Abstract: Aiming at the problems of abnormal operation condition monitoring for environmental protection in polluting enterprises at present, such as difficult implementation, large identification errors and easy tampering with the results, this paper proposes an identification method of abnormal operation conditions for environmental protection based on power quality monitoring data. The multi-dimensional power quality data obtained from non-intrustive load monitoring at the public power entrance of enterprise equipment are used to train the model of condition classification, to realize abnormal condition identification, which is different from the existing scheme of power consumption monitoring with separate meters installed for each device. First, the time series change-point detection and the clustering calculation for the characteristic data representing the production conditions are carried out to realize the division of production operation conditions of enterprises. Then, combined with the operation of environmental protection equipment, the categories of environmental protection operation conditions for classification are obtained. Furthermore,the operation condition scenarios related to environmental protection are classified and learned by the Stacking learning model.Finally, the trained classification model is used to identify the abnormal operation conditions for environmental protection in the enterprise. The effectiveness of the proposed method is verified by the simulation test data and the actual enterprise data.