Induction motors are one of the most widely used electrical machines. Statistics of bearing failures of induction motors indicate, that they constitute more than 40% of induction motor damage. Therefore, bearing diagnosis is so important for trouble-free work of induction motors. The most common methods of bearing diagnosis are based on vibration signal analysis. The main disadvantage of those methods is the need for physical access to the diagnosed machine, which is not always possible. Methods based on motor current signature analysis are free of this disadvantage. Preliminary studies have shown that motor current signature analysis based normalized triple covariance is a very good diagnostic indicator for induction motor bearings. This paper presents an attempt to find a more accurate diagnostic indicator based on normalized triple covariance. In this paper the author verifies how many diagnostic features (normalized triple covariances) included in diagnostic indicator can give better separation between healthy and unhealthy cases.

It is proposed, developed, investigated, and validated by experiments and modelling for the first time in worldwide terms new data processing technologies, higher order spectral multiple correlation technologies for fault identification for electromechanical systems via electrical data processing. Investigation of the higher order spectral triple correlation technology via modelling has shown that the proposed data processing technology effectively detects component faults. The higher order spectral triple correlation technology successfully applied for rolling bearing fault identification. Experimental investigation of the technology has shown, that the technology effectively identifies rolling bearing fault by electrical data processing at very early stage of fault development. Novel technology comparisons via modelling and experiments of the proposed higher order spectral triple correlation technology and the higher order spectra technology show the higher fault identification effectiveness of the proposed technology over the bicoherence technology.

In the last decade, research centered around the fault diagnosis of rotating machinery using non-contact techniques has been significantly on the rise. For the first time worldwide, innovative techniques for the diagnosis of rotating machinery, based on electrical motors, including generic, nonlinear, higher-order cross-correlations of spectral moduli of the third and fourth order (CCSM3 and CCSM4, respectively), have been comprehensively validated by modeling and experiments. The existing higher-order cross-correlations of complex spectra are not sufficiently effective for the fault diagnosis of rotating machinery. The novel technology CCSM3 was comprehensively experimentally validated for induction motor bearing diagnosis via motor current signals. Experimental results, provided by the validated technology, confirmed high overall probabilities of correct diagnosis for bearings at early stages of damage development. The novel diagnosis technologies were compared with existing diagnosis technologies, based on triple and fourth cross-correlations of the complex spectra. The comprehensive validation and comparison of the novel cross-correlation technologies confirmed an important non-traditional novel outcome: the technologies based on cross-correlations of spectral moduli were more effective for damage diagnosis than the technologies based on cross-correlations of the complex spectra. Experimental and simulation validations confirmed a high probability of correct diagnosis via the CCSM at the early stage of fault development. The average total probability of incorrect diagnosis for the CCSM3 for all experimental results of 8 tested bearings, estimated via 6528 diagnostic features, was 1.475%. The effectiveness gains in the total probability of incorrect diagnosis for the CCSM3 in comparison with the CCCS3 were 26.8 for the experimental validation and 18.9 for the simulation validation. The effectiveness gains in the Fisher criterion for the CCSM3 in comparison with the CCCS3 were 50.7 for the simulation validation and 104.7 for the experimental validation.

Diagnosis of induction motors, conducted remotely by measuring and analyzing the supply current is attractive with the lack of access to the engine. So far there is no solution, based on analysis of current, the credibility of which allow use in industry. Statistics of IM bearing failures of induction motors indicate, that they constitute more than 40% of IM damage, therefore bearing diagnosis is so important. The article provides an overview of selected methods of diagnosis of induction motor bearings, based on measurement of the supply current. The problem here is the high disturbance components level of the motor current in relation to diagnostic components. The paper presents the new approach to signal analysis solutions, based on statistical methods, which have been adapted to be used by this diagnostic system. First experimental results with use of this method are also presented, they confirm the advantages of this method.

For the first time worldwide, innovative techniques, generic non-linear higher-order unnormalized cross-correlations of spectral moduli, for the diagnosis of complex assets, are proposed. The normalization of the proposed techniques is based on the absolute central moments, that have been proposed and widely investigated in mathematical works. The existing higher-order, cross-covariances of complex spectral components are not sufficiently effective. The novel technology is comprehensively experimentally validated for induction motor bearing diagnosis via motor current signals. Experimental results, provided by the proposed technique, confirmed high overall probabilities of correct diagnoses for bearings at early stages of damage development. The proposed diagnosis technology is compared with existing diagnosis technology, based on the triple cross-covariance of complex spectral components.