Forming conditions and neural network control of continuously directional microstructure in directional solidification continuous casting process

Directional solidification continuous casting (DSCC) process is a new manufacturing technology for metallic materials which combines advantages of both directional solidification technology and continuous casting technology. Unlimited long shaped metal with directionally solidifying microstructure can be produced by this process. It is experimentally shown that controlling condition of stable and continuous growth of single crystal structure means the precise control of the location of the S/L interface, which is affected and determined by seven process parameters. Moreover, these parameters are also interacted each other, so the disturbance of any parameters may cause the failure of controlling of S/L interface. In this paper, on the basis of analyzing the forming conditions of continuously directional microstructures in DSCC process, the control model of DSCC procedure by neural network control (NNC) method was proposed and discussed. Combining with the experiments, we first used the computer to simulate the effects of the solidification parameters on destination control variable (S/L interface) and the interactions among these parameters during DSCC procedure. Secondly many training samples necessary for neural network calculation can be obtained through the simulation. Moreover, these samples are inputted into neural network software (NNs) and trained, then the control model can be built up.