Abstract
To address the problem of reduced comfort caused by vehicle tilt for hilly tractor driver, a novel seat posture omnidirectional levelling system is designed, and an omnidirectional rapid levelling control strategy (QBP-PID) is proposed, which fuses Q-learning, Back Propagation (BP) neural network, and Proportional-Integral-Derivative (PID) control. Firstly, an omnidirectional levelling system for seat posture is designed based on kinematic principles. On this basis, a model for the omnidirectional levelling system is established using valve-controlled hydraulic cylinder principles. Subsequently, addressing the challenge of difficult parameter tuning for the levelling system's PID control, a multi-level parameter update strategy employing QBP-PID is proposed for rapid omnidirectional levelling control. Simulation results show that under QBP-PID control, the 15° lateral levelling time is 2.98 s with an overshoot of 0.32°; The longitudinal levelling time at 20° is 3.41 s with an overshoot of 0.95°. Compared to BP-PID and PID, the lateral levelling time is reduced by 18.13% and 27.66% respectively, while the longitudinal levelling time decreased by 17.63% and 31.6% respectively. The superiority of the QBP-PID omnidirectional rapid levelling control strategy has been demonstrated.
Go to article