基于CNN-GRU控制算法的建筑隔震结构振动控制研究  

Research on vibration control of building isolation structure based on CNN-GRU control algorithm

作　　者：孙召英 山锋[1] 刘启泓 李玉峰 SUN Zhaoying;SHAN Feng;LIU Qihong;LI Yufeng(Shaanxi College of Communication Technology,Xi’an 710018,China;China Jikan Research Institute of Engineering Investigations and Design Co.,Ltd.,Xi’an 710005,China)

机构地区：[1]陕西交通职业技术学院,西安710018 [2]机械工业勘察设计研究院有限公司,西安710005

出　　处：《自动化与仪器仪表》2025年第3期220-223,共4页Automation & Instrumentation

基　　金：陕西省教育厅专项科研课题《TaC/TLSiC,陶瓷基复合材料的制备及其性研究》(18JK1101)。

摘　　要：地震是对建筑危害较大的自然灾害,但现有的建筑结构振动控制方法效果不佳。针对此问题,研究提出门控循环单元和卷积神经网络结合的结构振动控制系统,使用门控循环单元作为隐藏层,优化隐藏层数量。实验结果表明,门控循环单元和卷积神经网络控制的结构位移比优化前低0.005 m,预测精度提高了0.001,容错机制的位移误差不超过0.001 m。由此可得,门控循环单元和卷积神经网络控制算法能够有效预测建筑结构的位移变化情况,控制系统的建筑结构位移和加速度控制效果更优,能有效降低建筑结构的位移变化峰值和加速度变化峰值。优化后控制系统的效果评价指标在低楼层表现更好,控制系统的容错机制具有较高的稳定性和良好的振动控制效果,能有效保障系统的稳定运行。Earthquake is a natural disaster that does great harm to buildings,but the existing vibration control methods of building structures are not effective.To solve this problem,a structural vibration control system combining gated cycle unit and convolutional neural network is proposed.The gated cycle unit is used as the hidden layer to optimize the number of hidden layers.The experimental results show that the structure displacement of gated cyclic unit and convolutional neural network control is 0.005 m lower than before optimization,the prediction accuracy is improved by 0.001,and the displacement error of fault tolerant mechanism is no more than 0.001 m.It can be concluded that gated cycle unit and convolutional neural network control algorithm can effectively predict the displacement change of building structure,and the displacement and acceleration control effect of building structure in the control system is better,and can effectively reduce the displacement change peak value and acceleration change peak value of building structure.After optimization,the effect evaluation index of the control system is better in the low floor,and the fault-tolerant mechanism of the control system has higher stability and vibration control effect,which can effectively ensure the stable operation of the system.

关 键 词：门控循环单元 卷积神经网络 振动控制 建筑隔震 容错机制 

分 类 号：TD67[矿业工程—矿山机电]