树形神经网络的学习机理及其在电液伺服系统控制中的应用

针对具有多维输出信息的系统，采用了一种树形神经网络的构造方法，并解释了其物理意义。此外，本文还提出了该网络的一种新的训练结构，与传统的训练方法相比，保证了系统的稳定性与安全性，加快了网络的收敛速度，避免了训练过程中的虚收敛点，同时提高了神经网络的在线学习和实时控制性能。最后给出了针对电液伺服系统的仿真结果。

Learning mechanism of a tree structure neural network and the application of the network in electrohydraulic servo system control

Aiming at a system with multi-dimensional output, the construction method of a tree - structure neural network is adopted, and its physical meaning is explained.Moreover, a new training structure of hte network is put forward in this paper. Compared with the traditional training method, the stability and safety of the system are guaranteed,the convergence is speed up. the virtual convergence point is avoided, at the same time, the ability of on line learning and real time control of the network is enhanced. At the end of the paper, a simulating result for an electrohydraulic servo system is offered.