Short-Term Load Forecasting And Photovoltaic Power Generation Forecasting Based On Fuzzy Membrane Clustering

Power load forecasting is an important module of energy management system.It aims to provide a safe and reliable automatic management platform for power system.The accuracy of short-term load forecasting has a great impact on the operation and management of power system,power production and purchase plan,reliability and security analysis and system maintenance plan.Therefore,improving the accuracy of short-term load forecasting can not only improve the efficiency of schedule management,but also significantly reduce operating costs.With the increasing popularity of solar energy,the fluctuation and uncertainty of solar power generation have brought difficulties to grid management.Therefore,in order to ensure the stability of large power grid,realize the optimal input of photovoltaic generating units and economic dispatch of power system,and reduce the uncertainty of photovoltaic power generation cost and power generation,accurate prediction of solar power generation is very necessary.Membrane computing,as a new natural computing method,has various structure of membrane system for processing different data.In this paper,the historic data of power load and photovoltaic power generation are pre-processed by using the computing structure and rules of tissue membrane system combined with fuzzy clustering.The specific work is as follows:Aiming at the shortcomings of the output randomness,stability and generalization ability of the extreme learning machine,this paper improves the traditional extreme learning machine by combining the particle swarm optimization algorithm and the number traversal method of hidden layer neurons.On the basis of improving the extreme learning machine,this paper proposes an improved extreme learning machine prediction method based on Fuzzy membrane clustering,which greatly improves the accuracy and stability of short-term load forecasting and photovoltaic power generation forecasting.Combining with the characteristics of power load,this paper first deeply analyses the influence of load characteristics,temperature,time and other factors on short-term load change;secondly,the fuzzy membrane clustering algorithm is used to select input samples on load forecasting day to improve the validity of input samples;finally,the particle swarm optimization algorithm and hidden layer neuron number traversal method are used to improve the limit learning machine.Line forecasting,using different methods to simulate the same season load in two areas and the same season load in the same area.The experimental results show that the combination forecasting model of the fuzzy membrane clustering algorithm and the improved extreme learning machine has high accuracy,and improves the accuracy of short-term load forecasting in power system.In order to improve the prediction accuracy of photovoltaic output power,this paper firstly makes correlation analysis on the influencing factors of photovoltaic output,and selects the influencing factors with larger correlation coefficient as the input samples of photovoltaic output power prediction.Secondly,the historical photovoltaic output power samples and their influencing factors are analyzed by using the fuzzy membrane clustering algorithm and various types of typical days are selected.Then,using the historical data and meteorological data of the forecast day,four methods,namely BP neural network,extreme learning machine,improved extreme learning machine and improved extreme learning machine based on Fuzzy membrane clustering,are used to predict a photovoltaic power generation system.The effectiveness of the forecasting method proposed in this paper is verified by comparative analysis.