This paper presents an efficient approach for condition based maintenance (CBM) of three phase synchronous generators for diagnosing inter-turn faults using current signatures. Support vector machines (SVM) are one of the widely used algorithms for the purpose of decision making in CBM. To improve the performance of the classifier, either we can select the kernel according to the features or select the features according to the kernel or linearize the features into a higher dimensional space and use linear SVM. In this work, we experiment with the third approach for improving the performance of the system. Sparse coding is an effective feature mapping technique that can be used to linearize the features into a higher dimensional space. Sparse coding improves the performance from 58.19% to 91.14% for R phase fault and from 67.64% to 91.50% forY phase fault and 73.02% to 94.79% for B phase faults respectively. © Research India Publications
cited By 0
B. K. Vaisakh, Gopinath, R., Dr. Santhosh Kumar C., and Ganesan, M., “Condition monitoring of synchronous generators using sparse coding”, International Journal of Applied Engineering Research, vol. 10, pp. 26689-26697, 2015.