基于结构的神经网络在系统建模中的应用

Application of Neural Networks Based on System Architecture in Nonlinear Modeling

针对复杂非线性系统的建模问题,提出一种基于结构的神经网络建模方法。将一个复杂系统转化为若干个较简单的子系统,分别用函数链神经元建立各子系统的模型,然后根据子系统间的相互作用关系连接成一个完整的网络,即基于结构的神经网络。分析了网络的性能,探讨了在复杂系统建模中的应用。对Y2-Hc10型先导式溢流阀的仿真表明了该方法是可行和有效的。

A novel method of nonlinear system modeling is presented, which employs neural networks based on system architecture. This method translates a complicated system into some simple sub-systems. Each sub-system is modeled with a functional link neuron. Then the neurons are connected into an integrated network according to the relationship between these sub-systems. The performance of the networks is analyzed, and its application in complicated system modeling is discussed, too. The feasibility and validity of the method is illustrated by a simulation on Y2-Hc10 pilot operated relief valve.