基于径向基神经网络的热力参数虚拟传感器.pdf

第 44 卷 第 6 期 汽 　轮 　机 　技 　术 Vol. 44 No. 6 2002 年 12 月 TURBINE TECHNOLOGY Dec. 2002 基于径向基神经网络的热力参数虚拟传感器 1 1 2 赵　瑜 ,胡念苏 ,周宇阳 ( 1 武汉大学动力与机械工程学院 ,湖北 武汉 430072 ;2 新华控制工程有限公司 ,上海 200245) 摘要 :针对火电厂汽轮发电机组经济性监测系统中参数失效的问题 ,提出了利用径向基函数神经网络的仿真手段 作为虚拟传感器。分析了径向基函数神经网络的原理与特点 ,总结了求解各层权值和阈值的数学公式 , 阐述了参 数学习和仿真的方法与过程。 关键词 :径向基 ;神经网络 ;汽轮发电机组 ;经济性监测 ;仿真 ;虚拟传感器 ( ) 分类号 :TK247 ;TK284. 1 　　　文献标识码 :A 　　　文章编号 2002 Parameter Simulation of Economic Monitor System of Steam Turbine Unit Based on Radial Basis Function Network ZHAO Yu1 , HU Nianshu1 , ZHOU Yuyang2 ( 1 School of Power and Mechanical Engineering , Wuhan University , Wuhan 430072 , China ; 2 Xinhua Control Engineering Co. ,Ltd. , Shanghai 200245 , China) Abstract :In order to solve the problem of failure measure parameter in the economic monitoring system of steam turbine unit , a method employing radial basis function neural network (RBFN) to simulate the parameter is proposed. The principle and charac teristics of RBFN are analyzed , the mathematics method to evaluate the network weights and biases of each layer is summarized , and the training and simulation methods and processes are expounded. Besides , a more convenient method that can use MATLAB to training the network offline was proposed. Key words :radial basis; neural network; steam turbine unit; economic monitoring; simulation ; virtual sensor 0 　前 　言