Design of a neurocontroller and its application to time control of a prototype identified system using graphical programming

Abstract

The design of a control system in certain processes in the industry is the main problem for engineers. Since the input and output data of a certain process are obtained, we can know the response of the system, the same that allows us to study a model prototype and using the theory of nonlinear control the limit cycles are predicted, we can propose a function descriptive whose parameters will be necessary to implement the Neurocontroller that will be trained with the Back propagation algorithm. In our case, we address the design and implementation of a Neurocontroller in the LabVIEW graphical program environment under the supervision of the Simulation Toolkit. Our application is the control of a prototype physical system whose model can represent first and second order systems. The model of the process dynamics was obtained experimentally, for which the identification of parameters was applied using the curve fitting method by linear interpolation in the frequency domain. The accuracy of the model has been essential for the analysis of the descriptive function and consequently the adaptation of a neural model. The experimental results demonstrate that the designed control signal can make the output of the prototype system efficiently follow the imposed references with minimal overshoot and null steady-state error.