CN-121981377-A - Carbon emission accounting method and device for mixed combustion of coal and biomass

CN121981377ACN 121981377 ACN121981377 ACN 121981377ACN-121981377-A

Abstract

The invention discloses a carbon emission accounting method and device for blending combustion of coal and biomass, relates to the technical field of carbon emission accounting, and mainly aims to realize carbon emission accounting for blending combustion of coal and biomass. The method comprises the main technical scheme of obtaining a current observation value and a historical sequence value of an observation vector, obtaining a current state estimation value and a current boiler operation parameter of the state vector, obtaining theoretical observation values of a plurality of observation vectors according to the current state estimation value and a pre-established physical observation model of the state vector, inputting the historical sequence value, the state estimation value and the current boiler operation parameter into a machine learning model to obtain a systematic deviation predicted value of the physical observation model, observing residual errors according to the current observation value, the theoretical observation value and the systematic deviation predicted value, updating the state estimation value by adopting an extended Kalman filtering algorithm based on the observed residual errors, and calculating the coal-fired carbon emission and the biomass source carbon emission according to the updated state estimation value.

Inventors

SONG YUNCHANG
JIANG LONG
DU LEI
WANG YUE
LI JINJING
LI ZHANGUO
CHENG LIANG
ZHAO ZHENNING

Assignees

华北电力科学研究院有限责任公司
国家电网有限公司

Dates

Publication Date: 20260505
Application Date: 20251225

Claims (10)

