Research On Short-term Load And New Energy Power Generation Forecasting

Accurate renewable resource and load prediction plays a key role in the progress of power grid planning schemes,and is essential for the effective dispatch and stable operation of power systems.The proportion of wind and solar energy access continues to increase,leading to the occurrence of wind abandonment and light abandonment.So,the analysis of absorption for wind and photovoltaic power are particularly important.Based on accurately predicting short-term load,wind and photovoltaic output,this paper analyzes the accommodation capacity of wind power and photovoltaic.For load forecasting,this paper proposes a frequency domain prediction method based on empirical wavelet transform(EWT)and improved random forest(IRF).Firstly,EWT is for the decomposing of load to obtain different intrinsic mode functions(IMFs).Secondly,a suitable method is used for prediction according to the characteristics of each component.The low and intermediate frequency components are predicted via IRF.Considering the meteorological factors temperature and humidity,the high frequency component is clustered using density-based spatial clustering of applications with noise(DBSCAN).Then the processing method is selected according to the sample characteristics of each class.Finally,the prediction results of each component are superimposed to obtain the total prediction result.Experiments are carried out based on the measured load data in an area of Anhui Province.Comparing the prediction results with the prediction results of empirical mode decomposition(EMD)–IRF,IRF,and random forest(RF)models.The proposed model has higher prediction accuracy and reflects the randomness of the actual load.For wind power prediction,this paper put forward a combined method of prediction based on EWT and IRF.Firstly,isolation random(i Forest)is used to clean abnormal data in historical wind power for data preprocessing.Secondly,EWT is used to decompose wind power to obtain IMFs.Then,different IRF models are built according to the characteristics of each component.Finally,the predicted values of each component are added to achieve accurate prediction of wind power.The prediction results are compared with the prediction results of EMD–IRF,IRF,and RF models,it is concluded that the proposed model has higher prediction accuracy.For photovoltaic power forecasting,this paper proposes a short-term photovoltaic power combined prediction model based on fuzzy C-means clustering(FCM)and IRF.Firstly,from the perspective of data preprocessing,the historical photovoltaic power and meteorological data are preprocessed,including abnormal data cleaning,feature selection and normalization.Secondly,XB index-based FCM algorithm is used to cluster based on the meteorological factors of historical day and prediction day.Finally,combined with IRF,the photovoltaic output of each category is predicted,and the predicted results are integrated according to the time points.Further,the predicted results of photovoltaic output are obtained.The simulation experiments are carried out by the proposed method.Compared with IRF,RF,and BP methods,the prediction results verify that the proposed method can effectively promote prediction precision.The forecast value of load,wind power and photovoltaic output in an area of Anhui Province are comprehensively evaluated by the summarized prediction level indicators.Three consumption indicators are used for the accommodation capacity analysis of wind power and photovoltaic.The results show that the proposed forecasting methods of load,wind power and photovoltaic power can obtain better forecasting effect.The analysis results of supplementary prediction level and accommodation indices provide reference for effective grid dispatching of power grid,sustainable and healthy development of energy.