Furthermore, the preprocessing method for the time-synchronized real-time sensing data in step S3 includes:

[0020] Filtering and slope limiting, physical feasible interval comparison, bad value detection and handling, redundancy consistency comparison.

[0021] Furthermore, the process of calculating the overall difficulty coefficient K1 in step S3 includes:

[0022] The slope angle, vehicle mass, adhesion coefficient, driver demand intensity, and steering wheel angle from real-time sensor data are mapped to the [0,1] interval, and the slope factor, load factor, adhesion factor, driver demand factor, and steering factor are output.

[0023] Based on the confidence indices corresponding to the preprocessed slope angle, vehicle mass, adhesion coefficient, steering wheel angle, and driver demand intensity, a fusion algorithm is used to process them to obtain the comprehensive confidence indices corresponding to the slope angle, vehicle mass, adhesion coefficient, steering wheel angle, and driver demand intensity.

[0024] The original difficulty coefficient K0 is calculated by dynamically adjusting the K fusion weights corresponding to the slope factor, load factor, adhesion factor, steering factor and driver demand factor using the comprehensive confidence level as a weighting factor.

[0025] The overall difficulty coefficient K of the operating conditions in the previous control cycle is based on the control parameters loaded in step S1, either from the factory or stored data. prev The original difficulty coefficient K0 is used to obtain the final comprehensive difficulty coefficient K1 of the working condition after smoothing update, change rate limitation and shrinkage correction.

[0026] Furthermore, step S4 specifically includes:

[0027] S4.1. Based on the comprehensive difficulty coefficient K1 of the working condition, the slope angle, the vehicle mass, the adhesion coefficient and the driver's demand intensity, the pre-joint point PB is determined from the pre-joint point database. The pre-set pressure and pre-set torque are calculated, and the pre-set advance amount Δt is calculated based on the full-link delay model.

[0028] S4.2. Based on the preset pressure, preset torque, and preset lead time Δt, before the expected start-up moment, a preset pressure command is sent to the fast-engagement valve and slow-engagement valve of the clutch actuator, and an engine preset torque command is sent to the engine controller for start-up control, in order to ensure that the actual clutch pressure P... act The engine output torque and the engine reach a preset coordinated state at the start.

[0029] Furthermore, step S5 specifically includes:

[0030] The speed difference between the two sides of the clutch pressure plate and driven plate is calculated using the engine speed and gearbox input shaft speed from the real-time sensor data.

[0031] Based on the actual clutch pressure P act The estimated clutch torque is calculated using the engine speed, transmission input shaft speed and clutch temperature from the real-time sensor data.

[0032] Determining whether the clutch has reached the critical point of torque transmission: When the trend of engine speed change and the trend of speed difference change meet the preset conditions and the estimated torque of the clutch exceeds the critical torque threshold, it is determined that the clutch has reached the critical point of torque transmission, and then the adaptive slip control stage is entered.

[0033] Furthermore, in step S6, the three-layer closed-loop control process of acceleration, torque, inner layer pressure, and slip is specifically performed as follows:

[0034] ①Outer layer acceleration closed-loop control

[0035] Based on the driver's demand intensity and the overall difficulty coefficient K1 of the operating conditions, a desired longitudinal acceleration curve is set. By comparing the acceleration actually measured by the sensor with the corresponding expected value on the desired longitudinal acceleration curve, the outer layer correction torque ∆T is calculated. outer ;

[0036] ②Mid-level torque closed-loop control

[0037] The anti-rollback safety torque is calculated using wheel radius, slope angle, and vehicle mass.

[0038] Based on the intensity of driver demand, the required torque for the driver is obtained by looking up a table in combination with the engine torque characteristics.

[0039] The upper limit of the adhesion limit torque is calculated based on the wheel radius, vehicle mass, adhesion coefficient, slope angle, and longitudinal acceleration in the real-time sensor data.

[0040] Based on the outer layer modified torque ∆T outer The target clutch torque T is calculated based on the following parameters: anti-rollback safety torque, driver-required torque, upper limit of adhesion torque, and estimated clutch torque, according to the fusion rules. m The torque closed-loop correction is generated by a PI / PID control law with gain scheduling, so that the estimated clutch torque follows the target clutch torque T. m This is to ensure that the torque transmitted by the clutch is precisely controllable;

[0041] ③ Inner layer pressure and slip closed-loop control

[0042] Based on the pressure-torque characteristics of the clutch actuator, look up the table to determine the target clutch torque T. m Converted into feedforward pressure;

[0043] Based on the target clutch torque T m The error between the estimated clutch torque and the actual clutch torque generates a pressure correction amount;

[0044] The actual slip is calculated based on the engine speed and transmission input shaft speed in the real-time sensor data, and a slip correction amount is generated based on the error between the target slip and the actual slip in the driver's demand intensity.

