CN-122021985-A - Method and device for predicting carbon emission of power transmission line engineering

CN122021985ACN 122021985 ACN122021985 ACN 122021985ACN-122021985-A

Abstract

The application provides a method and a device for predicting carbon emission of a power transmission line project, and relates to the technical field of intersection of artificial intelligence and an energy system. The method comprises the steps of obtaining historical carbon emission data of a power transmission line project to be detected and external influence factors influencing the carbon emission data, generating a time sequence feature matrix based on the historical carbon emission data, generating a semantic vector based on the external influence factors, carrying out weight distribution on the time sequence feature matrix and the semantic vector by utilizing a multi-head attention mechanism, generating a fusion feature vector based on the time sequence feature matrix, the semantic vector, the weight of the time sequence feature matrix and the weight of the semantic vector, inputting the fusion feature vector into a carbon emission prediction model, outputting a carbon emission predicted value of the power transmission line project to be detected, and constructing the carbon emission prediction model based on BiLSTM networks. The method and the device can improve the accuracy of carbon emission prediction.

Inventors

HU SHIYAO
PANG NING
YUAN BO
ZHANG XINYUE
XIA JING
HONG CHONG
LI JINCHAO
XU XIAOBIN
LI SHIWEI
RAO YIWEN
HU PO
XI PENG

Assignees

国网河北省电力有限公司经济技术研究院
国家电网有限公司
华北电力大学

Dates

Publication Date: 20260512
Application Date: 20251208

Claims (10)

1. The carbon emission prediction method for the power transmission line engineering is characterized by comprising the following steps of: Acquiring historical carbon emission data of a power transmission line project to be tested and external influence factors influencing the carbon emission data, generating a time sequence feature matrix based on the historical carbon emission data, and generating a semantic vector based on the external influence factors; Performing weight distribution on the time sequence feature matrix and the semantic vector by utilizing a multi-head attention mechanism, and generating a fusion feature vector based on the time sequence feature matrix, the semantic vector, the weight of the time sequence feature matrix and the weight of the semantic vector; and inputting the fusion feature vector into a carbon emission prediction model, and outputting a carbon emission predicted value of the power transmission line project to be detected, wherein the carbon emission prediction model is constructed based on BiLSTM networks.
2. The transmission line engineering carbon emission prediction method according to claim 1, wherein the generating a time series feature matrix based on the historical carbon emission data includes: Performing modal decomposition on the historical carbon emission data by using empirical mode decomposition to obtain a plurality of eigenmode functions, wherein each eigenmode function is an oscillation component with different time scales in the historical carbon emission data; And carrying out normalization processing on each eigenmode function, and sliding all eigenmode functions subjected to normalization processing by utilizing a sliding window according to a preset window length and a preset sliding step length to generate the time sequence feature matrix.
3. The transmission line engineering carbon emission prediction method according to claim 1, wherein the generating a semantic vector based on the external influence factor includes: Extracting key features of the external influence factors by utilizing lasso regression, and taking the obtained key features affecting the carbon emission of the power transmission line project to be tested as first key features; Inputting the first key features into a semantic vector extraction model, and outputting the semantic vectors, wherein the semantic vector extraction model is constructed based on a Word2Vec model and a TextCNN model.
4. The method for predicting carbon emissions in power transmission line engineering according to claim 1, wherein the weighting of the timing feature matrix and the semantic vector by using a multi-head attention mechanism comprises: mapping the timing feature matrix to a first query vector, a first key vector, and a first value vector; mapping the semantic vector to a second query vector, a second key vector, and a second value vector; calculating the first query vector, the first key vector and the first value vector by adopting a multi-head attention mechanism to obtain the weight of the time sequence feature matrix; and calculating the second query vector, the second key vector and the second value vector by adopting a multi-head attention mechanism to obtain the weight of the semantic vector.
5. The method of claim 1, wherein the generating a fusion feature vector based on the timing feature matrix, the semantic vector, the weight of the timing feature matrix, and the weight of the semantic vector comprises: obtaining a first feature vector by using the time sequence feature matrix and the weight of the time sequence feature matrix; obtaining a second feature vector by using the semantic vector and the weight of the semantic vector; And generating the fusion feature vector by using the first feature vector and the second feature vector.
6. The transmission line engineering carbon emission prediction method according to claim 3, wherein after the inputting of the fusion feature vector into a carbon emission prediction model and outputting the predicted value of carbon emission of the transmission line engineering to be measured, the method further comprises: calculating to obtain feature importance weights by using the carbon emission predicted values and the fusion feature vectors; And calculating an activation graph by utilizing the feature importance weight and the fusion feature vector so as to quantify the contribution strength of the first key feature to the carbon emission predicted value of the power transmission line engineering to be measured.
7. The transmission line engineering carbon emission prediction method according to claim 1, wherein after the obtaining of the historical carbon emission data and the external influence factors of the transmission line engineering to be measured, the method further comprises: Filling missing values of the historical carbon emission data by adopting a linear interpolation method, and eliminating dimensions of the historical carbon emission data by adopting z fraction standardization; Performing word segmentation on the external influence factors by using a word segmentation tool, and constructing an initial characteristic index pool by the external influence factors subjected to the word segmentation; correspondingly, the generating the time sequence feature matrix based on the historical carbon emission data and the generating the semantic vector based on the external influence factors comprise the following steps: Generating a time sequence feature matrix based on the historical carbon emission data after the missing value filling and dimension elimination, and generating a semantic vector based on the initial feature index pool.
8. The transmission line engineering carbon emission prediction method according to claim 7, wherein the filling the historical carbon emission data with missing values by using a linear interpolation method comprises: Acquiring a first effective value and a second effective value of each missing time point, wherein the first effective value is an adjacent effective value nearest to the forward number of each missing time point, and the second effective value is an adjacent effective value nearest to the backward number of each missing time point; and calculating to obtain the missing value corresponding to the corresponding missing time point by using the first effective value, the second effective value, the time point of the first effective value and the time point of the second effective value of each missing time point.
9. The transmission line engineering carbon emission prediction method according to claim 7, wherein the step of performing dimension elimination on the historical carbon emission data by using z-score normalization includes: Calculating a characteristic mean value and a standard deviation of the historical carbon emission data; and obtaining the historical carbon emission data of the elimination dimension by using the historical carbon emission data, the characteristic mean value and the standard deviation of the historical carbon emission data.
10. The utility model provides a transmission line engineering carbon emission prediction device which characterized in that includes: The acquisition module is used for acquiring historical carbon emission data of the power transmission line project to be detected and external influence factors influencing the carbon emission data, generating a time sequence feature matrix based on the historical carbon emission data and generating semantic vectors based on the external influence factors; The fusion module is used for carrying out weight distribution on the time sequence feature matrix and the semantic vector by utilizing a multi-head attention mechanism, and generating a fusion feature vector based on the time sequence feature matrix, the semantic vector, the weight of the time sequence feature matrix and the weight of the semantic vector; And the prediction module is used for inputting the fusion feature vector into a carbon emission prediction model, outputting a carbon emission predicted value of the power transmission line project to be detected, and constructing the carbon emission prediction model based on BiLSTM networks.

