基于数据驱动的磁浮列车悬浮系统参数整定

Data-Driven Parameter Tuning for Maglev Train Levitation System

为了解决由于磁浮列车悬浮系统模型存在复杂非线性所带来的控制器参数整定难题,考虑仅依靠悬浮系统单次悬浮调试的输入输出数据来实现控制器参数整定,研究了基于数据驱动的磁浮列车悬浮系统快速参数整定方法.首先,通过磁浮列车悬浮系统的建模分析了悬浮系统的开环不稳定性和复杂非线性;其次,针对虚拟参考反馈参数整定方法存在的参考模型确定问题,利用估计闭环响应方法来实现数据驱动的控制器参数整定;并且考虑到数据中存在的干扰噪声会影响控制器参数整定,提出了基于信号投影的磁悬浮系统数据噪声抑制方法;最后,以单铁磁悬浮系统为研究对象,通过单铁悬浮实验验证了本文提出的基于数据驱动的磁浮列车悬浮系统参数整定方法的有效性.研究结果表明:悬浮系统的开环不稳定性和复杂非线性会给参数的快速整定带来较大困难;基于信号投影的噪声抑制方法能够将带噪声数据的方差降低54.1%;基于数据驱动的参数整定方法能够快速地整定好悬浮系统控制器参数,经参数整定后的系统相较于初始条件下只粗调好的PID反馈控制系统的阶跃响应超调量下降72.0%,平方误差积分(ISE)下降79.8%,绝对误差积分(IAE)下降54.5%.

In order to solve the controller parameter tuning problem caused by the complex nonlinearity of the maglev train levitation system model, a data-driven fast parameter tuning method for the maglev train levitation system is proposed, which is based on only the input and output data of a single levitation tuning of the levitation system. First, the open-loop instability and complex nonlinearity of the levitation system are analyzed by modeling of the maglev train levitation system. Aiming at the problem of determining the reference model in the virtual reference feedback tuning method, estimation of the closed-loop response is then used to realize the datadriven controller parameter tuning. Considering that the interference noise in the data will affect the tuning of controller parameters, a data noise suppression method of maglev system based on signal projection is proposed.Finally, taking a single-rail levitation system for example, the effectiveness of this data-driven parameter tuning method for maglev train levitation system was verified through the single-rail levitation experiment. The results show that the open-loop instability and complex nonlinearity of the suspension system will bring great difficulties to the rapid adjustment of parameters;the noise suppression method based on signal projection can reduce the variance of noise data by 54.1%;and the parameter tuning method based on data driven technique can quickly set the controller parameters of the suspension system. Compared with the PID feedback control system with only coarse tuning under the initial conditions, the step response overshoot of the system after parameter tuning is reduced by 72.0%, the square error integral(ISE) is reduced by 79.8%, and the absolute error integral(IAE) is reduced by 54.5%.