PID controller tuning algorithms based on Artificial Intelligence techniques

One of the common challenges in industrial process control is tuning PID controllers. In many instances, the most experienced operators rely on empirical knowledge or classical methods like the widely used Ziegler Nichols method for its simplicity and popularity. Nevertheless, in spite of the used values in the processes are useful, they are not the best, thus is necessary to find new ways of optimizing these values in order to improve the performance and robustness of the system, for instance, no energy is wasted on plants, and it is possible to increase the life of the equipment. This thesis presents the study of various artificial intelligence techniques in order to determine the most effective approach for optimizing and refining the PID control procedure. The performance of the designed controller is compared with another controller tuned via the Ziegler-Nichols method. This is achieved through the utilization of performance indices error such as ISE, ITSE, IAE, ITAE and system response metrics including Overshoot, Rise time, and Settling time. The study is carried out on two processes, the first one a conventional Mass Spring Damper model constructed using Simscape Multibody - MATLAB and the second one a complex dynamic system of a pendulum driven by a hydraulic piston.