Cascade-Forward Neural Network Based on Resilient Backpropagation for Simultaneous Parameters and State Space Estimations of Brushed DC Machines

TL;DR

This paper introduces a neural network-based sensorless estimation method for brushed DC machines that simultaneously predicts speed, temperature, and resistance using resilient backpropagation and cascade-forward neural networks, outperforming traditional estimators.

Contribution

It presents a novel neural network approach combining CFNN and resilient backpropagation for simultaneous parameter and state estimation in brushed DC machines.

Findings

The proposed method accurately estimates speed, temperature, and resistance.

It outperforms traditional estimators in noise sensitivity and convergence.

The approach is suitable for integration into thermal monitoring and high-performance drives.

Abstract

A sensorless speed, average temperature and resistance estimation technique based on Neural Network (NN) for brushed DC machines is proposed in this paper. The literature on parameters and state spaces estimations of the Brushed DC machines, shows a variety of approaches. However, these observers are sensitive to a noise, on the model accuracy also are difficult to stabilize and to converge. Furthermore, the majority of earlier works, estimate either the speed or the temperature or the winding resistance. According to the literatures, the Resilient backpropagation (RBP) as is the known as the faster BP algorithm, Cascade-Forward Neural Network (CFNN), is known as the among accelerated learning backpropagation algorithms, that's why where it is found in several researches, also in several applications in these few years. The main objective of this paper is to introduce an intelligent sensor based on resilient BP to estimate simultaneously the speed, armature temperature and resistance of brushed DC machines only from the measured current and voltage. A comparison between the obtained results and the results of traditional estimator has been made to prove the ability of the proposed method. This method can be embedded in thermal monitoring systems, in high performance motor drives.