基于BP神经网络仿真MR阻尼器逆动力性能

The Simulation of Inverse Dynamics of MR Damper Based on BP Neural Network

MR阻尼器的力学模型都是以电压为已知量,来求阻尼器的出力。而在结构控制中,通常是由控制算法先求出需要的控制力,由此控制力反推出相应的电压,从而控制阻尼器的输入电压来使其产生需要的力。由于MR阻尼器是一种强非线性半主动控制装置,由阻尼器的阻尼力反推其输入电压是一个复杂而困难的问题。本文利用神经网络强大的学习、非线性拟合等功能来模拟其逆动力性能,解决由力反推输入电压的难题。结果证明,神经网络可以很好的仿真MR阻尼器的逆动力性能。

In structure control,the control force is firstly determined in the control algorithm,then the MR damper provides its demanded force by the change of input voltage.But for the existed models of MR dampers,the force is derived from the voltage.Because of the strong nonlinear relationship,it＇s difficult to derive the voltage from the force for the MR dampers.In this article,the problem is handled by simulating the MR dampers＇inverse dynamics based on the neural network＇s functions such as strong studying and nonlinear fitting etc.The simulation conclusion is that neural network works well to simulate the inverse dynamics of MR dampers.