Influence of Different Input Variables on the Restoration of Abnormal Data of Photovoltaic Power
The output power data record of photovoltaic power stations and distributed photovoltaic equipment may exhibit a-nomalies due to measurement equipment abnormalities,communication failures,signal interference,and other factors,which will affect the power grid decision-making.Therefore,this study proposes a back-propagation neural network based on genetic algo-rithm optimization of initial values,utilizing the abnormal data repair of GA-BP neural network,and establishing a linear inter-polation data repair model as the control group.This article investigates the effects of using numerical meteorological records(radiation intensity,temperature,relative humidity,wind speed,and direction),weather types,power of nearby similar power stations,and different combinations of these parameters as input variables for neural networks on repair effectiveness.The ex-ample analysis shows that better repair effect can be obtained by using all input variables.
photovoltaicartificial neural networkabnormal data repairgenetic algorithminput variables
万方数据
维普
