|
Rolling complex industrial production process is a complex large system composed of multiple subsystems process , with severe nonlinear uncertainties , time-varying characteristics of the large time delay and strong coupling , and multi-parameter . Is very necessary for the realization of the a rolling complex industrial production process of quality control , establish its quality model for complex industrial processes , due to its complexity , coupled with industrial noise pollution , its quality modeling there are great difficulties . In recent years , many scholars at home and abroad using different neural network to quality modeling played a certain effect in the use of the built model to guide the production practice is still a certain distance . In this paper, based on the characteristics of rolling production process , using BP neural network , using wavelet and genetic algorithm to improve based on modeling of rolling industrial production process, product quality , meaningful results , and its main work and reads as follows: ( 1) for rolling complex industrial production process , its product quality modeling problem analysis , correlation analysis method to conduct a detailed study and analysis of the factors that affect rolling complex industrial production process, product quality , and to determine the 32 main factors affect the rolling complex industrial production process, product quality , laid the foundation for the establishment of the model and the implementation of quality steel rolling production process of product quality control . (2) rolling production process, product quality modeling method based on neural network research with BP network and BP neural network based on wavelet rolled steel production process, product quality , on the basis of the rolling product data preprocessing , modeling and its instance of simulation and analysis of simulation results and their analysis shows that the validity and reliability of the modeling approach . (3 ) on the basis of improved genetic algorithm (Improved Genetic Algorithm, IGA) optimization WNN (Wavelet Neural Network, WNN) structure analysis , the study gives Based on IGA - WNN product quality modeling methods , the advantages of the method has both the global search ability of genetic algorithms and WNN learning algorithm is simple and effective , the simulation results show that the validity and reliability of the modeling approach .
|