Online unit load prediction method based on ensemble learning

Abstract

The invention discloses an online unit load prediction method based on ensemble learning. The method comprises the following steps: firstly, reading historical unit load data of a past month from a database, secondly, preprocessing the data, detecting a gross value according to a Pauta criterion, and processing the detected gross value in a backward filling manner; further performing feature extraction on the preprocessed data, including time features, correlation features and aggregation features, and constructing a feature data set; training a limit gradient lifting model and a lightweight efficient gradient lifting model through the feature data set, and training a long-short-term memory neural network model through a prediction result of an LGBM model and original feature data; and finally, taking output results of the three models as three inputs to train a linear regression model to output a final load prediction result. In a model training stage, network parameters are adjusted by taking a minimum prediction average error as a principle, so that an optimal unit load prediction model is established.

Description

technical field [0001] The invention belongs to the field of thermal power station load forecasting, and relates to an online unit load forecasting method based on integrated learning. Background technique [0002] Affected by factors such as the environment and seasons, the power demand of the power grid fluctuates greatly over time. According to the actual characteristics of power production and consumption, it is necessary to maintain a balanced relationship between the supply and demand sides. The resulting changes in power generation on the power supply side are mainly regulated by thermal power units through unit load changes. The power grid issues load planning curves to the power plants according to the actual conditions of the units and the corresponding power dispatching principles. There is a large deviation from the actual unit load command. Therefore, accurate unit actual load forecasting is helpful for the power plant to choose the adjustment method, respond t...

AI Technical Summary

Problems solved by technology

However, the non-linearity of the coefficients in the regression method reduces the accuracy of the model; the time prediction model is greatly affected by factors such as weather and climate; the prediction accuracy of the gray model is inversely proportional to the gray level, and when the dispersion of the data increases, the prediction accuracy decreases; The design requirements of the neural network are relatively high, the number of hidden layers is difficult to judge, and the convergence speed is slow; the support vector machine is difficult to handle large-scale training samples, and cannot reflect the long-term change law of the unit load; The accuracy of data is high, and it is difficult to overcome the interference of complex factors; the fuzzy system lacks self-learning ability, and its fuzzy rules mainly rely on expert systems, and the scope of use has relatively large limitations

Images