关键词: 电静液起竖, 换级碰撞, 神经网络, 模型补偿, 积分鲁棒控制

Abstract: To improve the accuracy and stability of the rapid erection control of special vehicles, aiming at the strong nonlinearity, model uncertainty and inter stage collision dynamics of the heavy-duty erection system driven by the two-stage hydraulic cylinder, a neural network-based RISE control method is thus proposed for the electro-hydrostatic erection system with inter stage collision compensation. Taking the five-hinge erection system directly driven by the electro-hydrostatic actuator as the research object, the state-space equations are established considering the erection mechanism kinematics, the inter change collision dynamics, and the pressure-flow dynamics of the pump-controlled system. A modified L-N contact model is utilized to feedforward compensate the nonlinear collision force. Meanwhile, a neural network is designed to approximate the unmodeled collision and friction dynamics. This integrated approach establishes a comprehensive feedforward compensation term for the second channel. Subsequently, dual-channel RISE (Robust Integral of the Sign of Error) control laws are developed to address both matched and unmatched uncertainties. The semi-global asymptotic tracking performance of the electro-hydrostatic five-hinge erection system is proved based on Lyapunov theory. The effectiveness of the proposed method is verified through the AMESim-Adams-Simulink co-simulation. The research results show that under the constant-power medium-speed erection condition and the rapid erection condition, the stability accuracies of NNRISE + CFC are 0.0073° and 0.0095° respectively. Compared with other control strategies, the thrust fluctuation during stage-changing is significantly suppressed, which has high control accuracy and motion stability.

Key words: electro-hydrostatic erection, inter stage collision, neural network, model compensation, robust integral control of the sign of error