Research On Energy Scheduling Of Household Photovoltaic Energy Storage Inverter System Considering Forecast Of Source And Load

At present,photovoltaic power generation has become an important means to solve the energy crisis.As a product of photovoltaic practical application,the utilization rate of household photovoltaic energy storage inverters has gradually increased.In the complete system,photovoltaic modules and energy storage batteries are connected to the grid and household loads through photovoltaic inverter,so the internal energy flow modes are diverse.In order to improve energy utilization and user economy,the energy dispatching management of the system must be carried out in advance.The accurate prediction of photovoltaic output and household load is an important basis for efficient energy scheduling.Therefore,the photovoltaic power prediction,household load prediction and energy scheduling problems are studied in this paper.Firstly,this paper introduces the main influencing factors of photovoltaic power generation,and proposes a similar day selection method,which based on distance similarity and change similarity,so as to provide more accurate input samples for the prediction model.And then this paper uses the genetic algorithm with strong global optimization ability to optimize BP neural network to reduce the prediction error.Secondly this paper proposes a prediction model based on improved cuckoo algorithm optimized LSTM neural network,and uses this model to predict the household load.By adjusting some parameters in the cuckoo algorithm,the optimization ability of the algorithm is improved,and the accuracy of load forecasting is also improved.Finally,this paper builds an optimization model for the inverter system,and proposes an adaptive penalty function method for the complex constraints of the model,which is combined with the particle swarm optimization algorithm to carry out energy scheduling planning with the minimum daily power expenditure as the optimization goal.The final optimization result proves that the proposed algorithm has strong optimization ability,and the experiments on RT-LAB platform verify the practical feasibility of the scheduling scheme.48 figures,10 tables,65 references...