Metaheuristic algorithms for parameter estimation of DC servo motors with quantized sensor measurements

Abstract

Manufacturing, aviation, and robotics have increased servo motor use due to their precision, reliability, and adaptability in various applications. This study compares three metaheuristic techniques for servo motor model parameter estimation with sensor measurement quantization, focusing on their accuracy and efficiency. Armature resistance, back electromotive force (EMF) constant, torque constant, coil inductance, friction coefficient, and rotor-load inertia are crucial to servo motor behavior prediction, significantly impacting overall system performance. Each approach was rigorously tested and analyzed to evaluate its effectiveness in predicting servo motor characteristics. The results revealed that particle swarm optimization and the firefly algorithm delivered comparable performance, particularly excelling in scenarios where sensor measurement quantization introduced noise or imprecision in the data. These methods demonstrated strong resilience and accuracy under such challenging conditions. In contrast, the genetic algorithm did not perform as well, falling short when compared to the other two techniques in handling noisy or imprecise data, indicating its relative inefficiency in such environments. These findings give servo motor designers and engineers across industries a powerful tool for performance prediction.