1. A method for accounting carbon emissions from blending combustion of coal with biomass, the method comprising: Acquiring current observation values of a plurality of observation vectors by using an online monitoring system, and acquiring historical observation sequence values of the plurality of observation vectors; acquiring current state estimation values of a plurality of state vectors and current boiler operation parameters, wherein the state vectors comprise instantaneous furnace charging flow of fire coal and instantaneous furnace charging flow of biomass; Obtaining theoretical observation values of a plurality of observation vectors according to the current state estimation values of the plurality of state vectors and a pre-established physical observation model; Inputting the historical observation sequence values corresponding to the plurality of observation vectors, the current state estimation values of the plurality of state vectors and the current boiler operation parameters into a pre-trained machine learning model to obtain systematic deviation prediction values, wherein the systematic deviation prediction values represent systematic prediction deviations of the physical observation model on the observation vectors under the current boiler operation parameters and the historical observation conditions; Obtaining observation residual errors of the corrected multiple observation vectors according to the current observation values, the theoretical observation values and the systematic deviation prediction values of the multiple observation vectors; based on the observation residual errors of the plurality of observation vectors, updating the current state estimation values of the plurality of state vectors by adopting an extended Kalman filtering algorithm to obtain updated state estimation values of the plurality of state vectors; and respectively calculating the carbon emission of the coal and the carbon emission of the biomass source according to the updated state estimation values of the plurality of state vectors.
2. The method of claim 1, wherein updating the current state estimates for the plurality of state vectors using an extended kalman filter algorithm based on the observation residuals for the plurality of observation vectors to obtain updated state estimates for the plurality of state vectors comprises: acquiring a state prediction covariance matrix, an observation jacobian matrix and an observation noise covariance matrix at the current moment; Obtaining a Kalman gain coefficient matrix based on the state prediction covariance matrix, the observation jacobian matrix and the observation noise covariance matrix, wherein the number of lines of the Kalman gain coefficient matrix is equal to the dimension of a state vector, and the number of columns is equal to the dimension of an observation vector, wherein elements marked by each line and each column represent correction weights of observation residual errors of corresponding observation vectors on the corresponding state vectors; and updating the current state estimation values of the plurality of state vectors by using the Kalman gain coefficient matrix and the observation residual errors of the plurality of observation vectors to obtain updated state estimation values of the plurality of state vectors.
3. The method of claim 2, wherein updating the current state estimates of the plurality of state vectors using the kalman gain coefficient matrix and the observation residuals of the plurality of observation vectors to obtain updated state estimates of the plurality of state vectors comprises: And determining a sum value between a corresponding current state estimation value and a weighted correction amount as a corresponding updated state estimation value for each state vector in the state vectors, wherein the weighted correction amount is obtained by respectively carrying out weighted summation on each element of a corresponding row in the Kalman gain coefficient matrix and observation residues of the plurality of observation vectors.
4. The method of claim 1, wherein calculating the coal carbon emissions and the biomass source carbon emissions from the updated state estimates of the plurality of state vectors, respectively, comprises: Obtaining the carbon emission of the fire coal according to the mass fraction of the carbon element of the fire coal, the instantaneous charging flow of the fire coal and the preset mass ratio; and obtaining the biomass carbon emission according to the mass fraction of carbon elements of the biomass, the instantaneous furnace charging flow of the biomass and the preset mass ratio.
5. The method of claim 1, wherein deriving the corrected observation residuals for the plurality of observation vectors based on the current observations, the theoretical observations, and the systematic bias predictions for the plurality of observation vectors comprises: For each observation vector, determining a difference value between a current observation value and a corresponding theoretical observation value as an original observation residual; for each observation vector, determining the difference between the original observation residual and the corresponding systematic deviation predicted value as a corrected observation residual.
6. A method according to claim 3, wherein after obtaining updated state estimates for a plurality of state vectors, the method further comprises: updating a state prediction covariance matrix at the current moment based on the Kalman gain coefficient matrix and the observation jacobian matrix; And carrying out Kalman gain calculation at the next moment according to the updated state prediction covariance matrix.
7. The method of claim 1, wherein the training step of the pre-trained machine learning model is: collecting sample data in a historical operation period, wherein the sample data comprises a plurality of groups of historical observation sequence values, historical state vector estimated values, historical boiler operation parameters and deviation reality values obtained according to historical theoretical observed values and historical actual observed values of a plurality of observation vectors; taking the historical observation sequence value, the historical state vector estimated value and the historical boiler operation parameter as input characteristics, and taking the deviation true value as an output label to construct a training data set; Training a machine learning model according to the training data set, stopping training when the prediction deviation of the model meets a preset precision threshold value, and obtaining the pre-trained machine learning model, wherein the machine learning model is a long-term and short-term memory network.
8. A carbon emissions accounting device for co-firing coal and biomass, the device comprising: The first acquisition unit is used for acquiring current observation values of a plurality of observation vectors by using the online monitoring system and acquiring historical observation sequence values of the plurality of observation vectors; the second acquisition unit is used for acquiring current state estimation values of a plurality of state vectors and current boiler operation parameters, wherein the state vectors comprise instantaneous furnace charging flow of fire coal and instantaneous furnace charging flow of biomass; The theoretical determination unit is used for obtaining theoretical observation values of a plurality of observation vectors according to the current state estimation values of the plurality of state vectors and a pre-established physical observation model; The deviation prediction unit is used for inputting the historical observation sequence values corresponding to the plurality of observation vectors acquired by the first acquisition unit, the current state estimation values of the plurality of state vectors acquired by the second acquisition unit and the current boiler operation parameters into a pre-trained machine learning model to acquire a systematic deviation prediction value, wherein the systematic deviation prediction value represents systematic prediction deviation of the physical observation model on the observation vectors under the current boiler operation parameters and the historical observation conditions; The deviation correcting unit is used for obtaining the observation residual errors of the plurality of corrected observation vectors according to the current observation values, the theoretical observation values and the systematic deviation prediction values obtained by the deviation predicting unit; The state correction unit is used for updating the current state estimation values of the plurality of state vectors by adopting an extended Kalman filtering algorithm based on the observation residual errors of the plurality of observation vectors obtained by the deviation correction unit to obtain updated state estimation values of the plurality of state vectors; and the emission accounting unit is used for respectively calculating the carbon emission of the coal and the carbon emission of the biomass source according to the updated state estimated values of the plurality of state vectors obtained by the state correcting unit.
9. A storage medium comprising a stored program, wherein the program, when run, controls an apparatus in which the storage medium is located to perform the carbon emission accounting method of blending fire coal with biomass as claimed in any one of claims 1 to 7.
10. A processor for running a program, wherein the program is run to perform the method for accounting for carbon emissions by blending coal with biomass according to any one of claims 1 to 7.

