Research On Micro-grid Fault Diagnosis Based On Support Vector Machine

The consumption of traditional fossil energy sources and the serious environmental pollution problems that make it possible to develop renewable energy and optimize the energy structure are particularly important.The microgrid has become a hotspot in the research of renewable energy technologies due to its flexible operation mode,high energy efficiency,and environmental friendliness.However,due to the changeable micro-grid topology and distributed energy access,the fault diagnosis problem is extremely complicated.If the fault is not handled properly,the fault range may be expanded,which may affect the reliability of the power supply system and bring significant economics.Loss,it is of great significance to study a suitable microgrid fault diagnosis method.Support vector machines have the advantages of strong generalization and good classification in solving small sample,nonlinear and high-dimensional pattern recognition problems.In recent years,with the development of intelligent algorithms and data analysis technologies,support vector machines have broad application prospects in the field of fault diagnosis.Based on this,this paper mainly studies the micro-grid fault diagnosis algorithm based on support vector machine,and conducts in-depth research on micro-grid fault diagnosis from the aspects of fault sample processing and fault model optimization.The specific work is as follows:(1)Analyzed the working principle and mathematical model of micro-sources such as photovoltaic,wind turbines,micro gas turbines,etc.,built a corresponding simulation model on the PSCAD platform,constructed a low-voltage AC micro-grid model,and realized the stable operation of the grid-connected island.Simulate different types of line faults in the microgrid.(2)The extraction and optimization of fault characteristics of microgrid are studied.Aiming at the problem that single fault quantity is insufficient as diagnostic information of fault samples,this paper combines wavelet energy entropy with improved principal component analysis method,first extracts fault features from multiple fault quantities through wavelet energy entropy,and then uses improved principal component analysis method to The original fault features were selected to achieve the purpose of removing sample noise,increasing fault information and reducing the complexity of the system,and compared with single fault sample and multi-fault sample without dimension reduction.(3)A micro-grid fault diagnosis model based on support vector machine is established.Aiming at the phenomenon that the number of repeated training samples and the unevendistribution of samples on the SVM multi-classification problem,this paper proposes a support vector machine multi-classification algorithm with fault priority.This method combines the composition of microgrid fault samples,the decision tree idea and the support vector machine.It has the characteristics of less repeated training samples and simple and intuitive diagnosis process.(4)For the problem of low accuracy of the microgrid fault diagnosis model,this paper proposes to use a multiple population genetic algorithm to optimize the diagnosis model,establish a support vector machine fault diagnosis model based on the multi-group genetic algorithm,and conduct experimental analysis.Experimental results show that the method effectively improves the training speed and accuracy of the microgrid fault diagnosis model,has a higher generalization ability,and realizes the intelligent diagnosis of microgrid faults.