Product Quality Control For Cluster Hot Tandem Rolling Based On Fuzzy Wavelet Neural Network

The cluster hot tandem rolling processes is an important processes in the smelting the steel and iron.The burning hot billet is extruded under the large-scale cluster hot rolling mill so that it's thickness and width is ok.The modern society'need to quality of the steel production is updating,to improve the quality of the productions of iron and steel enterprise,Experts study the line of the product, according to the theoretical analysis and the actual operating experience, and abstract an model from the entire production line that has many key input variables and many main output variables and then find the nonlinear function relations between them. To find this key input variable and this main output variable nonlinear function relations,and to obtains the stable quality model using in the industrial production is always a problem for the world various Iron and steel enterprise.Following the progressing of the computer technology, artificial intelligence technology. The artificial neural networks, the fuzzy logic and the wavelet transformation already have the fast development,and have applied in each kind of control system, the analysis system and approaches.There are some attmpts in the areas of quality controlling process for cluster hot tandem rolling and had some chievements.But the process of the production of the icon and smelt have many misalignment, time-dependent factors and some random disturbances. So the theoretical model which have used in other areas is difficult to apply instructing the practice in the actual production process or its effect is not obvious.The wavelet analysis has the multi-resolution characteristic and the time-frequency localization characteristic, especially useful in qualifying the misalignment signal. The fuzzy neural network not only has the ability of mapping the misalignment that artificial neural networks has, but also has the superiority like the fuzzy logic portray to classify the fuzzy boundary. Based on mature wavelet analysis theoretical, fuzzy logical and the artificial neural networks,The paper has conducted the research regarding the wavelet analysis, the fuzzy logic and the artificial neural networks in the cluster hot tandem rolling production line to control the quality of the pruduction.And the paper has studied one kind fuzzy wavelet neural network model to apply in the cluster hot tandem rolling about controlling its quality. Therefore the paper has mainly done following work:At first it has studied the artificial neural network,the wavelet neural network and fuzzy reasoning about their algorithm research. Find the way to hidden layer of wavelet neural network.It has given fast algorithm for fuzzy neural network and BP neural network.Second, it has analyzed a steel multi-roll hot rolling process, make sure 32 key input variables and four output variables that affect the quality of hot-rolled products,and conducted data pre-processing.Finally, it has constructed a fuzzy wavelet neural network structure; has built the fuzzy wavelet neural network model using the data samples that were got in the steel gas bottle, the results show that the training sample hit rate is 94.3% and the test sample hit rate is 93.3% in line with business requirements.Next to do is to use the built fuzzy wavelet neural network model in the product quality control for steel business.