Self-Learning Method for Shift Hub Position in a Double-Input Shaft Hybrid Gearbox Using Linear Active Disturbance Rejection Control and Nutcracker Optimization Algorithm

<div>Due to manufacturing, assembly, and actuator wear, slight deviations between the actual and logical positions of various gears in a transmission system may accumulate, affecting shift quality, reducing shift accuracy, and causing operational anomalies. To address this issue, a self-learning method based on the top dead center (TDC) and lower dead center (LDC) was proposed, specifically for the hybrid gearbox of an electric torque converter (eTC) module and a double-input shaft gearbox (DIG). The linear active disturbance rejection control (LADRC) method was employed to estimate and manage the nonlinear resistance during the motion of the shifting motor. To simplify the controller parameter problem, the nutcracker optimization algorithm (NOA) was utilized to tune the LADRC parameters, thereby optimizing the position self-learning process. The control strategy was modeled using MATLAB/SIMULINK, and its reasonableness was verified through hardware-in-the-loop (HIL) tests. Based on these tests, the approach was applied to three controllers: the PID controller, LADRC, and NOA_LADRC. Subsequent gearbox bench experiments showed that the self-learning method successfully corrected gear positions during product launch and shifting. Among these controllers, NOA_LADRC effectively addresses nonlinear disturbances, reducing the time required for identifying the shift drum position by 0.06 s and 0.36 s, respectively. It provides critical parameters for the control of the shift actuator, thereby optimizing shift performance and indirectly enhancing overall performance.</div>