Description

Method and device for predicting carbon emission of power transmission line engineering Technical Field The application relates to the technical field of intersection of artificial intelligence and an energy system, in particular to a method and a device for predicting carbon emission of a power transmission line project. Background Along with acceleration of low-carbon transformation of the global energy structure, the power transmission line engineering is taken as a core infrastructure of the power system, and accurate measurement and calculation of the carbon emission of the whole life cycle and trend prejudgment become key technical links for realizing the 'double-carbon' target. The traditional carbon emission prediction model is mainly based on historical operation data to construct time series regression analysis, and information such as external policy specifications and economic and technical environments are not fully fused, so that response to unstructured influencing factors such as industry policy adjustment and technical standard change is delayed. The existing deep learning model can process multi-source data input, but is difficult to effectively analyze the cross correlation between the structural numerical characteristics and text semantic characteristics, and most black box models cannot position key driving factors of a prediction result, so that a prediction conclusion lacks traceable interpretation, and the falling to the ground in engineering design and environmental protection decision is limited. The prior art has obvious bottleneck in processing multi-mode carbon emission data, namely firstly, drive elements hidden in unstructured text are difficult to be effectively extracted by a traditional structured model, semantic gaps between text features and numerical data lead to insufficient information interaction between modes, secondly, the problem of multiple collinearity caused by redundant variables in a high-dimensional feature space can weaken the generalization capability of the model, and the conventional dimension reduction method can not realize feature optimization while retaining the semantic meaning of a key text. The problems cause the defects that a prediction model is distorted in response to sudden policy factors, a key driving mechanism cannot be accurately captured and the like in actual engineering, and the deep application of a digital carbon reduction technology is hindered. Disclosure of Invention The application provides a method and a device for predicting carbon emission of a power transmission line project, which are used for solving the problem of inaccurate carbon emission prediction in the power transmission line project in the prior art. In a first aspect, the present application provides a method for predicting carbon emission in power transmission line engineering, including: Acquiring historical carbon emission data of a power transmission line project to be tested and external influence factors influencing the carbon emission data, generating a time sequence feature matrix based on the historical carbon emission data, and generating a semantic vector based on the external influence factors; Performing weight distribution on the time sequence feature matrix and the semantic vector by utilizing a multi-head attention mechanism, and generating a fusion feature vector based on the time sequence feature matrix, the semantic vector, the weight of the time sequence feature matrix and the weight of the semantic vector; and inputting the fusion feature vector into a carbon emission prediction model, and outputting a carbon emission predicted value of the power transmission line project to be detected, wherein the carbon emission prediction model is constructed based on BiLSTM networks. In a second aspect, the present application provides a carbon emission prediction apparatus for power transmission line engineering, including: The acquisition module is used for acquiring historical carbon emission data of the power transmission line project to be detected and external influence factors influencing the carbon emission data, generating a time sequence feature matrix based on the historical carbon emission data and generating semantic vectors based on the external influence factors; The fusion module is used for carrying out weight distribution on the time sequence feature matrix and the semantic vector by utilizing a multi-head attention mechanism, and generating a fusion feature vector based on the time sequence feature matrix, the semantic vector, the weight of the time sequence feature matrix and the weight of the semantic vector; And the prediction module is used for inputting the fusion feature vector into a carbon emission prediction model, outputting a carbon emission predicted value of the power transmission line project to be detected, and constructing the carbon emission prediction model based on BiLSTM networks. The application provides a method and