[0045] The final target pressure is obtained by summing the feedforward pressure, pressure correction, and slip correction.

[0046] Calculate the actual clutch pressure P act The pressure error is compared with the final target pressure, and the equivalent target gas flow rate is calculated based on the pressure error and the rate of change of the target pressure.

[0047] The equivalent target airflow rate is decomposed into low-frequency and high-frequency target airflow rates using a frequency band decomposition algorithm. These are then calculated and converted into opening or duty cycle signals for the fast-action and slow-action valves, respectively, to control the actual clutch pressure P. act Follow the pressure of the ultimate goal.

[0048] Furthermore, the execution process of the safety bypass described in step S6 is as follows:

[0049] Based on the engine speed in the real-time sensor data, the engine stall risk index is calculated according to the degree and duration of the engine speed approaching the stall threshold.

[0050] Based on the clutch-estimated torque, and the slope angle, vehicle longitudinal velocity direction, and vehicle longitudinal acceleration direction in the real-time sensor data, the backward slip risk index is calculated.

[0051] The impact risk index is calculated based on the longitudinal acceleration in the real-time sensing data.

[0052] Determine whether the flameout risk index exceeds the high flameout risk threshold. If the flameout risk index exceeds the high flameout risk threshold, then execute a forced flameout prevention action by controlling the fast separation valve or the slow separation valve. If the flameout risk index does not exceed the high flameout risk threshold, or after executing the forced flameout prevention action to control the flameout risk index within the preset safe zone, determine whether the backflow risk index exceeds the backflow threshold.

[0053] If the slippage risk index exceeds the slippage threshold, then the slippage suppression action is executed, and the target torque lower limit is adjusted to be above the anti-slippage torque value. If the slippage risk index does not exceed the slippage threshold, then it is determined whether the impact risk index exceeds the impact threshold.

[0054] If the shock risk index exceeds the shock threshold, then anti-shock actions are executed; if the shock risk index does not exceed the shock threshold, then the current three-layer closed-loop control output is maintained.

[0055] Furthermore, in step S6, during the process of controlling clutch lock-up, the real-time impact risk index is continuously monitored. If the impact risk index rises above the impact threshold for a short period of time, the actual clutch pressure P is reduced by decreasing the opening degree or duty cycle of the fast engagement valve and the slow engagement valve. act The rise is to slow the rate of increase in the final target pressure;

[0056] After the clutch lock-up is completed, the system switches from the adaptive slip control phase to the steady-state pressure maintenance and monitoring phase to maintain sufficient connection force and prevent unnecessary continuous slip.

[0057] Furthermore, in step S7, the process of scoring the quality of this start-up process and determining whether to update and save the control parameters based on the quality score is as follows:

[0058] Record the number of safe overruns and their duration during this start-up process, and calculate the start-up quality score based on the rollback distance, total start-up time, minimum engine speed, maximum impact, and thermal load during this start-up process.

[0059] Determine whether the number and duration of safety overruns and the start-up quality score meet the update conditions. The update conditions are that safety overruns are not continuously triggered and the start-up quality score is higher than the preset quality score threshold.

[0060] If the update conditions are not met, the update process will end.

[0061] If the update conditions are met, an update permission instruction is output. The slope angle, vehicle mass, adhesion coefficient, driver demand intensity and steering wheel angle, pre-engagement point PB and start quality score of this start are used as candidate samples. Anomaly detection and distance measurement are performed on the candidate samples.

[0062] If the detection result of the candidate sample is a normal sample, the candidate sample is written into the pre-binding point database, and its sample weight is adjusted according to the density and importance of the candidate sample. The database is managed for lifecycle: sliding time window to ensure that the number and coverage of candidate samples are within a preset reasonable range.

[0063] If the start-up quality score does not decrease significantly and the safe override frequency does not increase within the preset time window, then a preset small-amplitude, bounded update is performed on some parameters of the control parameters, including the K fusion weight, the pre-engagement point estimation bias, and the torque estimator, and the update is saved; the significant decrease value is set empirically.

[0064] If the start-up quality score drops significantly or the safety override frequency increases within the preset time window, the system enters a learning freeze state, stops online learning updates, cancels all control parameter updates, and restores the system to the factory-calibrated control parameters or the last saved control parameters. The system will then prompt maintenance / replacement suggestions through the human-machine interface (HMI).

[0065] Compared with the prior art, the present invention has the following beneficial technical effects:

[0066] 1. The present invention provides a starting control method for an AMT pneumatic clutch actuator. Through quantitative calculation of the comprehensive difficulty coefficient K1 of the working condition, self-learning query of the pre-engagement point database, three-layer closed-loop control, safety override, and quality score update, the AMT pneumatic clutch actuator can achieve full-process coordinated control from working condition identification, starting preset, slip control to lock-up holding under complex and harsh working conditions of engineering vehicles, which significantly improves starting quality, transmission system reliability, and clutch durability.

