Study On Maximum Power Point Tracking And Sensor Fault Diagnosis Of Photovoltaic Systems

Solar energy,as a kind of renewable energy,can reduce the carbon dioxide emission from industrial production and conducive to less environmental pollution.An effective way to use solar energy is to generate electricity using the photovoltaic properties,Photovoltaic power plants will inevitably suffer from non-Gaussian random disturbance during the production process,the sensor is of critical significance for the stable running of the system.The main work of this paper includes three parts:the maximum power point tracking of photovoltaic systems,the prediction of solar irradiance and the sensor fault diagnosis of photovoltaic systems.In the first part,the equivalent model of photovoltaic cells and its output characteristic is studied,in addition,the incremental conductance method based on piecewise idea is used to track the maximum power point,the simulation results show that the performance of this method is greatly improved compared with the results obtained by traditional methods.In the second part,the long-short term memory neural network optimized by attention model is proposed to realize the function of predicting solar irradiance.The prediction performance of this neural network can achieve high accuracy compared with some traditional prediction methods.In the last part,a filter based on quantized minimum error entropy is designed for the distributed photovoltaic power plants suffering from non-Gaussian disturbances,the filter can accurately estimate the state of the system by the optimal gain matrix,in addition,a fault diagnose method,in which the states obtained by the mentioned filter inputted into a hierarchical extreme learning machine,is proposed,the fault diagnosis simulations of the sensor shows that the method has good performance.