Improved Generalized Predictive Adaptive Speed Tracking Control For High-Speed EMU

The research of operation control methods for high speed electric multiple unit(EMU) is a key to realize automatic train operation. Due to the complexity of its running environment, high-speed electric multiple unit operation process is a complex dynamic system with strong nonlinear, strong interference and more constraints. It has significance for design of adaptive and robust control schemes for high speed electric multiple unit to achieve its precision tracking control.Since the parameters of the high speed electric multiple unit process model are time- varying, this paper presents an improved generalized predictive adaptive control algorithm for high-speed electric multiple unit to achieve speed tracking. Specific studies are as follows:1. A high-speed EMU nonlinear model is established by using the echo sta te network of high-speed EMU operation process system identification, with the high-speed EMU operation data of CRH380 AL collected from the field. Particle swarm optimization algorithm is used to optimize key parameters of the echo state network, aiming to the difficult parameter selection problem. Such an identification performance is compared with ones using the dynamic recursion Elman neural network. Simulation results show that the echo state network can overcome the recursive neural network memories fade, it has an approximation capability of nonlinear dynamic systems and a stronger adaptive capacity, and the established nonlinear model can describe relatively complete high-speed EMU operation process.2. Aiming to the parameter time-varying problem and the parameter sudden change problem of the high speed EMU process dynamics model, a generalized predictive adaptive control algorithm(JGPC) is presented. Aimed at overcoming the shortcomings of the traditional GPC algorithm, such as a large amount of calculation, the complex algorithm and weak robustness, a modified generalized predictive control algorithm(JGPC) is presented which modifies the JGPC model by using FFRELS to estimate and update the model parameters. The generalized predictive control algor ithm can control of open-loop unstable systems, non- minimum phase systems and systems with large delay, by multi-step prediction, rolling optimization and feedback correction, to obtain desired system performance.3. The CRH380 AL high-speed EMU as the research object, JGPC adaptive control algorithm is presented in this paper to verify and test the control performance. The simulation results show that the algorithm can realize high speed EMU speed high precision tracking of improved generalized adaptive predictive control, strong robustness, and can meet the high speed EMU running safety and comfort requirements.