Adaptive Smith Predictor Based Fast Converging Genetic Algorithm

Smith predictor provides an effective method to improve for plants with pure delay in theory, but it depends on the accuracy of the model too much. However genetic algorithm (GA) has quite good robust and optimization. So an adaptive Smith predictor based fast converging genetic algorithm (GA) for a class of systems with pure delay is proposed. An improved genetic algorithm (GA) is applied for system identification online, which fitness may decrease or increase automatically to get the fast convergence, and probabilities of crossover and mutation are adaptive along with the evolution proceeding to get the global astringency. So the adaptive Smith predictor can estimate the dynamic model to compensate delayed time. This design is applied to optimize controller of furnace. It shows that the control effect is better than that of traditional PID controller. It has strong robust and restrained from disturbance.