Dynamic Modeling and Control Analysis of Industrial Electromechanical Servo Positioning System Using Machine Learning Technique

Authors:
Publisher website
Google Scholar
Abstract
Servo electromechanical systems are used in industrial automation to attain high accuracy, reliability, linearity, and high aspect ratio. Such technology possesses the advantage of compact structure and easy control over electro-pneumatic and electro-hydraulic systems. The major drawback of this technology is the high friction/vibration and also the jerk of servo electromechanical drives that are caused by load variation and speed regulation. When the load is varied, the force acting on the ball screw leading along the axial direction is varied, resulting in the creation of vibrations that lead to fatigue and wear. The major cause for this nature is magnetic loading and unloading capability of electrical machines, selection of controller tuning values, and feedback mechanism. It is necessary to control the magnitude of vibration to get smooth control on the toolpath during load variation. To arrest the vibration, the position control of the servo motor is implemented. In this proposed work, the design requirement of the servo mechanism, such as the stability of the driving mechanism, is examined in detail with mathematical modeling of the servo system. Simulation of the servo mechanism performance according to design and operating parameters is performed based on the derived mathematical model. To analyze the performance of the position control, gain-phase margin controller is compared with conventional Ziegler Nichols and auto-tune PI controllers. Further, the machine learning algorithm of K-means clustering is executed by taking the motor current parameter because the motor current is proportional to the torque, which gets direct impact by the load variations. Further, the cluster assignment on the motor current attributes is undertaken to infer either that the load variation is gradual or that it gives sudden fluctuations during the position control on the trajectory path.

This publication has 29 references indexed in Scilit: