The invention discloses a distribution network electrical topology identification method and system based on an edge calculation improved KNN, and relates to the technical field of topology identification in the field of distribution networks, and the method comprises the steps: S1, obtaining user voltage data, and recording the user voltage data as a data set A; S2, dividing the data set A into atraining set and a test set, performing mean value filling processing, and recording the data set A as a training set B and a test set C; s3, improving a KNN classification algorithm, taking a samplein the test set C to calculate with the training set B, and determining a phase to which the sample in the test set C belongs; S4, adding tested samples in the test set C into the training set B forupdating, and recording the samples as a training set D; s5, repeating the step S3, calculating the next sample in the test set C and the training set D, and determining the phase to which the next sample in the test set C belongs; and S6, repeating the training set updating in the step S4 and the classification calculation in the step S3 to obtain all phases to which the test set C belongs, thereby overcoming the problems of low topology identification efficiency and overlarge data processing pressure of the acquisition master station in the prior art.