Accurate estimation of voltage sag severity can provide a basis for sensitive users to adjust their production mode and avoid economic losses in advance. In view of lightning

which is the main cause of voltage sag

the widely deployed lightning location system has the function of predicting lightning activity parameters. So it is possible to estimate the severity of voltage sag at monitoring points caused by lightning strike. Based on the above background

a lightning-voltage sag severity estimation method based on adaptive association rule mining is proposed in this paper. The lightning-voltage sag sample set is constructed based on the monitoring data. An attribute discretization method based on Dunn index is proposed

and the best discretization results of attributes are obtained. The key condition attributes that affect voltage sag severity are screened by attribute reduction algorithm. An association rule mining algorithm based on parameter adaption is proposed

which overcomes the problem that the results of traditional association rules mining methods are affected by non-uniform data. Based on the weighted Euclidean distance

the matching model between the prediction scene and the historical rules is established to estimate voltage sag severity. The practicality and effectiveness of the method are verified based on the actual measurement data of a regional power grid.