Research On Hot Spot Fault Detection Method Of Photovoltaic Modules Based On Frequency Domain Characteristics

Photovoltaic modules play an integral role in photovoltaic power generation systems.However,due to the environmental problems,PV modules are subject to a variety of faults,among which hot spot faults have a greater impact on the power generation efficiency of PV modules and are more hazardous.To address the shortcomings of the traditional hot spot fault detection method,the impedance frequency domain characteristics of PV modules under the influence of hot spot faults are analysed in depth,based on which a hot spot fault detection method based on the frequency domain characteristics of PV modules combined with deep learning algorithms is proposed.Firstly,the working principle of the PV cell as well as the forward,reverse and dynamic characteristics are analysed.Based on the dynamic characteristics of the PV module,the impedance frequency domain characteristics of the PV module are discussed,the electrical characteristics parameters corresponding to each frequency band of the impedance frequency domain characteristics are given,and the influence of different electrical parameters on the impedance characteristics of the PV module is investigated.Secondly,an experimental test platform is built to analyse the variation of electrical characteristics parameters in the frequency domain characteristics of PV modules under different scenarios of bias voltage and hot spot fault area size,and the equivalent output impedance is simulated and verified based on the AC small signal model of the frequency domain characteristics of PV modules to verify the accuracy and validity of the impedance study method of the frequency domain characteristics of PV modules proposed in this paper.Finally,a convolutional neural network and bi-directional gated cyclic unit-based hot spot fault detection method for PV modules is proposed.By analysing the variation of electrical frequency domain characteristic parameters for different hot spot fault sizes of PV modules,a database including information on hot spot faults of different sizes is constructed.The database is used to perform diagnostic tests of the method adopted in this paper,and it is verified that the research method in this paper can effectively complete the diagnosis of hot spot faults of different areas of PV modules,and the accuracy rate is significantly improved when compared with different diagnostic methods.The findings of this paper provide an idea of combining frequency domain characteristics with intelligent algorithms for the detection of hot spot faults in PV modules,which is a reference for other detection methods to improve the fault diagnosis rate by combining with intelligent algorithms.