[0067] 2. This invention normalizes the slope angle, vehicle mass, adhesion coefficient, steering wheel angle, and driver demand intensity, and performs confidence-weighted linear fusion. After smoothing updates, rate of change limits, and contraction corrections, it calculates the comprehensive difficulty coefficient K1 of the working condition. This ensures that the comprehensive difficulty coefficient K1 of the working condition remains stable even when there is uncertainty in the perception of multiple source sensors, avoiding sudden changes in control strategy caused by single sensor errors or instantaneous noise, and providing a reliable basis for subsequent mode decisions and parameter scheduling.

[0068] 3. This invention decouples the starting control target layer by layer by constructing a cascade control architecture consisting of an outer acceleration closed loop, a middle torque closed loop, and an inner pressure and slip closed loop: the outer acceleration closed loop sets and corrects the desired acceleration trajectory in real time based on the driver's demand intensity and the comprehensive difficulty coefficient K1 of the working conditions; the middle torque closed loop integrates the anti-rollback safety torque, the driver's demand torque, and the upper limit of the adhesion limit to determine the target clutch torque; the inner pressure and slip closed loop uses a dual-frequency band decomposition and equivalent airflow distribution algorithm to enable the fast engagement valve and the slow engagement valve to work together under different working conditions, achieving precise control of the target pressure and slip, forming a top-down cascade link of "setting the target - calculating torque - controlling pressure". At the same time, each layer forms a top-down feedback closed loop through measured acceleration, estimated torque, actual pressure, and slip, which enables layer-by-layer suppression capabilities when facing complex and harsh working conditions and other highly disturbing working conditions. Under the premise of ensuring that the risks of stalling, rollback, and impact are controllable, it achieves a coordinated unity of starting response speed, control accuracy, and ride comfort. Attached Figure Description

[0069] Figure 1 This is a flowchart of a starting control method for an AMT pneumatic clutch actuator according to the present invention;

[0070] Figure 2 This is a flowchart of steps S1-S5 in an embodiment of the starting control method for an AMT pneumatic clutch actuator of the present invention.

[0071] Figure 3 This is a flowchart of steps S6-S7 in an embodiment of the starting control method for an AMT pneumatic clutch actuator of the present invention. Detailed Implementation

[0072] To make the objectives, advantages, and features of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art should understand that these embodiments are merely used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0073] To address the technical challenges of poor real-time performance and robustness in existing control algorithms when facing various highly nonlinear, strongly coupled, and highly disturbed operating conditions such as steep slopes, near-rated loads, sudden changes in road surface adhesion coefficients, and large steering angles leading to a surge in resistance, thus failing to guarantee smooth start-up, this embodiment designs a closed-loop control system integrating perception, decision-making, execution, and safeguards. This system, specifically designed for AMT clutch actuators, incorporates detailed calculation rules for the comprehensive operating condition difficulty coefficient K1, dual-valve frequency band decomposition and equivalent airflow distribution algorithms, online updates and confidence rules for the pre-engagement point database, time synchronization, and safe override conflict decision-making. This provides a starting control method for AMT pneumatic clutch actuators, such as… Figure 1 As shown.

[0074] This embodiment provides a starting control method for an AMT pneumatic clutch actuator, wherein the clutch actuator includes a fast engagement valve, a slow engagement valve, a fast disengagement valve, and a slow disengagement valve; the control method includes the following steps:

[0075] like Figure 2 As shown, the system is first powered on and initialized. Then, the comprehensive difficulty coefficient K1 of the operating condition is calculated based on the collected sensor data. Next, the start-up preset is performed and the critical point is determined to determine whether to enter the adaptive slip control stage. The specific steps are as follows:

[0076] S1. Read and verify the online status of multi-source sensors, fast engagement valve, slow engagement valve, fast disengagement valve and slow disengagement valve, and establish a unified time reference, while loading the factory or saved control parameters; the control parameters include clutch estimation model parameters, pre-engagement point database, pre-engagement point estimation bias, K fusion weight, thermal model parameters and safety override parameters.

[0077] S2. Collect real-time sensing data from multiple source sensors and synchronize the real-time sensing data in time.

[0078] In some specific embodiments, the process of synchronizing the real-time sensing data includes:

[0079] Define a unique reference clock to time-stamp the real-time sensing data from multiple sensors;

[0080] The real-time sensor data from different sampling periods is resampled to the main control period using an interpolation / hold algorithm.

[0081] For real-time sensor data with fixed communication delays, a model is created in the end-to-end delay model to compensate for time in subsequent decision-making processes, including calculating the comprehensive difficulty coefficient K1 of the operating condition, querying and determining the pre-engagement point PB, calculating the preset advance Δt, determining whether the clutch has reached the torque transmission critical point, three-layer closed-loop control, and determining the synchronization threshold.

