Dc motor identification based on artificial neural networks implementation in a microcontroller

Abstract EN

In the last few decades neural networks have been involved in many fields of researches and their application was restricted to be embedded on a computer.

This paper describes implementing of artificial neural network (ANN) based on a low cost microcontroller to identify a DC motor model.

The identifier will act as path for the output error feedback in Model Reference Adaptive Controller or to build an inverse controller.

To reduce the size of the proposed system, the DC Motor is controlled conventionally using the method of Pulse Width Modulation technique on the same microcontroller.

The proposed neural network is a multilayer feed-forward network which is of three layers (1-3-1) structure.

The training technique used the back propagation algorithm by using C program.

Identification process has been implemented successfully in the microcontroller and validated with different randomly picked data with least mean square error of 0.0007.