High Robustness Of The Lag Process Control Methods

In the industrial processes, most of the plant has the characteristic of the time delay, which can increase the overshoot and adjusting time. In one word, time delay worsens the control quality. So, it is very difficult to control the processes with time delay. Since the end of 1950's, lots of methods for time delay system are proposed. The robustness is one of key questions which are decided whether these control methods can apply in reality, it is attentive topic to improve the robustness of time delay system. Therefore the research of the high robustness control methods for time delay system has important theory and practical significance.At present the main methods for processes with time delay are PID control and Smith predictive control. By approximating the pure time delay factor with Pade series, the parameters of PID controller for time delay system can be got based on Internal Model Control (IMC). There is only one tuning parameter, the method is easy and feasible and can apply in reality. The PID or PI controller of Smith predictive control system is designed based on optimal transfer functions. The method is simple and practical. The Smith predictive control system designed by this method is provided with good dynamic performance, anti-jamming and high robustness.After introducing the Internal Model Control, the Smith predictive control and the Dahlin control, the relationships between these control methods are studied. It is presented that these control methods are equivalent in some cases. It is shown that the Dahlin controller is a special case of the Smith controller and the Smith predictive control can run as an IMC control under some constraint conditions and the Dahlin control is equivalent to the IMC in case that the model matchs the plant.Based on Smith predictive control and IMC, the predictive controller with multiple internal model is put forward. The parameters of plant is tested on-line, the model fast follows the plant. The controller is suitable for processes with time delay, it simultaneously satisfies the requirements of follow-up control system and constant value control system with two-degree-of-freedom, it has high robustness.It can't get ideal control response for large time delay plant using PID controller, so the method of decreasing time delay controller is presented. Using decreasing time delay controller, the large time delay plant is remodeled into small time delay equivalent plant, and then the large time delay system can be controlled by PID controller. A new control system, which consists of the incomplete differential PID algorithm and BP neural network and decreasing time delay controller, has high robustness, auto-adapted ability, anti-jamming ability.After introducing ITAE optimal transfer function, Butterworth optimal transfer function, improved high order Butterworth optimal transfer function, a method of optimal control for process with time delay based on state feedback is presented. This method firstly deduces a mathematics model to approximate pure time delay, and then designs the controller and state observer according to optimal transfer function. A design method coorinating work state between observer and controller is proposed. Based on the state feedback optimal control system, adding an integrator, a kind of cascade control system with state feedback is presented for process with time delay. The system is a two-degree-of-freedom control system, it can effectively reject parameter variations and disturbance. Good dynamic response performance and robustness can be achieved simultaneously.The predictive controller with multiple internal model is applied to heating furnace, the new controller which consists of the incomplete differential PID algorithm and BP neural network and decreasing time delay controller is applied to heating furnace temperature control system in steel rolling workshop, the cascade control system with state feedback is applied to rotating stove temperature control system. The simulations show that these methods have high robustness.