MR阻尼器的性能测试与逆模式神经网络智能控制方法的研究

Test of MR damper and intelligent control by using inverse neural network

自制了内置弹簧蓄能器的新型MR阻尼器,将自制的MR阻尼器在武汉理工大学测试中心完成了性能测试。研究了在不同电流输入下阻尼力-位移、阻尼力-速度之间的关系,提出了MR阻尼器的修正的Bingham力学模型。在此基础上,采用逆模式神经网络来模拟阻尼器的逆向模型,与其它主动控制算法形成了MR阻尼器的智能控制方法。通过对一栋三层剪切型结构进行仿真计算,我们知道通过逆模式神经网络可以得到连续的控制电流,实现阻尼力的连续可调,控制效果将优于半主动控制效果。

In this paper, a self-developed MR damper is tested at measure central in wuhan university of technology, force-displacement and force velocity relationship at different current is researched, and then, establish the revised Bingham model for MR dampers. On base of this, inverse neural network trained with data set calculated from revised Bingham model is used to simulate inverse dynamics of MR dampers. After the inverse neural network model is well trained, it can be linked with active control algorithms for control practice to form the intelligent control method. In a numerical example, the linear quadratic regulator （LQR） method is adopted to control vibration of a multi story structure, in which the inverse neural model gives continuously changing electrical current signal as expected. The control effect is rather satisfactory than semi-active control.