Classification method for multiple power quality disturbances via label distribution enhancement and multi‐granular feature optimization

Large-scale integration of distributed generation and widespread use power electronic equipment make quality disturbances (PQDs) more complicated. There are still unsolved problems for the classification multiple (MPQDs) that consist various kinds single disturbances: (1) since difference between contribution degrees every contained disturbance to composite is ignored, logical labels cannot completely describe disturbance; (2) optimal feature in one granular space may not be another. These drawbacks lead degradation MPQD classification, which should considered a multi-label model rather than single-label used existing methods. Therefore, this paper proposes novel method improve performance MPQDs. The label distribution, representing degree disturbance, introduced. In addition, high-dimensional reduced by multigranular optimization, where fuzziness redundancy removed modified rough-set method. To classifier, ensemble based on homogeneous classifier proposed integrate base classifiers constructed vectors from different granularity spaces. A large number field recordings applied validate results show performs better traditional methods, especially under noisy environments.