Application of Small Animals Control in Substation Based on GBDT and LR Fusion Algorithm

Animals control is an important task for the safe operation of substations. Aiming at the problems existing in the control of small animals, with the superior prediction ability of machine learning, a prediction model of small animals hazard grade is proposed, which combines gradient boosting decision (GBDT) and logistic regression (LR) algorithm. The model combined substation operation and maintenance data with local meteorological data, performs features screening by calculating the variance value, and achieves classes balance by using sampling technology. And finally the model achieves the prediction of small animals hazard grade in substation. By using different data sets and not using GBDT algorithm to train the model, the prediction results are compared and analyzed. The proposed model is better in all prediction performance indicators, which verifies the validity of the method.