Substation relay protection device service life prediction method

Abstract

The invention relates to a method for predicting the service life of a transformer substation relay protection device. The method comprises the following steps: S1, selecting evaluation indexes, and obtaining the sample data of the corresponding evaluation indexes; s2, quantifying the sample data; s3, establishing a multi-classification support vector machine model; s4, training and optimizing themulti-class support vector machine model by utilizing the quantized sample data; and S5, obtaining evaluation indexes of the relay protection device, inputting the evaluation indexes into the multi-classification support vector machine model, and outputting the predicted life length of the relay protection device by the multi-classification support vector machine model.

Description

technical field [0001] The invention relates to the field of relay protection, in particular to a method for predicting the service life of a substation relay protection device. Background technique [0002] As the first line of defense for power grid security, the abnormal status of relay protection devices has always been one of the most important factors that endanger the safety of the entire power grid. Once the relay protection device fails, the power grid will lose its protection, which will easily cause large-scale power outages and cause huge economic losses. Nowadays, the regular inspection scheme adopted by power grid companies is generally relatively intensive, resulting in a large manpower burden. Therefore, timely prediction and evaluation of the service life of relay protection devices will help power grid companies formulate more efficient regular inspection plans. [0003] Under the condition of long-term operation of the relay protection device, there are ...

AI Technical Summary

Problems solved by technology

But in addition, they all have some problems: the correct rate of Bayesian network classifier is low when the key attributes are missing; when the evidence reasoning method transmits attribute information layer by layer, there will be a certain degree of loss of uncertain information; The target theory method is difficult to obtain high accuracy for small sample data; the artificial neural network method will not be able to perform neural networks when the number of samples is small.

Images