CN-121993587-A - Automatic transmission steady-state self-learning method

CN121993587ACN 121993587 ACN121993587 ACN 121993587ACN-121993587-A

Abstract

A steady-state self-learning method of an automatic transmission belongs to the technical field of vehicle automatic transmission control. The method comprises the steps of judging self-learning enabling conditions, identifying and capturing steady-state working conditions, applying controlled test excitation and data acquisition, calculating and updating parameters, verifying and storing learning results, and managing and exiting learning states. According to the invention, the steady-state driving working condition of the vehicle is intelligently captured, controlled test excitation is applied on the premise of meeting preset enabling conditions, steady-state control parameters are calculated and updated on line through a built-in model based on excitation response data, and the learning process safety is ensured through a multi-level enabling judgment and real-time interrupt mechanism, so that the problems of high cost, low efficiency and easiness in shifting quality caused by parameter drift reduction due to lack of a reliable online self-learning scheme of steady-state core parameters of an automatic transmission in the prior art are solved, the self-adaptive compensation of performance attenuation in the whole life cycle of the vehicle is realized, and the precision and reliability of parameter learning are improved.

Inventors

WANG ZIQUAN
ZHAO HAISHENG
LI WENQIANG
CUI TIANLE
MA CHENYANG
SUN BENCHAO
SHENG LIANG
WANG JIANWU
LUO XIUCHUAN
WANG ZHANWEN
MA QUN
ZHANG HONGBIN
DING LIN
XIA HONGBIN
YANG SHIZHEN
LIU JINGBO

Assignees

哈尔滨东安汽车发动机制造有限公司
哈尔滨东安汽车动力股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260212

Claims (6)

1. The steady-state self-learning method for the automatic transmission is characterized by comprising the following steps of: S1, judging self-learning enabling conditions, namely continuously monitoring a group of global enabling conditions by a system, and entering a steady self-learning preparation state only when all the conditions are met at the same time; S2, steady-state working condition identification and capture, namely when all the global enabling conditions in S1 are met, the system starts to continuously identify steady-state working conditions meeting preset conditions, marks the steady-state working conditions as learning window periods, and starts a learning preparation flow; S3, applying controlled test excitation and data acquisition, namely, in a learning window period, the TCU actively applies test excitation aiming at different learning targets and synchronously acquires key response data before and after excitation application; S4, parameter calculation and updating, namely substituting the response data acquired in the S3 into a physical model or a statistical model preset in the TCU to calculate to obtain a corresponding new value of the steady-state control parameter, comparing the calculated new value of the parameter with an original calibration value stored in the TCU, and generating a smoothly updated parameter value through filtering; s5, verifying and storing a learning result, namely enabling the system to enter a short-time verification stage after single parameter updating, continuously monitoring a gear shifting quality index by the TCU in a subsequent gear shifting operation related to the updated parameter, and if the gear shifting quality is within an expected range, confirming that learning is effective, and writing a verified effective parameter value into the TCU; And S6, learning state management and exiting, namely adopting a state machine management mechanism in the whole self-learning process, immediately interrupting the self-learning process if the system monitors that the safety condition or the steady-state condition is not met in any stage of S2-S5, resetting all temporarily acquired data, recovering a normal driving mode, continuously monitoring working conditions by the system, and waiting for restarting the learning process in the next learning window period.
2. The method for steady-state self-learning of an automatic transmission of claim 1 wherein S1 said global enabling condition comprises: the first condition is that the oil temperature of the transmission is in the optimal learning range; the second condition is that the EMS and the TCU do not report fault codes; Thirdly, the driving mileage or the running time of the vehicle reaches a preset self-learning trigger period; The voltage of the storage battery of the vehicle is in a normal working range; and fifthly, the vehicle is in a non-violent driving mode.
3. The method for steady-state self-learning of an automatic transmission according to claim 2, wherein S2 the steady-state conditions simultaneously satisfy the following conditions: The vehicle is in a stable vehicle speed cruising state, a slow acceleration state or a slow deceleration state in which an absolute value of acceleration is lower than a preset threshold value, an engine torque output stable state in which a torque fluctuation rate in a past preset period of time is lower than a set threshold value, a transmission is not in a gear shifting process and in a current gear fixing state, and duration of all the states exceeds the preset period of time.
4. The method for steady-state self-learning of an automatic transmission according to claim 3, wherein S3 is a specific excitation method for different learning objectives: if the learning target is clutch kiss point, on the premise of keeping the current gear transmission torque unchanged, the control pressure of the target clutch is increased in a stepped manner to obtain a pressure increment, and meanwhile, the working states of other clutches are kept unchanged; and if the learning target is the system pressure reference, the duty ratio of the main oil pressure electromagnetic valve is adjusted, and the following characteristic of the actual system pressure is observed.
5. The method of steady state self learning of an automatic transmission of claim 4 wherein S3 said data collection includes input shaft and output shaft speed variation, turbine speed variation, target clutch actual pressure, actual gear ratio and engine output torque.
6. The method for steady-state self-learning of an automatic transmission according to claim 5, wherein S4 the calculation of the physical model or the statistical model comprises: for clutch kiss point learning, after analyzing the applied pressure increment, observing whether sliding friction occurs on the rotation speed of an input and output shaft, and calculating an instant pressure point of the current clutch for starting to transmit torque, namely real-time kiss point; and for system pressure reference learning, calculating to obtain a system pressure reference correction value by analyzing the deviation and response speed of the actual system pressure and the target pressure after the duty ratio of the main oil pressure electromagnetic valve.

