论文部分内容阅读
针对电液伺服系统非线性建模问题,研究了电液位置伺服系统神经网络辨识模型的基本结构。分析伺服系统动态模型的输入、输出关系,依据遗传算法优化神经网络权值和阈值,建立神经网络辨识模型的基本结构。利用xPC技术建立阀控缸实时电液伺服实验台,以实验台的阶跃输出信号作为改进BP神经网络系统辨识信号,以实验台正弦输出信号作为验证信号。结果表明:该神经网络辨识模型的基本结构可达到较高的辨识精度,其可信性得以验证,适用于非线性系统模型辨识。
Aiming at the problem of nonlinear modeling of electro-hydraulic servo system, the basic structure of neural network identification model of electro-hydraulic servo system is studied. The relationship between input and output of the dynamic model of servo system is analyzed. The neural network weight and threshold are optimized based on genetic algorithm to establish the basic structure of neural network identification model. The valve controlled cylinder real-time electro-hydraulic servo experimental bench is established by xPC technology. The step output signal of the test bench is taken as an improved BP neural network system identification signal, and the test bench sine output signal is used as a verification signal. The results show that the basic structure of the neural network identification model can reach a higher recognition accuracy, and its credibility can be verified, which is suitable for nonlinear system model identification.