Several Applications Of Neural Network In Photovoltaic Monitoring

The energy problem has been widely concerned,the new energy technology represented by the photovoltaic energy generation has been taken seriously.Photovoltaic monitoring is one of the important ways to make good use of solar energy.However,the photovoltaic monitoring systems have a complex nonlinear problems,lead the artificial intelligence technology to cope with the problems of the monitoring is the main motivation of this paper.In this paper,the neural network method is applied to the identification of the optical angle sensor model and the fault diagnosis.The main contents are arranged as follows:Part 1: This part mainly introduces the fundamental theory of neural network.Firstly,the principle of artificial neurons which invented from the biological neurons are introduced,also the working characteristics are introduced as well.Then,based on the level of information provided by the environment,the three learning methods of the neural network are summarized and described.Finally,the paper summarizes the three commonly used algorithm principles and the typical network structure used in the neural network learning process.Part 2: The BP neural network model is selected as the main research object of this paper.Firstly,the principle of the BP neural network model is briefly summarized.The feedforward calculation and the error back propagation in the model are deduced respectively,and the final error iterative formula is obtained.Then,the study on the structure design of the network model and the selection of the input and output samples,also the processing of the input and output samples are discussed respectively and deeply.In the end,the selection of parameters in network training is mainly studied and analyzed.Part 3: The BP neural network system identification model in proposed to solve the problem that the special optical angel sensor used in PV monitoring system which is difficult to identify.Firstly,a BP neural network model based on L-M algorithm for model identification is established for the nonlinear problem that cannot accurately measure the angle due to the manufacturing or installation error existing in the optical angle sensor.Then,the data acquisition system based on optical angle sensor is established and measure a set of experiment data which including the input and output sample data of training neural network.Finally,the neural network trained which meets the target can effectively predict the output angle value based on the current data measured by the optical angle sensor.This method can be effectively applied to the modeling of the mathematical problems where the parameters are unknown or even difficult to build.Part 4: The BP neural network fault diagnosis model based on self-adaptive learning rate is proposed to solve the complex nonlinear problems which existing between PV modules and operating parameters.Firstly,the mathematical model is used to analyze the complex nonlinear problems of operating parameters and environmental conditions,also the BP neural network fault diagnosis model with adaptive learning rate is established.Then,in order to verify the effectiveness of the proposed model,using the fundamental equation of photovoltaic modules for experiment data acquisition,the input and output data for the training network is acquired.Finally,a new set of operation data from the PV modules is used to verify the effectiveness of the neural network.The results show that the neural network can effectively identify and classify the running states of the components PV modules,and the effectiveness of the proposed method is also verified.