Desarrollo de un regulador neurodifuso adaptativo para el control de una planta de vuelo vertical de National Instrument.

Abstract

An adaptive neuro-diffuse regulator was developed, for the control of a vertical flight plant of National Instrument, by means of a controller of programmed gains or also called PID of gains programs with a fuzzy inference of the Takagi Sugeno Kang type of zero order that through its structure picks up the advantages of neural networks and fuzzy logic The plant modeling was started by means of the response of the system to a step input, obtaining a second order transfer function and which was then adjusted with the decreasing gradient technique. A classic controller was designed to observe the behavior of the plant. Then optimized with intelligent control techniques. Before getting to implement, the system was simulated with the controllers to verify the behavior of the plant and if it is appropriate to proceed to the implementation stage. The algorithms of the classic PID control and the Neuro-diffuse were programmed in the Lab view software, where the computer is responsible for carrying out the entire process and then send via serial communication port to the Arduino Uno. . This is responsible for receiving the PWM information to transmit it to the driver of the vertical flight plant engine. The loop closes when the angular position sensor sends the data acquisition card and it returns the information to the PC. The data obtained were evaluated using the integral criterion of the square of the error (ISE) and compared to each other, where the Neuro-diffuse regulator responds three times faster with the PID. Take into account in the sampling time for the acquisition of data to design the controllers.