Pattern recognition with cerebellar model articulation controller and fractal features on partial discharges

This paper presents a new partial discharge (PD) pattern recognition method based on the cerebellar model articulation controller (CMAC). CMAC is an adaptive system by which defect types for partial discharge can be identified by referring to a table rather than by mathematical solution of simultaneous equations. CMAC maps input features of partial discharge into an input vector which is used to address a memory where the appropriate defect types are stored. Five types of defect models are well-designed on the base of investigation of many power apparatus failures. A PD detector is used to measure the raw three-dimension (3D) PD patterns, from which the fractal dimension, the lacunarity, and the mean discharges of phase windows are extracted as PD features. These critical features form the cluster domains of defect types. Using the characteristics of self-learning, association, and generalization, like the cerebellum of human being, the proposed CMAC-based pattern recognition scheme enables a powerful, straightforward, and efficient pattern recognition method. Moreover, the CMAC has the advantages of higher accuracy, shorter learning times, and noise tolerance, which are useful in recognizing the PD patterns of electrical apparatus. To demonstrate the effectiveness of the proposed method, comparative studies using a multilayer neural network (MNN) and K-means method are conducted on 200 sets of field-test PD patterns with high accuracy and high tolerance in noise interference.