Line loss anomaly identification and correction method and system considering filling sequence

The invention discloses a line loss anomaly identification and correction method considering a filling sequence. The method comprises the following steps: performing high-dimensional mapping through a neural tangent kernel function to obtain a kernel matrix; selecting an initial clustering center by using a clustering center probability formula and a roulette method, and iteratively updating the clustering center to obtain a final clustering result; calculating the minimum value of the sum of the intra-cluster minimum distance and the reciprocal of the inter-cluster maximum distance to evaluate a clustering result; the similarity between data sources is obtained through a multi-dimensional similarity calculation method, and data filling is carried out through a Bayesian network problem solving method so as to correct line loss abnormal data. According to the method, accurate identification of the line loss abnormal data is realized through the neural tangent kernel K-Means clustering; a line loss abnormal dat