CEEMD-LSTM-MLR-based short-term power load prediction method

Abstract

The invention provides a short-term power load prediction method based on CEEMD-LSTM-MLR, and the method comprises the steps: 1, obtaining power load data, and carrying out the preprocessing of an obtained data set; 2, decomposing input data into limited IMF components and a residual component through CEEMD, and combining and recombining the components into a high-frequency component and a low-frequency component according to the fluctuation period of each component; step 3, predicting the high-frequency component by using an LSTM neural network, and performing hyper-parameter optimization on the LSTM network by using a Bayesian algorithm; step 4, predicting the low-frequency component by applying MLR; and 5, superposing and reconstructing the prediction results of the components to obtain a final prediction result, and comparing the prediction result with a real load data value. According to the method, the CEEMD decomposition method is adopted, the problem of mode aliasing of a traditional EMD decomposition method and the problem of large EEMD reconstruction error are solved, and the Bayesian optimization algorithm is introduced based on the thought of respective prediction of different frequencies, so that the model prediction precision is further improved.

Description

technical field [0001] The invention relates to the technical field of power system load forecasting, in particular to a short-term power load forecasting method based on CEEMD-LSTM-MLR. Background technique [0002] With the continuous development of my country's power grid, the change of power load is becoming more and more complex, and the research on power load forecasting has become an important content of power grid management. Short-term power load forecasting usually refers to the forecasting of the load from one day to seven days in the future. It is the basis for dispatching centers to formulate power generation plans and power plant quotations. It is also an important part of the energy management system (EMS). It controls the operation, control and Planning has a very important impact. Improving the accuracy of power system short-term load forecasting can not only enhance the safety of power system operation, but also improve the economy of power system operatio...

AI Technical Summary

Problems solved by technology

Traditional forecasting methods include linear regression, gray model, autoregressive sliding average, etc. These methods usually use linear models. Although the structure is simple, there are problems such as low prediction accuracy; machine learning methods include support vector regression, random forest, artificial neural Networks, deep learning methods, etc. have great advantages in dealing with nonlinear problems, but most of them ignore the correlation of load data in time series

Images