Error Correction Of Time Serial Wind Speed Prediction Model

In the context of the global shortage of non-renewable energy,clean and renewable energy has become a key research object affecting social and economic development,and wind resources are a clean and easily available renewable resource,which has gradually become the backbone of the energy field.However,the wind speed sequence has serious problems such as random fluctuation,instability,and intermittency,which brings many technical problems to the wind power generation process.Wind speed is one of the main factors affecting wind power generation,and the accuracy of its prediction is very important for the safe and stable operation of wind power generation.In order to further improve the accuracy of wind speed prediction,this paper proposes a CNN-LSTM short-term wind speed prediction model based on error correction,which not only takes into account the wind speed sequence,but also takes into account other relevant meteorological factors affecting wind speed,in order to weaken the non-stationary characteristics of wind speed,a new influencing factor is constructed,and finally the error is analyzed and predicted to improve the model prediction accuracy,the main work is as follows:(1)Fully understand the relevant knowledge,theory and important concepts of wind power generation.Firstly,the neural network models involved in the research process are introduced,and then the theoretical knowledge and modeling process of decomposition algorithms and sample entropy algorithms are introduced.Ningxia wind farm data was selected to perform preprocessing work on the data and establish the wind farm data set to lay a foundation for the smooth progress of the follow-up work.(2)The CNN-LSTM wind speed prediction model is applied.Firstly,the LSTM model with meteorological factors as input variables and the LSTM model with single wind speed as input variables were explored to predict wind speeds,and the comparison results showed that the addition of meteorological factors could improve the prediction accuracy.Considering the instability of wind speed,a new influencing factor is established by decomposition and reconstruction,and the meteorological factor is input into the prediction model together with the new influencing factor,and the prediction results show that the accuracy of the model with the new factor is indeed higher than that of the model without the new influencing factor.The CNN-LSTM model is used in the modeling process,and the model established in this way can extract useful feature information for prediction and solve the problem of long-distance dependence.(3)An error correction model is added on the basis of the CNN-LSTM model.In order to make the model prediction effect better,a CEEMDAN-SE-LSTM model is designed to correct the error,first decompose the error to obtain the high-frequency and low-frequency components,and then superimpose the components with approximate sample entropy to obtain new components,and then LSTM prediction on the components.By comparing the model with the LSTM model and the EMD-SE-LSTM model,the proposed model has higher prediction accuracy,and finally the error prediction value and the initial predicted value are summed to obtain the final wind speed prediction value.The empirical analysis results show that the CNN-LSTM-error correction model adopted in this paper with new influencing factors has higher prediction accuracy and good fitting effect than the LSTM model and CNN-LSTM model.