The Research On Prediction Of The Rolling Force Based On Wavelet Neural Networks And Automatic Gauge Control System

With the technology development at very fast speed and production scale-up, a lot of industrial systems become more and more complex. Therefore, traditional models and modeling theory already far from can't adapt to the requirements presented by complex industrial processes. Neural network has very strong nonlinear approximation ability, self-learning ability, self-adaptive and stronger fault tolerance, thus it can trail and catch object's change and change tendency very well because of the influence of various uncertain factor. But when it is applied in practice, the training time of network is long, convergence speed is slow and local minimum value can't be avoided, so that its application have been restricted in the modern industrial processes. In order to make neural network get better application, wavelet transform is led into neural network to form a new type wavelet neural network. Wavelet neural network combines the local property of wavelet transform with the advantages of neural network, it has better approximation ability, and fast convergence speed and can avoid the local minimum value efficiently. In the hot strip mill process, the reduction computation is preset by the rolling force allotment, therefore the rolling force is a very important parameter. Because of the spring phenomenon in the slab and strip rolling, the prediction of rolling force has became the core of models in the finishing rolling mill preset. Therefore its prediction precision directly influences the preset of the roll-gap, sequentially it can influence thickness precision, the shape of strip quality and the stability of threading. Because the rolling process is the multivariable, the non-linear, the time-variable and the strong coupling, the traditional modeling methods have been far from adequate to meet the requirements of the high accuracy predicting rolling force.Combining with the characteristics of the complex industrial process, and taking the prediction of the rolling force as the object, this thesis aims at studying and discussing complex system modeling based on the wavelet neural network, and the predicted rolling force will be applied in the automatic gauge control system.The main contributions of the thesis are as follows:(1) After going to a hot strip mill factory production, the rolling technologies and the computer control system are studied, the computer control system is the SIROLL.(2) The way of the combination with neural network and the wavelet are studied, wavelet neural network structure and the study algorithm are embedded analyzed and researched, and a kind of fast BP algorithm is obtained. (3) Aiming at the modeling questions in the present complex industry processes, and taking prediction of the rolling force as the object in the research process, the wavelet neural network through the Matlab simulation experiments are studied, the choose of the network structure, wavelet function, network study algorithm, network training objective, the parameter the coefficient and values in network etc are mainly studied, and the studied wavelet neural network is applied to predict the rolling force.(4) The automatic gauge control system in the hot strip milling is studied, and the predicted rolling force and the hydraulic automatic gauge control technology are used to realize the gauge setting and controlling of the strip. In view of the problems that exist in the current system, the corresponding improved measures are proposed, and some of them have already been adopted. Thus the precision of automatic gauge control is effectively enhanced in a hot strip rolling mill.(5) As the conclusion, a summary of the research works is given, and some future research directions are forecast in the thesis.