Study Of Heating Current Prediction Model In Electric Upsetting BP NN Based On Mechanistic Model

In the producing of the valve, the discarding rate in electric upsetting accounts for 50% of that of whole process, therefore, the quality of the air valve roughcast to that of the whole product rises the prominent function, and in order to decrease the discarding rate, we must improve the quality of roughcast. In the course of taking shapely electric upsetting, the velocities of upsetting cylinder and anvil cylinder, upsetting press, heating current and preheating length are the key parameters affecting the quality of air valve roughcast. The determination of those parameters is complex, because different materials have the different mechanics, electric and thermal characteristics, and the parameters are interactional together with interconditional. Besides, there are other accessorial parameters. Only on the condition that all parameters could be moderated, matched with mutually, can we produce the qualified product. Among the whole parameters, heating current is the most important. So it is significant that predict the model of heating current. Based on the principle of taking shape, this paper discusses the effects of parameters on product quality, puts forward the computing method of heating current, deduces mathematics model of heating current. Furthermore, the prediction model are also established by employing such four projects as BP NN, BP NN + mathematic model, BP NN * mathematic model, synthetic method of neural network and mechanistic model. Velocities of upsetting cylinder and anvil cylinder, pressure of upsetting cylinder used as input parameters, heating current was as target parameter. The results of the research could be applied to manufacturing practice directly, and advance developing of intelligence electric upsetting machine.There are four chapters in this paper, in the first section, isthe description of research recurrent situation and some documents, including the origin of the task and the significance of research. The second section describes the process of electric upsetting and hardware of autocontrol system. The third section describes the prediction theory of heating current, where method of establishing mathematics model is introduced. Then basic knowledge of neural network, BP network and its learning methods, and BP network' s designing are described. Lastly, the synthetic method of BP neural network and mathematics model is shown. The last section includes the application of the models, the simulation results and the discussion on on- line application technology of property heating current prediction.A new synthetic method of neural network and mechanistic model is chosen among the five models above. This method integrates the mechanistic model knowledge into the neural network structure , which can reduce the number of network weight , improve the learning speed, nonlinear approximation precision, as well as generalization performance of the network . And the super duper effect could be achieved if it is employed to predict the heating current of electric upsetting.