Tuning of industrial controllers usign a multiobjective optimization paradigm

Number: 322-B4-218


Industrial processes are continually subjected to undesirable disturbances that make the system out of its normal operating point. It is for this reason that there are instruments called controllers that manipulate some variables of the process, so that the effect of these disturbances is counteracted. However, many of the controllers are not tuned in a suitable way, making the system inefficient or even prone to failure.

This project proposes a tuning algorithm that considers various disturbances sources simultaneously, so as to obtain the optimum parameters of an industrial controller. This will require to define a multiobjective optimization problem that includes all the requirements, both from the standpoint of performance and robustness. Finally, the main objective is to create a computational tool to assist in tuning PID controllers with two degrees of freedom for second order overdamped plants, using a multi-objective optimization paradigm, taken into account at the same time, at least three sources of disturbances besides the robustness.

Objectives

  1. Design, implement and test a program for optimizing the parameters of a PID controller with two degrees of freedom for at least 3 objective functions
  2. Create a database with the optimal parameters of 2DoF PID controllers for overdamped second-order plants
  3. Obtain tuning  rules as simple as possible, based on data obtained from optimization
  4. Design, implement and test an interactive software that allows you to select the best set of parameters of the controller to the control system under study

Although there are examples in the literature which has attempted to apply optimization techniques to control systems, there are very few that consider several sources of disturbance at the same time. In fact, many of them try to optimize different performance criteria, but only considering the response to a disturbance at the entrance of the system or to a change in the reference signal. This project will have a significant impact on the control community, since propose a generalization of the classical problem of servo-regulator PID controllers compromise. In addition, a visualization software which can be directly used by practicing engineers to tune PID controllers in the industry, is created.

Results

The results of the optimization can be found here, and can be freely used.


Researchers