【摘要】 为了有效抑制多层建筑的地震响应,提出了自适应神经网络控制(MGA_ANNC)策略.首先对非确定性和非线性结构的参考轨迹进行了追踪,并采用径向基函数网络来保证追踪的精度.然后利用改进的遗传算法(MGA)对结果参数向量的初始值进行了选定.最后结合改进的剪切最优(MCO)控制算法提出了适合调谐质量-磁流变阻尼器(TM-MRD)的MGA_ANNC/MCO半主动控制算法.分别对一座9层框架结构在无控制、MGA_ANNC/MCO半主动控制、MGA_ANNC主动控制和LQG主动控制下的各项评价指标值进行了计算.结果表明:MGA_ANNC/MCO和MGA_ANNC的减震效果均比LQG的要显著.更多还原

【Abstract】 To effectively suppress the seismic responses of multi-storey buildings,a newadaptive neural network control strategy based on the modified genetic algorithm( MGA_ ANNC) was proposed. First,the reference trajectory of uncertain and nonlinear structures was tracked,and the tracking accuracy was ensured by using the radial basis function network. Next,the initial values of consequent parameter vectors were selected by using the modified genetic algorithm. Finally,the semi-active control strategy MGA_ANNC / MCO used for TM-MRD was proposed by using the modified clipped optimal( MCO) control algorithm. The various evaluation criteria of a 9-storey frame structure under the non-control,MGA_ ANNC / MCO semi-active control,MGA_ ANNC active control and linear quadratic gaussian( LQG) active control were calcalated,respectively.The results indicated that the seismic reduction effects of both MGA_ ANNC / MCO and MGA_ANNC are obviously superior to those of LQG.更多还原