Adaptive neural network controller for the flush material Belt Weigh Feeder

The flush material belt weigh feeder (BWF) is used in many material handling plants. The stability and the performance of the layer control system will affect the quality of the production. In general, the behavior of the flush material on the BWF is non-linear, time-lag, and disturbance character. The layer of the flush material on the belt is hard to be stably controlled especially the occurrence of unstable situation in flush material or the Prefeeder feeding rate. This paper adopted two adaptive neural network (ANN) controller in the general PID control system to improve the performance. The proposed ANN controllers are utilized instead of the original PID controllers to deal with the problems of the unstable pheomena. This paper offers the simulation result of the material character (coarse size, moisture content) or the Prefeeder feeding rate stable and unstable condition (involve the Interference Signal) for the original and the proposed control system. The simulation results are distinct better than the original control system. The variations of the material character (coarse size, moisture content) or the Prefeeder feeding rate can be successfully predicted by the ANN controllers and adjusted the proportional actuator immediately. Finally the proposed control system remains convergence with smooth and stable. Therefore, the accuracy, stability and the performance of the control system are improved.