Geometric Analysis of Signals for Inference of Multiple Faults in Induction Motors

Abstract

Multiple fault identification in induction motors is essential in industrial processes due to the high costs that unexpected failures can cause. In real cases, the motor could present multiple faults, influencing systems that classify isolated failures. This paper presents a novel methodology for detecting multiple motor faults based on quaternion signal analysis (QSA). This method couples the measured signals from the motor current and the triaxial accelerometer mounted on the induction motor chassis to the quaternion coefficients. The QSA calculates the quaternion rotation and applies statistics such as mean, variance, kurtosis, skewness, standard deviation, root mean square, and shape factor to obtain their features. After that, four classification algorithms are applied to predict motor states. The results of the QSA method are validated for ten classes: four single classes (healthy condition, unbalanced pulley, bearing fault, and half-broken bar) and six combined classes. The proposed method achieves high accuracy and performance compared to similar works in the state of the art.

Fault diagnosis in asynchronous motors based on an optimal deep bidirectional long short-term memory networks

“…It is difficult to develop a unified theoretical model for these models; More generally, the mechanisms and operations of numerous smart swarm optimization algorithms tend to be homogeneous, and a large number of current algorithms are current disguises of older ones. Therefore, the task of the current generation of intelligent optimization algorithms should be to establish universally applicable rules and strategies, focusing on optimization problems and algorithmic common problems; (5) In the modified 3D Logistics-sine complex chaotic map, the distribution of chaotic attractors in its chaotic track diagram was sparse in the middle and tight around, indicating that its distribution was not uniform. Therefore, the mathematical model of 3D logistics-sine complex chaotic map can be further improved.…”

“…The fault index is chosen based on the analysis of the fault index, which provides a valuable reference for motor fault diagnosis. Contreras-Hernandez et al [5] proposed a multi-motor fault detection method based on quaternion signal analysis (QSA). In this method, the motor current and the measured signal from a triaxial accelerometer mounted on the chassis of the induction motor are coupled to the quaternion coefficients.…”

Detection of Broken Bars in Induction Motors Using Histogram Analysis of Current Signals

“…Additionally, most of the traditional signal processing techniques are used in the time, frequency, and time-frequency domains; the choice of the domain depends on the information required to perform motor fault classification [1]. Some methods in the time domain are MUSIC [3], PCA and LDA [4], quaternions [5], statistics [6], and machine learning based [7,8]. These methods provide low computational cost and quickness in processing.…”