Intelligent Predictive Control And Its Application Research

This paper summarizes the current research situation of intelligent predictive control. After this, the neural network nonlinear prediction based single neuron control and fuzzy logic control methods are proposed for the nonlinear three-tank system, and the experiment research is carried through.Firstly, the neural network nonlinear prediction based single neuron control is proposed for the nonlinear three-tank system. It includes two parts: the neural network predictive model, which predicts the nonlinear system output; the single neuron controller, which does the optimization operation according to current error, error variation, error variation rate and predictive error, and then gives out the control output. The simulation result shows that this method has quicker response speed, smaller overshoot and better adaptability for time-delay, and has the bright future in real-time control.Secondly, the neural network nonlinear prediction based fuzzy logic control is proposed for the nonlinear three-tank system. It includes two parts: the neural network nonlinear predictive model, which predicts the nonlinear system output; the weighted fuzzy logic controllers, which do the fuzzy operation according to predictive error between set value and predictive value, and then give out the control output. The simulation result shows that this method has excellent tracking performance, anti-interference capability as well as good dynamic and static performance indexes.Finally, according to the real three-tank system, construct the system hardware, design the MCGS monitoring interface, write the script algorithm program and do the experiment research using the second control strategy. The experiment results show this method has quicker response speed, shorter adjusting time and stronger disturbance rejection performance than traditional FC method or PID method, and this method is effective for the control of real three-tank system.