摘要
在工业过程控制中,常常存在一些重要的变量难以测量,为了解决这个问题,出现了软仪表。软仪表的实质是建立被测量参数与影响该参数的其它操作参数之间的数学模型,通过计算得出此类难于测量的变量的数值。小波神经网络就是软测量的一种方法。在传统的小波神经网络的基础上进行了改进,利用小波对工业现场过来的数据进行了降噪,并使用主元分析法去除了数据的相关性。然后对处理过的数据建立小波神经网络模型,最后通过计算机仿真证实了该方法的良好的收敛速度快,不容易陷入极度最小等辨识效果。
In the industrial process control,there are usually some variables that are difficult to be measured, we often use the simulating instrument,The simulating instrument is to constitute a model of the measured parameters and the variables influenced by the parameters. Wavelet neural networks is a kind of simulating instrument. This paper presents an improved universal tool for wavelet neural network in which the data from the field are denoised and the variables are decorrelated by using PAC. Then the model of WNN is built and the computer simulation results show that the method is practical.
出处
《计算机仿真》
CSCD
2007年第4期152-154,共3页
Computer Simulation
关键词
小波分析
神经网络
主元分析
Wavelet analysis
Neural network
Principle component analysis