[0082] S3. Preprocess the real-time sensing data after time synchronization and calculate the slope angle θ, vehicle mass m, adhesion coefficient μ, and driver demand intensity P. drive Then, by utilizing the slope angle θ, vehicle mass m, adhesion coefficient μ, and driver's required intensity P... drive The overall difficulty coefficient K1 of the working condition is calculated by fusing the steering wheel angle δ in the real-time sensor data with the weight of K.

[0083] In some specific embodiments, the preprocessing of the time-synchronized real-time sensing data includes filtering and slope limiting, physical feasible interval comparison, bad value detection and processing, and redundancy consistency comparison.

[0084] In some specific embodiments, the process of calculating the overall difficulty coefficient K1 of the working condition includes:

[0085] The slope angle θ, vehicle mass m, adhesion coefficient μ, and driver's required intensity P are considered. drive The steering wheel angle δ from the real-time sensor data is mapped to the interval [0, 1] and the slope factor x is output. θ Loading factor x m Adhesion factor x μ Driver demand factor x p and turning factor x δ ;

[0086] Based on the preprocessed slope angle θ, vehicle mass m, adhesion coefficient μ, steering wheel angle δ, and driver demand intensity P drive The corresponding confidence indices are processed through a fusion algorithm to obtain the slope angle θ, vehicle mass m, adhesion coefficient μ, steering wheel angle δ, and driver demand intensity P. drive The corresponding overall confidence level C K ;

[0087] Using the overall confidence level C K As a weighting factor, the slope factor x is dynamically adjusted. θ Loading factor x m Adhesion factor x μ , turning factor x δ and driver demand factor x p The original difficulty coefficient K0 is calculated after the K-fusion weights are used;

[0088] The overall difficulty coefficient K of the operating conditions in the previous control cycle is based on the control parameters loaded in step S1, either from the factory or stored data. prev The original difficulty coefficient K0 is used to obtain the final comprehensive difficulty coefficient K1 of the working condition after smoothing update, change rate limitation and shrinkage correction.

[0089] S4. Based on the comprehensive difficulty coefficient K1, slope angle θ, vehicle mass m, adhesion coefficient μ, and driver's required intensity P. drive Perform a start-up preset to make the actual clutch pressure P act The engine output torque and the engine reach a preset coordinated state at the start.

[0090] S4.1, based on the comprehensive difficulty coefficient K1, slope angle θ, vehicle mass m, adhesion coefficient μ, and driver's required intensity P. driveThe pre-joint point PB is determined by querying the pre-joint point database, the pre-set pressure and pre-set torque are calculated, and the pre-set lead time Δt is calculated based on the full-link delay model.

[0091] S4.2. Based on the preset pressure, preset torque, and preset lead time Δt, before the expected start-up moment, a preset pressure command is sent to the fast-engagement valve and slow-engagement valve of the clutch actuator, and an engine preset torque command is sent to the engine controller for start-up control, in order to ensure that the actual clutch pressure P... act The engine output torque and the engine reach a preset coordinated state at the start.

[0092] In some specific embodiments, if the driver cancels the start signal, pedal brake signal, gear shift signal, or transmission start request cancellation signal is detected during the start preset process, the engine preset torque command is canceled, and the fast release valve or slow release valve is controlled to release the working chamber pressure on the engagement side of the clutch actuator.

[0093] S5. Based on real-time sensor data and actual clutch pressure P act The speed difference between the clutch pressure plate and the driven plate and the estimated clutch torque are calculated to determine whether the clutch has reached the critical point of torque transmission. If the critical point of torque transmission is reached, the adaptive slip control stage is entered.

[0094] S5.1 Calculate the speed difference between the two sides of the clutch pressure plate and the driven plate using the engine speed and gearbox input shaft speed in the real-time sensing data;

[0095] S5.2, Based on the actual clutch pressure P act The estimated clutch torque is calculated using the engine speed, transmission input shaft speed and clutch temperature from the real-time sensor data.

[0096] S5.3 Determine whether the clutch has reached the critical point of torque transmission: When the rate of decrease of the engine speed within a preset time period exceeds a preset decrease rate threshold, the rate of change of the speed difference within a preset time period is less than a preset change rate threshold, and the estimated torque of the clutch exceeds the critical torque threshold, it is determined that the clutch has reached the critical point of torque transmission, and then the adaptive slip control stage is entered.

