A Generic Severity Estimation Method for Lightning-Caused Voltage Sags by Using Multiplatform Monitoring Data Mining

Estimating the severity of lightning-caused voltage sag events lays the foundation of voltage sag mitigation by differentiated lightning protection. Traditional methods require numerous parameters for modeling power grid and lightning-caused voltage sags. In real-world scenarios, obtaining these parameters is challenging due to security issues or management problems. Therefore, a generic severity estimation method for lightning-caused voltage sags by using multiplatform monitoring data mining is proposed in this paper. The generic estimation method can be divided into three main modules: data discretization, association rule mining, and rule matching. Association rules are mined to discover hidden patterns between available lightning-related parameters and the corresponding sag severity based on historical records in multiplatform. For unmonitored sites, the severity of voltage sag caused by newly recorded lightning can be estimated by matching lightning-parameters with mined rules. Field records from multiplatform are applied to verify the performance and effectiveness of the proposed method. The proposed method yields acceptable performance as long as a few essential monitoring parameters are available. All the parameters are monitored in two widely used monitoring platforms, i.e., Lightning Location System (LLS) and Power Quality Monitoring System (PQMS), indicating the proposed method is generic. Moreover, the supplementation of more parameters from other platforms can further improve the estimated accuracy.