Research On Overall Linearized Modeling Of Photovoltaic System And Associated Application In Power Prediction

With the increasingly severe energy and environmental crisis,renewable energy sources are developing rapidly around the world.Among them,photovoltaic power generation has rapidly become one of the hot spots of new energy research with its advantages of low cost and low pollution.The intermittent and random nature of photovoltaic power generation poses a great challenge to the operation and adjustment of the grid system.In order to ensure that PV power plants are safely and stably connected to the grid,giving full play to the advantages of new energy sources in terms of efficiency and environmental protection,while overcoming their intermittent and fluctuating disadvantages,it is necessary to make reasonable predictions of the output power of PV power generation.This can reduce the operational risk of the grid and improve the operational efficiency.At present,scholars at home and abroad have proposed many power generation prediction methods,which are effective in realizing the power generation management of PV systems.On the one hand,almost all methods use power prediction methods based on historical data,and this method must provide a large amount of historical data for model training.On the other hand,because the process from the change of output parameters caused by external weather changes to the controller’s response generates a large delay,this power prediction method has poor weather adaptive capability,slow prediction speed,and low real-time prediction accuracy.In order to solve these problems,this Dissertation conducts an innovative research on the active prediction method based on delay model variable weather parameters(DTM-VWP)and proposes a DTM-VWP active power prediction algorithm.In addition,in order to obtain better speed and accuracy of power prediction algorithm at the same time,this Dissertation also conducts an in-depth study on the complementary application of DTM-VWP method and particle swarm optimization neural network(PSO-BP)power prediction method.The main contents are as follows.(1)An overall linearized model of variable weather parameter(VWP)is established by analyzing the PV cell model and PV system structure.Based on this,a SOLM-VWP and its DTM-VWP are proposed,which can avoid the use of linearization theory involving the analysis of complex higher-order system delay characteristics,thus reducing the hardware cost,shortening the computational cycle and speeding up the acquisition of transient performance to carry out real-time power prediction work.(2)A DTM-VWP power prediction method based on OLM-VWP is proposed by the mathematical relationship between PV cell model and circuit theorem,thus maximizing the speed of the power prediction method.Among them,finding the direct relationship between the delay model of PV system and the output power is the core of these prediction methods and the main theoretical basis of these prediction methods.(3)Based on the DTM-VWP power prediction method,the DTM-VWP power prediction algorithm is set up in the design of the BP neural network algorithm controller to determine the relationship between the two weather parameters,irradiance and temperature,on the PV output power,so as to determine the input variables of the PSO-BP prediction model.Based on this,this Dissertation proposes a prediction algorithm based on the DTM-VWP method to optimize PSO-BP,which can fully utilize the advantages of both DTM-VWP method and PSO-BP neural network prediction method and overcome the shortcomings of the traditional PSO-BP neural network.Finally,MATLAB simulations and a hardware circuit experimental platform are built to verify the theoretical correctness of the proposed OLM-VWP,SOLM-VWP,DTM-VWP and DTM-VWP power prediction methods.The experimental results demonstrate the feasibility of the overall linearized modeling idea,the effectiveness of the DTM-VWP prediction method and the DTM-VWP method to optimize the PSOBP neural network prediction method.It is also confirmed by comparison experiments that the DTM-VWP method can improve the rapidity of PV system power prediction,and the combination of DTM-VWP method and PSO-BP prediction method can combine the rapidity and accuracy performance of the prediction algorithm at the same time.This power prediction method provides effective theoretical support for the optimization of PV system scheduling,efficient energy utilization,and safe and stable grid operation.