Power transmission line fault classification method

Abstract

The invention discloses a power transmission line fault classification method, which comprises the following steps of: A, extracting current fault sample data, classifying current fault samples, and establishing a current sample matrix according to current fault types; B, setting a decision value, and establishing a mathematical model of deep dictionary learning according to the sample matrix; C, training the mathematical model; D, judging whether the mathematical model is trained or not according to the judgment value, if the judgment value condition is not met, repeating the step C, and if the judgment value condition is met, skipping to the step E; and E, establishing and outputting the mathematical model. The power transmission line fault classification method has three technical advantages that: features can be automatically extracted; the physical meaning is clear; and the model performance is good.

Description

technical field [0001] The invention relates to the technical field of relay protection, and more specifically relates to a fault classification method for transmission lines. Background technique [0002] The operating environment of transmission lines in power systems is complex, and they are susceptible to failures caused by various human or natural factors. Quickly and accurately identifying the fault type after a line fault is of great significance for improving the accuracy of fault location, shortening the recovery time of the fault line, and improving the stability of the power system. [0003] At present, the main transmission line fault classification methods can be roughly divided into classification methods based on physical models and classification methods based on artificial intelligence. The method based on the physical model is to use the difference of different types of faults in the predefined characteristic indicators to realize the judgment of the fault...

AI Technical Summary

Problems solved by technology

However, the methods based on physical models also have disadvantages such as being easily affected by fault conditions or changes in system parameters, and some classification methods require high hardware requirements; while the design process of fault classification methods based on artificial intelligence is relatively simple and does not require complex modeling, but In terms of fault feature extraction, etc., human participation is still required, which is time-consuming and laborious

Moreover, the features extracted by the existing deep learning classification methods lack interpretability, and the physical meaning is not clear.

Images