Optimization Of Modeling Industrial Process Based On General Regression Neural Network

The working conditions (or operating conditions) of the industrial process are closed-loop feedback controlled by the controller. Changes in the environment, aging of the equipment and changes in raw material composition and so on factors make the slow disturbance to the industrial process. Therefor we should continuously change the controller setting according some provisions of certain indicators, which make the industrial process keep in optimal working condition in order to increase production, reduce raw material and energy consumption, as well as improve product quality, this is the purpose of optimization of industrial process. However, how to set up an accurate system model and find an effective optimization algorithm to complete the optimum design program are the two key issues to optimization.Artificial neural network is used to model intelligently focused on the characteristics of multivariate, nonlinearity and inaccuracy. Artificial neural network not only can approach any nonlinearity, but also has large-scaled parallel process, knowledge distributed store, strong self-learning and fault-tolerance well and so on. Where, multilayer feedforward neural network can approximate any continuous function at arbitrary precision. General regression neural network (GRNN) is a typical form, its structure and training are simple, but learning convergence is fast. Because of these advantages, this article uses GRNN to set up the model. Traditional general regression neural network has been improved, a new parameter-optimized rotated general regression neural network based on particle swarm optimization is advanced and simulation tests verify the result. In order to obtain the accurate optimization of parameters, this paper use particle swarm optimization to find the optimal values of the network parameter. The improved method can not only accurate fit, predicte, but also avoid the drawback of man-made selection structure parameters. At last, this method is used to simulate two examples. In this paper, the background of research is shells automatic injection lines of Heilongjiang Huaan Industial Company. Because the field test datas of the production line are not collected, so we can only determine the optimized parameters and objectives combined with the requirements of the production process. we use the method advanced in this paper to set up the model, the tese results prove effectiveness of the algorithm.