Induction Motor Static Eccentricity Severity Estimation Using Evidence Theory

On-line condition monitoring of induction motors generally requires analysis of a range of signal features from multiple sensors to be able to accurately detect the presence of a fault and estimate its severity. Even so, variations in motor design or construction, operating conditions or other factors cause uncertainty in the relationship of the feature magnitudes to the presence and severity of a fault. This paper investigates a multisensor fusion algorithm based on evidence theory to estimate the severity of static eccentricity faults in a squirrel- cage induction motor. The paper reports a wide range of test results from a 2.2 kW 3-phase induction motor under varying degrees of eccentricity faults. In addition, the implementation details of the evidence theory based algorithm are given and the ability of the algorithm to accurately estimate the level of static eccentricity to within 12.5% is demonstrated.