Networked computing-based system identification and control of electromechanical systems with industrial IoT

Abstract

A system is a set of elements brought together to perform a specific function or task. It is also an abstracted and modeled representation of a particular part of the real world. Electro-mechanical systems are systems created by integrating electrical and mechanical components. These systems are used to convert electrical energy into mechanical motion or mechanical energy into electrical signals. In this study, studies were carried out on resonant load and crankshaft electro-mechanical experiment systems in Istanbul Technical University (ITU) Rockwell Automation Industry 4.0 Control and Automation Engineering Laboratory. System identification, a fundamental aspect of control and automation engineering, involves constructing a model of the system using data that has been observed and collected. This process allows for understanding and predicting the dynamic behavior of the system. Models obtained through system identification are used for various purposes, such as fault detection, improving system performance, and applying adaptive control techniques. Thus, possible malfunctions can be detected in advance, performance improvements can be made, and control strategies that can adapt to changing conditions can be developed. It is necessary to guarantee the stability, durability, precision, and performance criteria of systems in system identification, control, and automation applications. Thanks to system identification, the properties, behaviors, and relationships of systems are described with mathematical expressions. In this way, analytical analyses can be made on the systems and more effective control strategies or automation processes can be developed. These mathematical expressions are differential equations or transfer functions for continuous-time system models; the difference equations or discrete transfer functions for discrete-time system models. This sequence of experiments (system identification experiment) includes the steps that must be followed to accurately model and analyze the system's dynamic behavior. Additionally, these experimental steps convert the system from a black box with unknown properties to a gray/white box with known properties. System identification experiments first start with determining the linear operating region of the system and selecting the sampling period. This linear operating region helps determine the selected input signals and experimental conditions to make the system model more accurate. The choice of sampling period is also important because it determines the time interval between data points that will be used to analyze the collected data and build the system model. This period should be small enough to accurately respond to the dynamic behavior of the system but, at the same time, make the data collection and processing processes effectively manageable. Obtaining the Bode diagram using the systems' frequency response is one method that can be used to select the sampling period. The second step of system identification experiments is selecting the input signal to be applied to the system.