Forecast Modeling Establishing of Friction Welding Attachment Intensity Based on Compensatory Fuzzy Neural Network

Through carrying on the friction welding to the 45 steel and the W8Co3N high-speed steel and measuring the joint, the data can be obtained which the network simulation need. With fuzzy logic and neural network, we can establish the compensation fuzzy neural network model which has used in the welding process parameter forecast, and carry on the simulation by using it. The result indicates that this model may carry on a more accurate predict to the welding process parameters and which has the merit that the model is simple, the forecast speed is quick, forecast precision is high and pan-ability is strong. Thus it can provide an effective way for intensity of the welding joint.