Lightning arrester leakage current prediction method based on multivariable reconstruction combined dynamic weight

• Main Authors: LIU YIFAN, CAO YOUJIN, PANG WEISHENG, MA MINGZHONG, XU BAOHONG, DAI ZHENSHAN, LI GANG, ZHU JINWEI, YANG ZHENYU, CHEN PINGPING, SUN YONGKE, WEI ZHONG, CHEN KAI, LIU QIANGMING, LI DONGSHENG, PANG LEI, DING HAIBO, ZENG BO, HUANG TENG, LUO YI, ZHU CHAOPING, ZHANG YANJUN, ZHANG YUANYUE, ZHENG GAOJIE, MA YIXIN, LI ZHENXING, CHANG KUAN, LI LIANG
• Format: Patent
• Language: Chinese; English
• Published: 16.01.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a lightning arrester leakage current prediction method based on multivariable reconstruction combined with dynamic weight. The method comprises the following steps: extracting current data of a historical zinc oxide lightning arrester and corresponding external environment data; preprocessing the multi-channel data; constructing a multivariable spatial data structure of leakage current and external factors; a neural network leakage current prediction model based on multivariable phase space reconstruction and dynamic weight optimization is established by combining the strong nonlinear fitting capability of a nonlinear autoregressive neural network (NARX) and a radial basis function (RBF) neural network. According to the method, the accuracy of the prediction model is optimized while the nonlinear relationship between external factors and historical data is comprehensively considered, so that the maintenance of the zinc oxide arrester (MOA) has more reliable real-time performance and su