Study On Dynamic Moldel Of Multidimcnsional Stochastic Process Lor Solar Power Generation

Solar energy is an inexhaustible renewable energy for mankind.It has many superiorities on full cleanliness,absolute safety,relative universality,long term maintenance free,resource sufficiency and potential economy.The development of solar photovoltaic power generation is to solve the over exploitation of non-renewable energy,traditional fossil fuel burning effectively caused severe environmental pollution and other problems of today’s energy,is to promote the sustainable development of modern economic society,it is one of the important measures of creating energy intensive countries,and adhere to the implementation of the national new energy strategy.However,solar energy,as a dispersive and highly volatile intermittent energy,has many related factors in its power generation process,and photovoltaic power generation affects the rationality of power grid planning and the stability of system operation.Therefore,it is of great practical significance to study the stochastic process of solar power generation,build a dynamic model of photovoltaic generation process,and calculate the intensity of solar radiation and output of PV power station.This paper begins with a summary of the domestic and foreign research on the process of photovoltaic power generation random process,including the overall development of the dynamic stochastic process of photovoltaic power generation,model research dynamics of solar radiation value and photovoltaic output forecasting and application of improved neural network in photovoltaic power generation model.Secondly,it concreting on the analysis of solar photovoltaic power generation process:summarizing the characteristics of solar light resources and photovoltaic power generation systems,focusing on the internal and external factors of photovoltaic power generation process,and comparing of many improved neural network model advantages and disadvantages,select the advantage algorithm.Then,considering the characteristics of the randomness and intermittency of the solar light resources.A hidden Markov model of double random process is established to predict the intensity of solar radiation.Regarded the cloudiness as the hidden layer of the model,and the clear air obscuring coefficient is regarded as the model observation layer,and the relation function is studied.The double random variables are determined,and the solar radiation values are calculated.Subsequently,constructed a multi-dimensional wavelet neural network model one of the input of the solar radiation value.Wavelet transform is used to refine the input signal according to scale and replace the hidden layer function of neural network transmission process with wavelet function to expand and translate the transformation accordingly,making wavelet analysis and neural network vertically and horizontally combined,and complementary advantages,so as to predict PV output power.Finally,some management suggestions are put forward to improve the prediction effect of the PV output model from the aspects of data sampling,information management and equipment operation.This paper simulates the photovoltaic power generation process.Accord to historical data,predicting that solar radiation intensity shield by clouds in different degrees and the output power of photovoltaic power generation under the influence of various seasonal climate environments.The model has strong adaptability under different weather conditions and high accuracy of simulation prediction results,which can provide data support for load forecasting and reference for power grid dispatching and power planning.