[0097] like Figure 3 As shown, a three-layer closed-loop control is executed in parallel. Based on whether the speed difference between the engine speed and the transmission input shaft speed is less than a preset synchronization threshold, the clutch is locked, and a start-up quality assessment is performed after the lock-up is completed. The specific steps are as follows:

[0098] S6. In the adaptive slip control phase, a three-layer closed-loop control is executed based on acceleration, torque, inner layer pressure, and slip. During the three-layer closed-loop control process, safe override based on a risk index and graded thermal protection based on a thermal model are executed in parallel. This continues until the speed difference between the engine speed and the transmission input shaft speed in the real-time sensor data is less than a preset synchronization threshold and the vehicle longitudinal acceleration change rate is less than a preset impact threshold. At this point, the control valve quickly engages the clutch to reduce the actual clutch pressure P. act Increase the pressure to full engagement to control clutch lock-up.

[0099] In some specific embodiments, the process of executing the three-layer closed-loop control includes:

[0100] ①Outer layer acceleration closed-loop control:

[0101] Based on the driver's demand intensity and the overall difficulty coefficient K1 of the operating conditions, a desired longitudinal acceleration curve is set. By comparing the acceleration actually measured by the sensor with the corresponding expected value on the desired longitudinal acceleration curve, the outer layer correction torque ∆T is calculated. outer ;

[0102] ②Mid-level torque closed-loop control:

[0103] The anti-rollback safety torque is calculated using wheel radius, slope angle, and vehicle mass.

[0104] Based on the intensity of driver demand, the required torque for the driver is obtained by looking up a table in combination with the engine torque characteristics.

[0105] The upper limit of the adhesion limit torque is calculated based on the wheel radius, vehicle mass, adhesion coefficient, slope angle, and longitudinal acceleration in the real-time sensor data.

[0106] Based on the outer layer modified torque ∆T outer The target clutch torque T is calculated based on the following parameters: anti-rollback safety torque, driver-required torque, upper limit of adhesion torque, and estimated clutch torque, according to the fusion rules. m The torque closed-loop correction is generated by a PI / PID control law with gain scheduling, so that the estimated clutch torque follows the target clutch torque T. m This is to ensure that the torque transmitted by the clutch is precisely controllable;

[0107] ③ Inner layer pressure and slip closed-loop control:

[0108] Based on the pressure-torque characteristics of the clutch actuator, look up the table to determine the target clutch torque T. m Converted into feedforward pressure;

[0109] Based on the target clutch torque T m The error between the estimated clutch torque and the actual clutch torque generates a pressure correction amount;

[0110] The actual slip is calculated based on the engine speed and transmission input shaft speed in the real-time sensor data, and a slip correction amount is generated based on the error between the target slip and the actual slip in the driver's demand intensity.

[0111] The final target pressure is obtained by summing the feedforward pressure, pressure correction, and slip correction.

[0112] Calculate the actual clutch pressure P act The pressure error is compared with the final target pressure, and the equivalent target gas flow rate is calculated based on the pressure error and the rate of change of the target pressure.

[0113] The equivalent target airflow rate is decomposed into low-frequency and high-frequency target airflow rates using a frequency band decomposition algorithm. These are then calculated and converted into opening or duty cycle signals for the fast-action and slow-action valves, respectively, to control the actual clutch pressure P. act Follow the pressure of the ultimate goal.

[0114] In some specific embodiments, the execution process of the security override is as follows:

[0115] Based on the engine speed in the real-time sensor data, the engine stall risk index is calculated according to the degree and duration of the engine speed approaching the stall threshold.

[0116] Based on the clutch-estimated torque, and the slope angle, vehicle longitudinal velocity direction, and vehicle longitudinal acceleration direction in the real-time sensor data, the backward slip risk index is calculated.

[0117] The impact risk index is calculated based on the longitudinal acceleration in the real-time sensing data.

[0118] Determine if the stall risk index exceeds the high stall risk threshold. If the stall risk index exceeds the high stall risk threshold, reduce the actual clutch pressure P by controlling the quick release valve or slow release valve to release the working chamber pressure on the engagement side of the clutch actuator. act The clutch transmits torque to perform a forced anti-stalling action; if the stall risk index does not exceed the high stall risk threshold, or after the stall risk index is controlled within the preset safe zone by performing a forced anti-stalling action, it is determined whether the rollback risk index exceeds the rollback threshold.

[0119] If the slippage risk index exceeds the slippage threshold, then the slippage suppression action is executed, and the target torque lower limit is adjusted to be above the anti-slippage torque value. If the slippage risk index does not exceed the slippage threshold, then it is determined whether the impact risk index exceeds the impact threshold.

[0120] If the shock risk index exceeds the shock threshold, then anti-shock actions are executed; if the shock risk index does not exceed the shock threshold, then the current three-layer closed-loop control output is maintained.

