Fault Diagnosis And Early Warning Of Low-voltage Switchgear Based On Multivariate Data Analysis

The safe and reliable operation of power supply and distribution is an important guarantee for the sustainable development of national economy and the vital interests of users.As an important protection and control equipment in the power system,the switchgear has complex components.The aging of the insulation inside the switchgear in long-term operation,the corona discharge caused by excessive humidity,and the temperature rise of the bus bar crimp and bus bar may cause power failure or fire accidents,which may cause serious economic losses or even a great threat to personal safety.In this paper,we develop a low-voltage switchgear fault diagnosis and early warning system that can integrate multiple data for analysis by considering various electrical and non-electrical quantities that may cause low-voltage switchgear faults.First,the structure of low-voltage switchgear and its common faults are analyzed,and the necessary monitoring devices for low-voltage switchgear are determined accordingly,the selection of temperature sensors is completed,and the installation methods of each type of sensor are determined.The monitoring system is composed of four parts:terminal object layer,data communication layer,middle station layer and management layer.Then,the SVM-SMOTE oversampling algorithm is proposed,which can solve the above problems and make the sample data set more suitable for the subsequent artificial intelligence algorithm,and improve the accuracy of fault diagnosis and early warning as well as the accuracy of the system.The algorithm can solve the above problems and make the sample data set more suitable for subsequent artificial intelligence algorithms,improving the accuracy of fault diagnosis and early warning as well as the efficiency of the algorithm.It is verified that the algorithm can be adapted to any number of parameters and types of parameters in low-voltage switchgear operation.Finally,This Dissertation studies three different fault diagnosis and early warning algorithms,based on which an AdaBoost-RBF algorithm based low-voltage switchgear fault diagnosis and early warning method is proposed,and the optimal training parameters are selected,which can be adapted to any number and type of parameters in the operation of low-voltage switchgear.The inner layer of the algorithm uses RBF neural network to learn multivariate data and realize fault diagnosis and early warning based on multivariate data,and the outer layer uses AdaBoost algorithm to improve the fitting degree of the algorithm and reduce the misjudgment rate.Then,on the basis of the algorithm,the implementation of the dynamic criterion is completed,so that the judgment threshold of the algorithm is dynamically adjusted with the change of the operation mode of the power system,and the early warning accuracy of the algorithm model is improved.