The Research Of Fractional PID Controller With Computational Intelligence

PID controller is a widely used control strategy in the field of process control. With the industry production scale expanded and product refinement enhanced, about10%-20%of the process control can’t simply use the traditional PID control strategy. So it is necessary to look for more advanced control strategies. The fractional order PID controller is one of the advanced control strategy put forward in recent years. The introduction of two fractional orders makes the fractional order PID controller has a more flexible and adjustment range and a more wide application prospects, but at the same time, the setting of the complex five-dimension parameters is also one of the difficulties to restrict its application in practice. In order to solve the problem, this article studies from the fractional order calculus and use the knowledge of computational intelligence to put forward a method of combining the off-line and on-line setting of parameters to optimize the complex parameters of fractional order PID controller. Finally through the test with hardware platform, control effect of traditional PID controller and fractional order PID controller were compared.This paper studies the relevant contents of fractional order calculus and combined with the traditional PID control theory, and then discusses the structure and mathematical expression of the fractional order PID controller. An improved operator approximation method is used to approximate operator calculus, and then make the simulation with Matlab/Simulink to build the fractional order PID controller with five parameters of Kp, Ki, Kd, α, β, which is used as the simulation model of the off-line setting of the parameters in the next step.Use the method of off-line optimization to get the initial sample of five parameters. In order to search the optimal parameter of the fractional controller, this paper proposes an intelligent algorithm, glowworm swarm-genetic algorithm, which combines individual evolution and swarm intelligence. Take the traditional genetic algorithm and hybrid particle swarm optimization algorithm as the comparison algorithms, and find the optimal parameter setting of fractional order PID controller in the environment of simulation with Matlab/Simulink. The simulation experiments prove that the glowworm swarm-genetic algorithm has more powerful searching abilities.In order to apply the fractional PID controller to the actual system, this paper designed a BP neural network fractional order PID controller which can optimize the parameters on-line. In order to improve the precision and speed of on-line parameter setting, the off-line optimal parameters which are searched by glowworm swarm-genetic algorithm are used as the initial learning sample for the neural network. Simulation experiments prove that BP neural network fractional order PID controller is better than traditional PID controller in reducing the dynamic-state error.Take the brushless DC motor system as hardware test platform and platform system transfer function is built. Search optimal parameters of fractional order PID controller off-line with the glowworm swarm-genetic algorithm then take searching results as initial learning samples for BP neural network PID controller on-line. Test the control effect of traditional PID controller and BP neural network controller in the hardware testing platform. After tests, we found that the BP neural network fractional order PID controller can shorten the adjustment time by76.9%than traditional integer order PID controller when the system starts at the speed of1500r/min, and shorten the adjustment time by63.6%and reduce the overshoot by the amount of55%when system is given sudden load of0.5Nm. This research provides some reference for the application of fractional order PID controller.