Due to the high penetration, diversification and decentralization of distributed generation (DG) and power electronic loads, the distribution network is faced the challenge of power quality control. With the application of the monitoring system and communication system in distribution network, the power quality time series data from the above systems is available for distribution network control. In this paper, a regional power quality control strategy for power quality based on data-driven was proposed. The piecewise linear representation method is used to identify the crucial trend turning points of the power quality time series. A feature matrix is constructed with both numerical and temporal features to comprehensively capture the phased evolution patterns of power quality. The grey correlation analysis is applied to identify the degree between node sequences which is used to classify nodes region for power quality control. This control strategy can improve the power quality and save the power quality control device. Simulation was constructed with the IEEE 14-bus distribution system model to validate the rationality and effectiveness of the proposed control strategy.