Description

Carbon emission accounting method and device for mixed combustion of coal and biomass Technical Field The invention relates to the technical field of carbon emission accounting, in particular to a carbon emission accounting method and device for mixed combustion of coal and biomass. Background As an important field of carbon emission, the power industry is undergoing a deep shift from pure energy consumption control to precise carbon management. The coal-fired thermal generator set has become a key technical path for reducing the emission of clean carbon and realizing the flexibility of fuel by blending renewable energy sources such as biomass (such as straw and wood dust) and the like. However, the implementation of this technical path faces a core challenge-the need to accurately distinguish and account for fossil and biomass sources in a hybrid combustion process in real timeDischarge amount. According to international carbon accounting rules, biomass combustion occursBecause the plant growth stage realizes the closed loop of carbon absorption, charging is avoided, and if the plant growth stage cannot be clearly distinguished, enterprises are forced to pay additional cost for the part of free zero carbon emission. The current mainstream carbon emission accounting method comprises an emission factor method, a material balance algorithm and a continuous emission monitoring system, wherein the emission factor method is calculated based on fuel consumption and a preset emission coefficient, the material balance algorithm is used for carrying out balance calculation on the carbon content of input and output materials according to the law of conservation of carbon elements, and the continuous emission monitoring system is used for directly measuring through a flue analysis instrumentConcentration and flow to achieve total monitoring. However, the specific contribution ratio of coal (fossil source) and biomass in the mixed fuel cannot be analyzed, so that the mixed accounting of zero carbon emission and fossil source carbon emission is caused, and the risk of accounting distortion is formed. Disclosure of Invention In view of the above problems, the present invention provides a method and an apparatus for carbon emission accounting for co-firing of coal and biomass, which mainly aims to implement carbon emission accounting for co-firing of coal and biomass. In order to solve the technical problems, the invention provides the following scheme: In a first aspect, the present invention provides a carbon emission accounting method for co-firing of coal and biomass, the method comprising: Acquiring current observation values of a plurality of observation vectors by using an online monitoring system, and acquiring historical observation sequence values of the plurality of observation vectors; acquiring current state estimation values of a plurality of state vectors and current boiler operation parameters, wherein the state vectors comprise instantaneous furnace charging flow of fire coal and instantaneous furnace charging flow of biomass; Obtaining theoretical observation values of a plurality of observation vectors according to the current state estimation values of the plurality of state vectors and a pre-established physical observation model; Inputting the historical observation sequence values corresponding to the plurality of observation vectors, the current state estimation values of the plurality of state vectors and the current boiler operation parameters into a pre-trained machine learning model to obtain systematic deviation prediction values, wherein the systematic deviation prediction values represent systematic prediction deviations of the physical observation model on the observation vectors under the current boiler operation parameters and the historical observation conditions; Obtaining observation residual errors of the corrected multiple observation vectors according to the current observation values, the theoretical observation values and the systematic deviation prediction values of the multiple observation vectors; based on the observation residual errors of the plurality of observation vectors, updating the current state estimation values of the plurality of state vectors by adopting an extended Kalman filtering algorithm to obtain updated state estimation values of the plurality of state vectors; and respectively calculating the carbon emission of the coal and the carbon emission of the biomass source according to the updated state estimation values of the plurality of state vectors. In a second aspect, the present invention provides a carbon emission accounting device for co-firing of coal and biomass, the device comprising: The first acquisition unit is used for acquiring current observation values of a plurality of observation vectors by using the online monitoring system and acquiring historical observation sequence values of the plurality of observation vectors; the second acquisition uni