Application Of Deep Belief Networks In Fault Diagnosis Of Power Generation Equipment

In order to meet the needs of people,our country are invest a lot of money on the construction of electric power industry.Electricity as energy security for the steady development of national economy.Stable attaches great importance to the safety of the power plant in our country.Use the technology of artificial intelligence in fault diagnosis of power equipment,to ensure the security and stability of electric power industry production,reduce the fault.This article by comparing the effect of several kinds of algorithm,finally selected with that deep belief networks solve the fault diagnosis problem,the selected support vector machine regression algorithm to solve the fault prediction problem.Due to the fault diagnosis problem belongs to the unbalanced classification problem,need for data preprocessing before constructing classifier.In this paper,the normal data according to the time interval sampling and random sampling method to owe sampling,the fault data we use SMOTE algorithm expanded samples.Select three typical classification algorithm to model,comparison model effect,determine suitable algorithm.The fault prediction as part of the fault diagnosis technology,can discover the hidden fault equipment.This article selected using correlation analysis and support vector machine regression method to construct fault prediction model.By comparing regression model of locally weighted linear,the BP neural network regression model and the model of support vector machine in the actual data of a power equipment,the results show that the model of support vector machine can better predict the equipment in the future time,more conducive to the fault prediction.