Research On State Discrimination And Intelligent Decision Technology Of Power Equipment Based On Machine Learning

With the continuous development of economy and society,the demand for power supply in China is rising rapidly,which puts forward high requirements for the intelligent level of power grid.As an important part of the power transmission and transformation system,the status identification of the core equipment and the timely discovery,decision-making and processing of fault events are the key problems to be solved to ensure the normal and stable operation of the power system.Substation equipment running status is complex,fault type and causes of diversity,equipment failure,usually spread to many related devices and equipment,but as a result of monitoring and control system can capture the fault information is not complete,the weak correlation between failure data,only through the operations staff to find the problem and remedy has been unable to meet the needs of the current smart grid.In this paper,in-depth research has been carried out on the problems of complex and diverse fault features between data sources and equipment and poor correlation between fault data in the fault diagnosis of smart grid equipment.The main research contents and innovation points are as follows:(1)Aiming at the problems of complex data source and state characteristics of substation equipment and high data dimension,a fault diagnosis model based on optimized parameter SVM was proposed.The model firstly analyzes the main influencing factors through PCA,extracts the key fault features,and realizes the preprocessing of equipment fault information and data dimensionality reduction.Then,combining with the operation mode of substation,a fault classification method based on hierarchical binary tree is proposed,and the parameters are optimized by empire competition algorithm.Finally,the optimized parameters are used to construct the MCHBT classifier.The experimental results show that this model can effectively realize the fault identification of smart substation equipment with small samples and multicategories,and significantly reduce the complexity of calculation.(2)Aiming at the problems of incomplete data acquisition and poor correlation of fault features in complex environment of substation,a fault processing model based on improved CBR is proposed.Firstly,the fault case database is established according to the classification and storage mode of the fault types.Then,case search was carried out by using Bayesian probability and possibility theory,and different weights were assigned to different features through correlation analysis,and the assigned fault cases were matched.Finally,the case base is updated according to the principle of updating historical cases and adding new cases.The case analysis results show that the model can effectively solve the problem of poor correlation of substation fault features and improve the usability of fault diagnosis.Meanwhile,the self-learning function of the case base also provides a powerful reference for the follow-up work.