With the large-scale construction of low-voltage distributed photovoltaic in China, the promotion of variable frequency agricultural equipment, the harmonic content of 10kV grade power grid has increased rapidly, and the risk of resonance occurrence is greatly increased too, which is resulting in frequent abnormal accidents of operating equipment. In order to accurately analyse the causes of abnormal accidents of 10 kV grade equipment, this paper proposes a resonance monitoring and early warning method based on power quality characteristic index value, and puts forward the concept of index correlation degree for the first time. By establishing a benchmark database of characteristic indicators, all the operating data of indicators under normality is recorded and analysed, and variation range of correlation degree indicators is locked too. When the operating amplitude of characteristic indicators exceeds the range of the benchmark database, the judgment of resonance phenomenon and risk warning function can be realized by combining the variation of correlation degree amplitude of each indicator. It helps the power supply companies to adjust the line or install the corresponding treatment device in time to avoid more serious resonance accidents.