Research Of Spheriodisation Treatment On Wire Injection Value For Ductile Iron Based On Neural Network

The wire-injection treatment technology has been used widely as an important spheroidisation in ductile iron. The accurate formula of wire-injection value hasn't been proposed and it is operated by empirical value.The way of measure system of wire-injection value is adopted based on the Artificial Neural Network (ANN) and Expert System in this paper. The ANN model is establied by means of Matlab software which can divine the parameters of wire-injection, the error percentage is within 7%.The ANN system, ratiocination, emulation, data save mode, interface design are discussed in this paper. The system uses series of Windows as its OS, Access as its DBMS and Visual C++ as its IDE and presents oriented-object C++ algorithm class. The system's generalization ability is analysized through verifying sample data, it turns out that the model of wire-injection which based on ANN, has goodness of fit, and the determine the wire-injection value. By using the wire-injection value model from simulation, the amount of melting iron, and it's the temperature, sulfur content are analysized. The range of melting iron temperature is confirmed according to the effect of the spheroidisation and economic performance, the result agrees with industrial production. It has been found that the amount of Mg is linear to the amount of melting iron, but when the amount of melting iron is lower than 4t, the actual amount of Mg is larger than the theoretical value and when greater than 4t, the amount of Mg agrees with the theoretical value.The ANN system in this paper will bring into a useful method in the establishment of ANN and its application on foundry, it can direct the industrial.