摘要
针对人工神经网络在结构系统辨识中存在的问题,提出一种基于BP神经网络的跟踪辨识方法.通过将实际结构模型分为一个机理模型和一个实时误差模型,前者基于常规的BP神经网路通过离线训练而成,而后者通过小型的BP神经网络实时辨识系统误差,进而使这种经过改进的系统识别网络能够具有动态递阶识别系统的能力.计算机仿真分析表明,这种方法可有效地减小由于不同外荷载作用引起的结构系统辨识误差,提高人工神经网络在系统辨识中的精度和可靠度.
In this paper, in order to solve the existing problems of the application of the artificial neural network to the identification of structure systems, a tracing identification method to the structural systems is suggested based on BP neural network, which divides the actual structure into a mechanism model part and a time-varying error model part. The former is created off line on the basis of a computing model, which is in accordance with the actual situation of the structure systems, and the latter is identified on line by using a small-scale BP neural network, which employs the system identification ability of the dynamic recurrent network. Through analysis of computer simulation, it indicates that this method can effectively reduce the identification errors caused by the action of different earthquake loads, which improves prominently the precision and reliability for artificial neural network in identifying the structure systems.
出处
《郑州大学学报（工学版）》
CAS
2005年第1期50-53,共4页
Journal of Zhengzhou University（Engineering Science）
基金
国家自然科学基金资助项目(50078037)