[0121] In some specific embodiments, during the process of controlling clutch lock-up, the real-time impact risk index is continuously monitored. If the impact risk index rises above the impact threshold for a short period of time, the actual clutch pressure P is reduced by decreasing the opening degree or duty cycle of the fast engagement valve and the slow engagement valve. act The rate of increase in the final target pressure can be slowed down by limiting the rate of change in engine torque; if necessary, the fast release valve and slow release valve can also be controlled to briefly participate in releasing the working chamber pressure on the engagement side of the clutch actuator to reduce the actual clutch pressure P. act The rise.

[0122] In some specific embodiments, after the clutch lock-up is completed, the adaptive slip control phase is switched to the steady-state pressure maintenance and monitoring phase to maintain sufficient connection force and prevent unnecessary continuous slip.

[0123] In some specific embodiments, the graded thermal protection based on the thermal model adopts a first-order thermal accumulation model, and performs corresponding graded thermal protection by setting multiple temperature thresholds and their corresponding thermal protection strategies.

[0124] S7. After the clutch locks up, i.e. the start is completed, the quality of the start process is scored. Based on the quality score, it is decided whether to update the control parameters and save them. The updated control parameters are used for loading during the next start.

[0125] In some specific embodiments, the process of scoring the quality of this start-up process and determining whether to update and save the control parameters based on the quality score results is as follows:

[0126] Record the number of safe overruns and their duration during this start-up process, and calculate the start-up quality score based on the rollback distance, total start-up time, minimum engine speed, maximum impact, and thermal load during this start-up process.

[0127] Determine whether the number and duration of safety overruns and the start-up quality score meet the update conditions. The update conditions are that safety overruns are not continuously triggered and the start-up quality score is higher than the preset quality score threshold.

[0128] If the update conditions are not met, the update process will end.

[0129] If the update conditions are met, an update permission instruction is output. The slope angle, vehicle mass, adhesion coefficient, driver demand intensity and steering wheel angle, pre-engagement point PB and start quality score of this start are used as candidate samples. Anomaly detection and distance measurement are performed on the candidate samples.

[0130] If the detection result of the candidate sample is a normal sample, the candidate sample is written into the pre-binding point database, and its sample weight is adjusted according to the density and importance of the candidate sample. The database is managed for lifecycle: sliding time window to ensure that the number and coverage of candidate samples are within a preset reasonable range.

[0131] If the start-up quality score does not decrease significantly and the safe override frequency does not increase within the preset time window, then a preset small-amplitude, bounded update is performed on some parameters of the control parameters, including the K fusion weight, the pre-engagement point estimation bias, and the torque estimator, and the update is saved; the significant decrease value is set empirically.

[0132] If the start-up quality score drops significantly or the safety override frequency increases within the preset time window, the system enters a learning freeze state, stops online learning updates, cancels all control parameter updates, and restores the system to the factory-calibrated control parameters or the last saved control parameters. The system will then prompt maintenance / replacement suggestions through the human-machine interface (HMI).

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present invention.

Claims

1. A starting control method for an AMT pneumatic clutch actuator, wherein the clutch actuator includes a fast engagement valve, a slow engagement valve, a fast disengagement valve, and a slow disengagement valve; characterized in that, Includes the following steps: S1. Read and verify the online status of multi-source sensors, fast engagement valve, slow engagement valve, fast disengagement valve and slow disengagement valve, and establish a unified time reference, while loading the factory or saved control parameters; the control parameters include clutch estimation model parameters, pre-engagement point database, pre-engagement point estimation bias, K fusion weight, thermal model parameters and safety override parameters; S2. Collect real-time sensing data from multiple source sensors and synchronize the real-time sensing data in time. S3. The real-time sensing data after time synchronization is preprocessed, and the slope angle, vehicle mass, adhesion coefficient and driver demand intensity are calculated. Then, the comprehensive difficulty coefficient K1 of the working condition is calculated by using the slope angle, vehicle mass, adhesion coefficient, driver demand intensity, and steering wheel angle in the real-time sensing data and the K fusion weight. S4. Based on the comprehensive difficulty coefficient K1, slope angle, vehicle weight, adhesion coefficient, and driver's required intensity, control the fast engagement valve and slow engagement valve to preset the start-up, so that the actual clutch pressure P act The engine output torque reaches a preset coordinated state at the moment of start-up; S5. Based on real-time sensor data and actual clutch pressure P act The speed difference between the two sides of the clutch pressure plate and driven plate and the estimated torque of the clutch are calculated to determine whether the clutch has reached the critical point of torque transmission. If the critical point of torque transmission is reached, the adaptive slip control stage is entered. S6. In the adaptive slip control phase, a three-layer closed-loop control is executed based on acceleration, torque, inner layer pressure, and slip. During this three-layer closed-loop control, safety override based on a risk index and graded thermal protection based on a thermal model are executed in parallel. This continues until the speed difference between the engine speed and the transmission input shaft speed in the real-time sensor data is less than a preset synchronization threshold and the vehicle's longitudinal acceleration change rate is less than a preset impact threshold. At this point, the fast engagement valve, slow engagement valve, fast disengagement valve, and slow disengagement valve are activated to adjust the actual clutch pressure P. act Increase the pressure to full engagement and control clutch lock-up; S7. After the clutch locks up, i.e. the start-up is completed, the quality of the start-up process is scored. Based on the quality score, it is decided whether to update the control parameters and save them, thus completing the start-up control of the AMT pneumatic clutch actuator.

2. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, The process of synchronizing the real-time sensing data in step S2 includes: Define a unique reference clock to time-stamp the real-time sensing data from multiple sensors; The real-time sensor data from different sampling periods is resampled to the main control period using an interpolation / hold algorithm. For real-time sensor data with fixed communication delays, a model is created in the end-to-end delay model for time compensation in steps S3-S6.

3. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, The preprocessing of the time-synchronized real-time sensing data in step S3 includes: Filtering and slope limiting, physical feasible interval comparison, bad value detection and handling, redundancy consistency comparison.

4. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, The process of calculating the overall difficulty coefficient K1 in step S3 includes: The slope angle, vehicle mass, adhesion coefficient, driver demand intensity, and steering wheel angle from real-time sensor data are mapped to the [0,1] interval, and the slope factor, load factor, adhesion factor, driver demand factor, and steering factor are output. Based on the confidence indices corresponding to the preprocessed slope angle, vehicle mass, adhesion coefficient, steering wheel angle, and driver demand intensity, a fusion algorithm is used to process them to obtain the comprehensive confidence indices corresponding to the slope angle, vehicle mass, adhesion coefficient, steering wheel angle, and driver demand intensity. The original difficulty coefficient K0 is calculated by dynamically adjusting the K fusion weights corresponding to the slope factor, load factor, adhesion factor, steering factor and driver demand factor using the comprehensive confidence level as a weighting factor. The overall difficulty coefficient K of the operating conditions in the previous control cycle is based on the control parameters loaded in step S1, either from the factory or stored data. prev The original difficulty coefficient K0 is used to obtain the final comprehensive difficulty coefficient K1 of the working condition after smoothing update, change rate limitation and shrinkage correction.

5. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, Step S4 is as follows: S4.

1. Based on the comprehensive difficulty coefficient K1 of the working condition, the slope angle, the vehicle mass, the adhesion coefficient and the driver's demand intensity, the pre-joint point PB is determined from the pre-joint point database. The pre-set pressure and pre-set torque are calculated, and the pre-set advance amount Δt is calculated based on the full-link delay model. S4.

2. Based on the preset pressure, preset torque, and preset lead time Δt, before the expected start-up moment, a preset pressure command is sent to the fast-engagement valve and slow-engagement valve of the clutch actuator, and an engine preset torque command is sent to the engine controller for start-up control, in order to ensure that the actual clutch pressure P... act The engine output torque and the engine reach a preset coordinated state at the start.

6. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, Step S5 is as follows: The speed difference between the two sides of the clutch pressure plate and driven plate is calculated using the engine speed and gearbox input shaft speed from the real-time sensor data. Based on the actual clutch pressure P act The estimated clutch torque is calculated using the engine speed, transmission input shaft speed and clutch temperature from the real-time sensor data. Determining whether the clutch has reached the critical point of torque transmission: When the trend of engine speed change and the trend of speed difference change meet the preset conditions and the estimated torque of the clutch exceeds the critical torque threshold, it is determined that the clutch has reached the critical point of torque transmission, and then the adaptive slip control stage is entered.

7. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, In step S6, the three-layer closed-loop control process of acceleration, torque, inner layer pressure, and slip is specifically performed as follows: ①Outer layer acceleration closed-loop control Based on the driver's demand intensity and the overall difficulty coefficient K1 of the operating conditions, a desired longitudinal acceleration curve is set. By comparing the acceleration actually measured by the sensor with the corresponding expected value on the desired longitudinal acceleration curve, the outer layer correction torque ∆T is calculated. outer ; ②Mid-level torque closed-loop control The anti-rollback safety torque is calculated using wheel radius, slope angle, and vehicle mass. Based on the intensity of driver demand, the required torque for the driver is obtained by looking up a table in combination with the engine torque characteristics. The upper limit of the adhesion limit torque is calculated based on the wheel radius, vehicle mass, adhesion coefficient, slope angle, and longitudinal acceleration in the real-time sensor data. Based on the outer layer modified torque ∆T outer The target clutch torque T is calculated based on the following parameters: anti-rollback safety torque, driver-required torque, upper limit of adhesion torque, and estimated clutch torque, according to the fusion rules. m ; A PI / PID control law with gain scheduling is used to generate a torque closed-loop correction, enabling the estimated clutch torque to follow the target clutch torque T. m This is to ensure that the torque transmitted by the clutch is precisely controllable; ③ Inner layer pressure and slip closed-loop control Based on the pressure-torque characteristics of the clutch actuator, look up the table to determine the target clutch torque T. m Converted into feedforward pressure; Based on the target clutch torque T m The error between the estimated clutch torque and the actual clutch torque generates a pressure correction amount; The actual slip is calculated based on the engine speed and transmission input shaft speed in the real-time sensor data, and a slip correction amount is generated based on the error between the target slip and the actual slip in the driver's demand intensity. The final target pressure is obtained by summing the feedforward pressure, pressure correction, and slip correction. Calculate the actual clutch pressure P act The pressure error is compared with the final target pressure, and the equivalent target gas flow rate is calculated based on the pressure error and the rate of change of the target pressure. The equivalent target airflow rate is decomposed into low-frequency and high-frequency target airflow rates using a frequency band decomposition algorithm. These are then calculated and converted into opening or duty cycle signals for the fast-action and slow-action valves, respectively, to control the actual clutch pressure P. act Follow the pressure of the ultimate goal.

