Intelligent improvement of Kalman filter based on artificial intelligence for sensorless speed estimation and control of DC motors

The state estimation is considered as an essential and complex task for accurate and efficient plant control and monitoring in industrial applications. The measuring system including sensors is a significant investments for any control systems to monitor both non-measurable and measurable variables of state for dynamic systems. As a result, the limitation of cost can be reduced by using sensorless strategies that estimate variables of state. The aim of this paper is to implement an intelligent improved Kalman Filter (KF) based on different machine learning algorithms for sensorless speed estimation of DC motor. The intelligent methods are Artificial Neural Network (ANN), Adaptive Neuro Fuzzy Inference System (ANFIS), Genetic Algorithm (GA) and Particle Swarm Optimization (PSO). These algorithms are used to improve and tune the KF. To improve accuracy of estimation, the parameters of KF were optimized using PSO and GA. The research explores three kinds of architectures of ANN were implemented and compared with ANFIS to estimate the motor speed, employing collected data that involved voltage, current, and outputs speed of traditional KF. The models were tested and evaluated using multiple error criteria metrics. Results indicated that ANN-based Baysian Regulation Algorithm (BR) significantly outperformed other models, achieving minimum values of error metrics. The proposed intelligent sensorless speed estimation based on ANN-based BR strategy proves potential as adaptive solution, accurate, and cost-effective methodology for speed control of DC motor. The study findings offer valuable and distinct insights for investigation cost-effective and efficient sensorless cost-effective and efficient sensorless control schemes.