BP神经网络控制在复合传动试验台中的应用

Application of BP Neural Network Control in Compound Coupled Hydro-mechanical Transmission Test Bench

为了实现对混合式液压-机械无极传动(CCHMT)的调速性能优化,需在多种工况下采集数据,而搭建的试验台在实际应用过程中取值准确率低、精度不高,且存在控制值和电磁阀的开度呈弱相关性的问题。为此,对试验台控制系统应用BP神经网络控制方法,以输出转速比平稳为控制目标。该方法引入多传感器作为控制系统反馈环节,从而将开环系统转变成闭环系统,并且实现控制系统从点动控制转变为连续控制。通过AMESim和Matlab/Simulink的仿真结果与实际实验数据的对比分析表明,引入该控制策略后输出转速明显平稳,并且控制值和电磁阀的开度呈强相关性;在不同输入转速下,实际开度都与理论计算拟合良好,为该系统下一步优化研究提供基础。

In order to optimize the speed regulation performance of compound coupled hydro-mechanical transmission(CCHMT),it is necessary to collect data under various working conditions,the test bench built has poor accuracy in the actual application process,and the control value and the opening of the solenoid valve are weakly correlated.Hence,the BP neural network control method is applied in the test bench control system,with the stable output speed ratio as the control goal,the method introduces multiple sensors as the feedback link of the control system,so as to transform the open-loop system into a closed-loop system,and realize the control system from jog control to continuous control.By comparing the simulation results of AMESim and Matlab/Simulink with the actual experimental data,it is show that the output speed is obviously stable after the introduction of the control strategy,the control value is strongly correlated with the opening degree of the solenoid valve,and the actual opening is well fitted with the theorelical calculation at different input speeds,which provides a basis for the next optimization study of the system.