8. The starting control method for an AMT pneumatic clutch actuator according to claim 7, characterized in that, The execution process of the safety override described in step S6 is as follows: Based on the engine speed in the real-time sensor data, the engine stall risk index is calculated according to the degree and duration of the engine speed approaching the stall threshold. Based on the clutch-estimated torque, and the slope angle, vehicle longitudinal velocity direction, and vehicle longitudinal acceleration direction in the real-time sensor data, the backward slip risk index is calculated. The impact risk index is calculated based on the longitudinal acceleration in the real-time sensing data. Determine whether the flameout risk index exceeds the high flameout risk threshold. If the flameout risk index exceeds the high flameout risk threshold, then execute a forced flameout prevention action by controlling the fast separation valve or the slow separation valve. If the stall risk index does not exceed the high stall risk threshold, or after the stall risk index is controlled within the preset safe zone by performing a forced anti-stalling action, determine whether the rollback risk index exceeds the rollback threshold. If the slippage risk index exceeds the slippage threshold, then the slippage suppression action is executed, and the target torque lower limit is adjusted to be above the anti-slippage torque value. If the slippage risk index does not exceed the slippage threshold, then it is determined whether the impact risk index exceeds the impact threshold. If the shock risk index exceeds the shock threshold, then anti-shock actions are executed; if the shock risk index does not exceed the shock threshold, then the current three-layer closed-loop control output is maintained.

9. The starting control method for an AMT pneumatic clutch actuator according to claim 8, characterized in that: In step S6, during the clutch lock-up process, the real-time impact risk index is continuously monitored. If the impact risk index rises above the impact threshold for a short period, the actual clutch pressure P is reduced by decreasing the opening degree or duty cycle of the fast engagement valve and the slow engagement valve. act The rise is to slow the rate of increase in the final target pressure; After the clutch lock-up is completed, the system switches from the adaptive slip control phase to the steady-state pressure maintenance and monitoring phase to maintain sufficient connection force and prevent unnecessary continuous slip.

10. The starting control method for an AMT pneumatic clutch actuator according to claim 1, characterized in that, In step S7, the quality of this start-up process is scored, and the process of deciding whether to update the control parameters and save them based on the quality score results is as follows: Record the number of safe overruns and their duration during this start-up process, and calculate the start-up quality score based on the rollback distance, total start-up time, minimum engine speed, maximum impact, and thermal load during this start-up process. Determine whether the number and duration of safety overruns and the start-up quality score meet the update conditions. The update conditions are that safety overruns are not continuously triggered and the start-up quality score is higher than the preset quality score threshold. If the update conditions are not met, the update process will end. If the update conditions are met, an update permission instruction is output. The slope angle, vehicle mass, adhesion coefficient, driver demand intensity and steering wheel angle, pre-engagement point PB and start quality score of this start are used as candidate samples. Anomaly detection and distance measurement are performed on the candidate samples. If the detection result of the candidate sample is a normal sample, the candidate sample is written into the pre-binding point database, and its sample weight is adjusted according to the density and importance of the candidate sample. The database is managed for lifecycle: sliding time window to ensure that the number and coverage of candidate samples are within a preset reasonable range. If the starting quality score does not decrease significantly and the safe override frequency does not increase within the preset time window, then the K fusion weight, pre-engagement point estimation bias, and some parameters of the torque estimator in the control parameters will be updated in a preset small-amplitude bounded manner and saved. The significant decrease value is set based on experience; If the start-up quality score drops significantly or the safety override frequency increases within the preset time window, the system enters a learning freeze state, stops online learning updates, cancels all control parameter updates, and restores the system to the factory-calibrated control parameters or the last saved control parameters. The system will then prompt maintenance / replacement suggestions through the human-machine interface (HMI).

Images