Short-term load prediction method for household electricity utilization

Abstract

The invention discloses a short-term load prediction method for household electricity utilization, and the method comprises the steps: introducing a residual mechanism into an LSTM network, introducing a Scaled Dot-Product Attention mechanism into a decoding process, and constructing an Encoder-Decoder model. According to the method, similar day data are extracted by using a fuzzy clustering algorithm, and the data are normalized, so that the problems of high similarity and non-uniform dimension among the data are solved; in order to solve the problem that correlation between long input sequence information and sequences is lost, data are input into an Encoder-Decoder model combined with Scaled Dot-Product Attention, so that the weights of elements in an intermediate code on output at each moment of Decoder are different, and the influence of key factors is highlighted.

Description

technical field [0001] The invention belongs to the technical field of load forecasting, in particular to a short-term load forecasting method for household electricity consumption. Background technique [0002] Power load forecasting can be divided into short-term load forecasting and medium- and long-term load forecasting. Short-term load forecasting refers to daily load forecasting and weekly load forecasting, which are mostly used for short-term grid operation arrangement, static safety analysis, planned maintenance arrangement, etc. [0003] In the field of short-term load forecasting, there are artificial neural network (Artificial Neural Networks, ANN), wavelet transform (Wavelet Transform), fuzzy logic (Fuzzy Logic, FL), combined forecasting methods and other methods. The method of load forecasting is gradually transformed into a forecasted sequence, and the model is transformed from a single LSTM model to a Sequence to Sequence model with an attention mechanism. Th...

AI Technical Summary

Problems solved by technology

[0008] Aiming at the deficiencies in the existing technology, the present invention provides a short-term load forecasting method for household electricity consumption, based on the LSTM network combined with the Scaled Dot-Product Attention mechanism, and introduces the Attention mechanism into the Encoder-Decoder model to effectively highlight the factors that affect the load. Factors, solve the time series problems and nonlinear regression problems in the power system load data, so as to improve the forecasting effect

Images