The invention relates to a power quality disturbance source working condition identifying method comprising the following steps: (1) acquiring power data of a power quality disturbance source; (2) acquiring a power data format which can be used for clustering; (3) analyzing the acquired power data; and (4) classifying the acquired power data using a clustering analysis method. The method provided by the invention overcomes the defects of the traditional power quality analysis method such as low efficiency, high cost and unsuitability for large-scale data set processing, is practical and efficient in the aspect of power quality disturbance source working condition identification, and is of low cost and good performance.