Identification and Process Control with Artificial Neural Networks and PID Controller

摘要

The objective of this research is focused on the development of an automatic control strategy and algorithm to identify a process by using process historical records, as well as knowledge from the operation itself. The algorithm is based on artificial neural network (ANN). The output of the ANN feed a classic PID controller to perform control actions forming a hybrid controller. Here, a local control loop associated with the manipulated process variable will be modeled estimating its space-state dynamic equation. This kind of automatic control strategies can automatically control complex processes in steady state, for example, a process with high and rapid rate changes (high variance variables) and final control elements with large-time constants. The control strategy seeks to eliminate manual control actions in a steady state, and self-regulation regarding the variations for on or multiple process variables, identifying for this the process behavior to take automatic control actions. The proposed control algorithm was tested in the process of gases transport in a copper smelter plant at Empresa Nacional de Minera, ENAMI that was chosen for this purpose. The necessary data and scenarios to testing the proposed control algorithm was obtained. The chosen process consists of transporting gases from the Teniente converter hood to the acid plant inlet converting this process in a MISO (Multiple Input Single Output) system. This application will try to present a solution to inherent problems of this process like manual control, multiple key variables coexisting in a system, mechanical stress in equipment because of manual actions, et., The control strategy presented will be based on a computer simulation made with real process data as was mentioned before; it shows improvement of the transient periods in the final actuators due to the control signals, as well as it shows that these kinds of technologies could be implemented in the existing plant hardware/software or in a conventional control system.