Multi-scene design analysis of integrated energy system based on feature extraction algorithm

The specific analysis of a region's energy needs to model and simulate various types of energy, quantify energy information, and clearly and intuitively reflect the energy situation and energy potential of a region. In this paper, according to the input attributes of various energy load forecasting models, the correlation degree of main control factors is analyzed, and the influence degrees of environmental factors on electric power, gas, heating and cooling loads are obtained respectively. Then, convolution neural network is used to extract the feature vectors of comprehensive environmental factors. Finally, according to the given feature vectors, the feature clustering models of various energy loads are established by using K-means clustering algorithm, and the load forecasting results of multi-energy systems are obtained. The errors between the predicted results of various energy loads and the actual load records in the study area are 1.105%, 1.876%, 3.102% and 2.834%, respectively. The load forecasting method based on feature clustering proposed in this paper can effectively extract the influence of different environmental factors on the load forecasting results, and get more accurate load forecasting results. (C) 2022 The Author(s). Published by Elsevier Ltd.

Automated summary

This paper proposes a load forecasting method based on feature clustering, which analyzes the correlation degree of control factors and the influence of environmental factors on loads, extracts feature vectors using convolutional neural networks, establishes clustering models for various energy loads, and obtains accurate load forecasting results.