Song Qi, Song Yong-Duan
State Key Laboratory of Rail Traffic Control and Safety, Beijng Jiaotong University, Beijing 100044, China.
IEEE Trans Neural Netw. 2011 Dec;22(12):2250-61. doi: 10.1109/TNN.2011.2175451.
This paper investigates the position and velocity tracking control problem of high-speed trains with multiple vehicles connected through couplers. A dynamic model reflecting nonlinear and elastic impacts between adjacent vehicles as well as traction/braking nonlinearities and actuation faults is derived. Neuroadaptive fault-tolerant control algorithms are developed to account for various factors such as input nonlinearities, actuator failures, and uncertain impacts of in-train forces in the system simultaneously. The resultant control scheme is essentially independent of system model and is primarily data-driven because with the appropriate input-output data, the proposed control algorithms are capable of automatically generating the intermediate control parameters, neuro-weights, and the compensation signals, literally producing the traction/braking force based upon input and response data only--the whole process does not require precise information on system model or system parameter, nor human intervention. The effectiveness of the proposed approach is also confirmed through numerical simulations.
本文研究了通过联轴节连接的多节高速列车的位置和速度跟踪控制问题。推导了一个反映相邻车辆间非线性和弹性冲击以及牵引/制动非线性和驱动故障的动力学模型。开发了神经自适应容错控制算法，以同时考虑系统中的各种因素，如输入非线性、执行器故障和列车内力的不确定冲击。由此产生的控制方案基本上独立于系统模型，并且主要是数据驱动的，因为利用适当的输入-输出数据，所提出的控制算法能够自动生成中间控制参数、神经权重和补偿信号，仅根据输入和响应数据就可直接产生牵引/制动力——整个过程不需要系统模型或系统参数的精确信息，也不需要人工干预。通过数值模拟也证实了所提方法的有效性。
