Optimal Design Of Controllers Based On Libraries And Intelligent Optimization

Automatic control system occupies a dominant position in the fields of military weapons and equipment, aerospace and industrial production. In modern industrial applications, there is a wide variety of control methods including PID control, fuzzy control, adaptive control, active disturbance rejection control and so on. In these control methods above, PID controllers are widely used in process control and motion control due to the characteristics of simple structure and robustness, and especially applied for the control systems with precise mathematical models. However, PID controller often can not achieve the optimum performance for complex and non-linear control systems because its structure is fixed and only parameters can be adjusted. In addition, a general process for controller design is that researchers rely on experiences to determine the control strategy as to object’s characteristics(delay, saturation, etc.) firstly. Then, they tune the parameters based on prior knowledge or intelligent optimization algorithms. Because of the limitations of prior knowledge and experiences, the structure designed in this way is limited within a small range and the type of structure is relatively simple, which leads to poor performance likely. Controller includes structure and parameters, while parameters can be tuned only if the structure is determined. To design a controller with better performance, the shackles of experience should be breaked and the structure should be designed automatically.Inspired by the law of nature, human beings designed intelligent optimization algorithms. They have strong versatility and are suitable for parallel computing, global optimization. Compared with traditional optimization algorithms, they are applied in the optimal design of control systems more and more, which improves the performance of systems effectively. Intelligent optimization provides an effective sulution for the optimal design of structure and parameters automatically.In order to design optimal controller automatically, the "library" is proposed at first and a component library, a performance index library and an optimization algorithm library are built in this paper. The component library contains basic units for consisting structure, the performance index library contains a variety of indicators for performance evaluation, and the optimization algorithm library contains algorithms for searching and optimizing the structure and parameters of controllers. In this paper, the design process of controller is that a plurality of basic units is selected from the component library as constructional elements of controller’s structure and to generate specific structure according to a given combined rules. And then, one performance index is chosen as the fitness evaluation function. At last, algorithm in the optimization algorithm library is called to optimize the structure and parameters.PID controller is constituted by three typical blocks, namely proportional, integral and derivative, which are connected parallelly. Drawing on this composition, this paper presents the optimal design of maximum structural paradigm controller. All components are connected in the form of parallel. Differential evolution is applied to optimize the parameters of all components. The integral of absolute error(IAE) is chosen as the performance index and simulations on several typical linear systems(non-minimum phase systems, high-order systems) are carried out. The performance is compared with that of PID controllers based on ZN tuning rule and intelligent optimization algorithms. Experimental results show that the maximum structural paradigm controller has a smaller IAE value which indicating better overall performance. The smaller overshoot, shorter rise time and settling time prove the effectiveness and superiority of this method. In addition, the maximum structural paradigm controller is also applied in a single-stage inverted pendulum system. It is found that the maximum structural paradigm controller can control the inverted pendulum system rapidly and stably. Also, the feasibility of the method in nonlinear systems is verified.In order to design controller’s structure in a more flexible way, the synchronous design of structure and parameters is studied and the double-deck optimization method is proposed in this artical. A modification of genetic algorithm is used to optimize controller structure in the outer, while differential evolution is utilized to find the optimal parameters for a defined structure in the inner. For a two-axis turntable control system, an angular velocity controller and an angle position controller are designed in the method, respectively. Results indicate that the two-axis turntable system has a fast response and can track sinusoidal signal precisely in this method.