Mathematical model by parametric identification and fuzzy controller design in the cooling section of a fluid heating-cooling plant

Abstract

In most industrial processes, it is required to know the dynamics of the process to achieve better adjustment and precision of the controllers; often, obtaining models by solving equations is complex to perform. With the development of technology, data can be obtained in real-time from the process. Identification methods are of great help in obtaining mathematical models experimentally, which is why engineering students must carry out practices in this countryside. A fluid heating-cooling plant is designed and built to have equipment for automatic control and instrumentation practices. The model was obtained from input-output data using parametric identification techniques; tests were carried out with ARX, ARMAX, OE, and BJ models, and the model was selected using similarity percentage, step response, and residual analysis. The best results were obtained with the BJ 22321 model, reaching a similarity percentage of 81.71%, for the model selection, parsimony was also considered, i.e., the one with the lowest number of poles and zeros. For the adjustment of the controller, the IMC-PID tuning was used. The response was tested with different controllers, and the PID-Fuzzy with 25 rules was selected; with this controller, a 10% surplus level percentage and a stabilization time of 200 seconds are obtained.