AI techniques in induction machines diagnosis including the speed ripple effect

TLDR

Artificial intelligence methods such as expert systems, neural networks, and fuzzy logic have been shown to improve online diagnostics of induction machines, yet accurate fault models remain essential to quantify fault severity and balance simulation accuracy with simplicity. The study proposes a new, simple procedure for detecting rotor electrical faults in induction machines that incorporates the speed ripple effect. The procedure is based on a model that integrates the speed ripple effect to generate a diagnostic index. The resulting diagnostic index is independent of operating condition and inertia, enabling a diagnostic system with minimal configuration intelligence.

Abstract

Various applications of artificial intelligence (AI) techniques (expert systems, neural networks, and fuzzy logic) presented in the literature prove that such technologies are well suited to cope with on-line diagnostic tasks for induction machines. The features of these techniques and the improvements that they introduce in the diagnostic process are recalled, showing that, in order to obtain an indication on the fault extent, faulty machine models are still essential. Moreover, by the models, that must trade off between simulation result effectiveness and simplicity, it is possible to overcome crucial points of the diagnosis. With reference to rotor electrical faults of induction machines, a new and simple procedure based on a model which includes the speed ripple effect is developed. This procedure leads to a new diagnostic index, independent of the machine operating condition and inertia value, that allows the implementation of the diagnostic system with a minimum configuration intelligence.