Description

Automatic transmission steady-state self-learning method Technical Field The invention relates to a steady-state self-learning method of an automatic transmission, and belongs to the technical field of control of automatic transmissions of vehicles. Background The shift quality of an automatic transmission (including key indexes such as shift shock degree, sliding friction control effect, clutch engagement speed and the like) is highly dependent on a set of control parameters such as clutch control pressure, torque phase parameter, inertia phase parameter, system pressure reference and the like which are determined in advance through bench test and road test calibration. However, in the actual use process of the vehicle, various factors may cause drift of the preset optimal control parameters, so as to reduce the shift quality: firstly, part wear and aging, such as clutch friction plate wear, seal aging, hydraulic oil path characteristic change and the like; Secondly, the consistency of products is different, and the automatic transmissions of the same model have individual performance differences due to manufacturing tolerances; Third, the usage environment and the working conditions are changed, such as fluctuation of the oil temperature of the transmission, difference of the driving altitude, different driving habits (load control modes) of the driver, and the like. The traditional solution mainly relies on fixed initial calibration data or performs rough compensation through a simple fault diagnosis mechanism, and cannot realize fine online self-adaptive adjustment. Although some of the prior art involves self-learning strategies, they are limited to transient learning (e.g., inertia phase learning) during shifting, and lack a set of reliable and safe online self-learning methods for steady-state operating mode core parameters (including clutch kiss point, steady-state torque transfer characteristics, system pressure references, etc.) that determine basic shift characteristics. At present, aiming at the steady-state parameter drift problem, a vehicle is usually required to return to a factory or is required to be recalibrated by a professional technician through special diagnosis equipment, so that the operation cost is high, the efficiency is low, and the use experience of a user is seriously influenced. Therefore, it is needed to develop a method capable of automatically, accurately and safely completing the self-learning of the steady-state core parameters during the normal running process of the vehicle. Disclosure of Invention In order to solve the problems in the background technology, the invention provides a steady-state self-learning method of an automatic transmission. The invention adopts the following technical scheme that the steady-state self-learning method of the automatic transmission comprises the following steps: S1, judging self-learning enabling conditions, namely continuously monitoring a group of global enabling conditions by a system, and entering a steady self-learning preparation state only when all the conditions are met at the same time; the global enabling conditions include: the first condition is that the oil temperature of the transmission is in the optimal learning range; the second condition is that the EMS and the TCU do not report fault codes; Thirdly, the driving mileage or the running time of the vehicle reaches a preset self-learning trigger period; The voltage of the storage battery of the vehicle is in a normal working range; and fifthly, the vehicle is in a non-violent driving mode. S2, steady-state working condition identification and capture, namely when all the global enabling conditions in S1 are met, the system starts to continuously identify steady-state working conditions meeting preset conditions, marks the steady-state working conditions as learning window periods, and starts a learning preparation flow; the steady-state working condition needs to meet the following conditions simultaneously: The vehicle is in a stable vehicle speed cruising state, a slow acceleration state or a slow deceleration state in which an absolute value of acceleration is lower than a preset threshold value, an engine torque output stable state in which a torque fluctuation rate in a past preset period of time is lower than a set threshold value, a transmission is not in a gear shifting process and in a current gear fixing state, and duration of all the states exceeds the preset period of time. S3, applying controlled test excitation and data acquisition, namely, in a learning window period, the TCU actively applies test excitation aiming at different learning targets and synchronously acquires key response data before and after excitation application; The specific excitation modes aiming at different learning targets are as follows: if the learning target is clutch kiss point, on the premise of keeping the current gear transmission torque unchanged, the control pressure of the