DefinePK

Abstract

A large part of the industrial processes, when entering competitiveness, must be subject to flexibility so that related aspects can be adapted according to demands at the production level as well as current technological trends. One strategy to appropriate these processes is to adopt the use of control techniques such as Artificial Neural Networks (ANN) inspired by the biological neural networks of the human brain; its advantage is the ability to provide abstract dynamic features from a series of experimental data. Under this concept, an ANN controller system applied to a test hydraulic system was developed, which was compared with a classic PI strategy. Said comparison at the simulation level presented satisfactory results, demonstrating the quality and optimization in the processing, emulation, and control of a physical system with non-linear characteristics. The performance of the networks is noteworthy, the Tau response times for both controllers when the level of the tank decreases are similar, however, the settling time of the neural network was between 20% and 40% faster than the controller PI. The presence of overshoot above 20% was identified by the PI control in response to changes in the setpoint for the size of the tank level.