Research On Fault Diagnosis Methods For Key Components Of Photovoltaic Generation System

With the influence of many factors,such as the growth of energy demand,the depletion and costs rise of fossil energy,the global warming,people are seeking for the development of renewable energy on a global scale.Solar energy has become an important part of renewable energy for the reason of its advantages,such as easy availability,no noise,cleanliness and endless.Chinese photovoltaic industry has developed rapidly,and the installed capacity of photovoltaics has been increasing year by year.With the rapid expansion of the installed capacity of photovoltaics,the demand for photovoltaic modules and related power equipment has also increased yearly.However,due to the lack of industry standards,the related products in the market are good and bad mixed together.Some products begin to fault after running for several years,even causing accidents.Therefore,it is of great practical significance to carry out fault diagnosis in key components of photovoltaic power generation systems.The main content of this article is as follows:1.The inverter is the core component of the photovoltaic power generation system.However,it is also the most prone to failure due to the influence of its self,operating environment and other factors.Before making the fault diagnosis of the inverter,the photovoltaic inverter should be selected first.This paper selects the three-level inverter as the research object,which is widely used in large capacity photovoltaic power generation applications,and establishes a simulation model.Photovoltaic arrays often fail due to its long-term outdoor work.So it is necessary to carry out fault diagnosis as well.At first,the mathematical model of the photovoltaic cell is introduced and its output characteristics are simulated.Then its working characteristics under shadow conditions are analyzed.Finally,the simulation model of the photovoltaic array is established,which sets the foundation of the following fault diagnosis.2.For the problem of open-circuit fault arising in diode-clamped three-level inverter,a new fault diagnosis method is proposed based on decision tree support vector machine.Firstly,the diagnosis strategy of the inverter is introduced and the fault is classified.Then,in terms of the multi-scale decomposition of wavelet analysis,the middle,upper and down bridge voltages are selected to extract the fault features,respectively.Moreover,particle swarm clustering algorithm is built to construct the DTSVM classify model,and the multi-model fault diagnosis of power component in three-level inverter is finally accomplished.By adding disturbances and comparing with other algorithms,it is verified that the algorithm can complete the fault diagnosis task with fewer classification models,and has high diagnosis accuracy and robustness.3.Both feature extraction and fault diagnosis algorithm are impoved.A fault diagnosis method based on empirical mode decomposition and decision tree relevance vector machine is proposed.In terms of feature extraction,aiming at the deficiency of wavelet decomposition algorithm,empirical mode decomposition for adaptive decomposition of signal is used.The feature vectors are in the form of energy and energy entropy.In the aspect of the diagnosis algorithm,a relevance vector machine,which does not need to set parameters,has more sparsely and shorter test time,is selected as the classification model.By comparing the two aspects of fault diagnosis and feature extraction,it is verified that this method has the advantages of less parameters,high efficiency,fast diagnosis speed and high accuracy.4.A photovoltaic array fault diagnosis method based on improved random forest is proposed for the purpose of another key component of the photovoltaic power generation system.This paper first introduces the diagnosis strategy of photovoltaic array and the fault was classified into two levels.Then a random forest algorithm is introduced.The weight of each decision tree is calculated by using the out-of-bag data during the training process.The voting and decision-making process of the decision tree is improved as well.At the same time,the importance of the variables is measured,and some secondary important variables are deleted.After retraining,the improved random forest fault diagnosis model is finally constructed.Through the fault diagnosis of 3×3 and 4×4 photovoltaic array,it shows that this method can achieve multi-mode fault diagnosis of the photovoltaic array,with high diagnostic accuracy and easy implementation.