论文部分内容阅读
用神经网络方法对系统状态转移矩阵进行识别的研究，首先建立结构动力学运动方程的状态转移矩阵的识别与神经网络的权值矩阵识别间的相互等价关系，其次利用神经网络功能强大的并行运算能力和丰富的学习功能以单层线性神经网络来识别线性时不变系统的广义状态转移矩阵以获取结构的状态参数。最后通过识别获得的神经网络仿真系统结构在任意激励下的响应，可用以抗震性能评估或振动控制。
The neural network method is used to identify the state transition matrix of the system. Firstly, the mutual equivalence between the state transition matrix identification and the weight matrix identification of the neural network is established. Secondly, with the powerful parallel of neural network Computing Capabilities and Extensive Learning Capabilities The monolayer linear neural network is used to identify the generalized state transition matrix of linearly invariant systems to obtain structural state parameters. In the end, the neural network simulation system is used to simulate the response of the system under arbitrary excitation, which can be used to evaluate the seismic performance or control the vibration.