Implement of neural network predictive controller based on Matlab and VC++ mixed programming

Industrial control requirements have been elevated from basic control to an optimal control height. Various advanced control algorithms, such as predictive control, has obtained plenty of approvals in industrial control with its good control effect. Advanced control algorithms are difficult to be used in practical industrial control effectively at present. To solve this problem, a new method is proposed based on mixed programming technique of Matlab and VC++. Taking Siemens S7-300 PLC as the controller, neural network predictive control can be achieved after registering the algorithm´s ActiveX in the SCADA software of WinCC. The operation can be detached from the experimental environment of Matlab completely. An experimental verification is conducted in the four tank water system. The results indicate that the method is feasible and effect of control is excellent. Predictive control algorithm, for example, is widely accepted in industrial control with its satisfactory control effect.