遗传优化BP神经网络在电液力伺服系统辨识中的应用  

Application of BP Neural Network Based on Genetic Optimization in Electro-hydraulic Force Servo System

作　　者：周挺 王虎 ZHOU Ting;WANG Hu(Aviation Industry Aircraft Strength Research Institute,Xi’an 710065,China)

机构地区：[1]航空工业飞机强度研究所,西安710065

出　　处：《机械工程师》2020年第12期60-63,66,共5页Mechanical Engineer

基　　金：航空工业强度所创新基金(20170923001)。

摘　　要：介绍电液力伺服系统的结构及原理,针对系统建模中参数时变和非线性问题,采用BP神经网络进行系统辨识建模;利用遗传算法优化BP神经网络,克服单纯BP算法容易局部收敛、训练速度慢的问题;借助MATLAB神经网络工具箱、全局优化工具箱编写系统辨识算法,建立系统的神经网络辨识模型。分析神经网络模型辨识结果,将其与ARMAX线性参数模型的辨识结果作对比,验证遗传优化BP神经网络系统辨识建模的高效性和适用性。The structure and principle of electro-hydraulic force servo system are introduced.BP neural network is used for system identification modeling aiming at parameter time-varying and nonlinear problems in system modeling.By using genetic algorithm,BP neural network is optimized to overcome the problem of easy local convergence and slow training speed.With the help of MATLAB neural network toolbox and global optimization toolbox,system identification algorithm is written and neural network system identification model is established.The identification results of the neural network model are analyzed and compared with those of ARMAX linear parameter model to verify the efficiency and applicability of genetic optimized BP neural network system identification modeling.

关 键 词：电液力伺服系统 BP神经网络 遗传算法 系统辨识建模 

分 类 号：TP273.4[自动化与计算机技术—检测技术与自动化装置]