A prediction method of relay protection defects

An improved logistic regression model applied to the prediction of relay protection defects is proposed. On the basis of the existing stepwise logistic regression model, by adding additional conditions restricting the training of the logistic regression model to fall into over fitting, the particle swarm optimization algorithm is proposed to improve the existing stepwise regression gradient descent model parameter fitting method, which solves the problem that the multicollinearity characteristics between influencing factors can not be found by the existing indicators caused by the randomness of relay protection defects and insufficient data accuracy, and obtains the regression parameters that follow the reasonable influence direction of independent variables on dependent variables. So as to reasonably predict whether defects occur according to equipment information and provide support for equipment operation and maintenance.