基于回声状态网络速度预测的高速动车组优化控制

Optimal control of high-speed EMU based on speed prediction with echo state networks

精确的速度预测控制是实现高速动车组安全、准点、节能优化控制的基础。然而,高速动车组运行过程干扰因素多、动力学非线性复杂,难以通过数学分析的方式建立其精确的速度预测模型。文章充分利用回声状态网络在非线性时间序列预测方面的优势,进行高速动车组运行速度的在线预测,同时采用多目标粒子群优化算法对预测控制输入进行多目标优化,实现高速动车组运行速度的精确控制,并基于某型号高速动车组的现场运行数据,仿真验证了本文方法的有效性。

The precise speed prediction is the basis of the optimal control of high-speed EMU(HEMU) with respect to safe, punctual and energy-efficient. However, it is difficult to establish an accurate speed prediction model for the HEMU through mathematical method, because of the complex nonlinear dynamics and various disturbs along HEMU’s running process. In this paper, an online speed prediction approach for the HEMU is proposed by taking advantage of ESNs’ excellent prediction ability. Besides, multi-objective particle swarm optimization algorithm is employed to conduct the prediction control, so as to fulfill the precisely speed tracking control of HEMU. Finally, based on the field operation data of some type of HEMU, simulation experiments are carried out to validate the effectiveness of the proposed method.