CN-121981610-A - Carbon emission evaluation method and equipment for building component production procedure based on Bayesian network

CN121981610ACN 121981610 ACN121981610 ACN 121981610ACN-121981610-A

Abstract

The disclosure provides a carbon emission evaluation method and equipment for a building component production procedure based on a Bayesian network, and relates to the technical field of data processing. The implementation scheme is that a Bayesian network, which is formed by leading each value father node of each influence factor variable of each production process of a building component to point to a carbon emission index child node of each production process, is constructed, the structure of the Bayesian network is subjected to iterative learning by utilizing historical production data, so that the conditional probability distribution of each father node of each child node in a target structure of the Bayesian network is obtained, the causal sensitivity of each influence factor variable of each process to the carbon emission index under different values is subjected to network inference, and the causal sensitivity is utilized, so that the production process with carbon reduction value and the key influence factors with the carbon reduction value of the production process can be accurately screened.

Inventors

HUANG ZUJIAN
LU PEIJUN
LUO AO
LENG TIANXIANG
WANG YICHENG
ZHANG WENYU

Assignees

华南理工大学
广州建筑股份有限公司

Dates

Publication Date: 20260505
Application Date: 20260128

Claims (10)

1. A bayesian network-based carbon emission assessment method for a building element production process, comprising: Taking each value of each influencing factor variable of each first production process of a building component as a father node, and taking carbon emission indexes of each first production process as child nodes respectively, so as to construct a Bayesian network, wherein the influencing factor variables comprise component sizes and external environment factors, and the carbon emission indexes comprise carbon emission factors and process activity; Iteratively learning the structure of the Bayesian network based on the historical production data of each first production procedure to obtain the conditional probability distribution of each father node of each child node in the target structure of the Bayesian network; for each influencing factor variable of each first production process, respectively carrying out network inference on causal sensitivity of the influencing factor variable to the carbon emission index under different values in a target structure based on conditional probability distribution of each father node of each child node in the target structure; Determining a second production process having a carbon reduction value and a key influencing factor of the second production process having a carbon reduction value based on causal sensitivity of each influencing factor variable of each first production process to the carbon emission index at different values.
2. The method as recited in claim 1, further comprising: Determining the carbon emission importance level of each production process of the building component based on the historical carbon emission average value, the historical carbon emission peak value and the historical carbon emission variation coefficient of each production process; And determining a plurality of first production processes from the production processes based on the carbon emission importance level of each production process.
3. The method of claim 1, wherein the bayesian network comprises one-way connection edges from parent nodes to child nodes, excluding reverse connection edges from child nodes to parent nodes.
4. The method according to claim 1, wherein the iteratively learning the bayesian network based on the historical production data of each of the first production processes to obtain a conditional probability distribution of each parent node of each child node comprises: Determining an initial structure of the Bayesian network based on the directed relationship from the father node to the child node between the child node and the father node corresponding to each of the first production processes recorded in the history; Starting from an initial structure of the Bayesian network, iteratively executing a structure updating operation until a scoring difference between a first neighborhood structure corresponding to the Bayesian network and a current structure of the Bayesian network is smaller than a preset difference threshold, wherein the structure updating operation comprises the following steps of: Generating a domain structure meeting the constraint condition corresponding to the pointing relation based on the current structure of the Bayesian network; The scoring operation is carried out to obtain the score of the current structure and the score of each neighborhood structure, wherein the scoring operation is carried out to calculate the conditional probability distribution of each father node of each child node in the structure based on the historical production data, and determine the score of the structure based on the conditional probability distribution of each father node of each child node in the structure and the sample size of the historical production data; After the scoring operation is executed, selecting the first neighborhood structure from the neighborhood structures based on scoring differences between the scoring of the neighborhood structures and the current structure, and adding the first neighborhood structure into the current structure so as to update the current structure; And after the structure updating operation is executed to obtain the target structure of the Bayesian network, acquiring the conditional probability distribution of each father node of each child node in the target structure from the structure updating operation.
5. The method of claim 4, wherein determining the score for the structure based on the conditional probability distribution of each parent node of each child node in the structure and the sample size of the historical production data comprises: for each child node in the structure, respectively calculating the product between the value of each father node of the child node and the conditional probability of pointing to the child node, wherein the logarithmic value of the sample size of the historical production data related to the conditional probability, and determining the score of the child node based on the product and the logarithmic value; And summing the scores of the child nodes in the structure to obtain the score of the structure.
6. The method of claim 1, wherein the network inferring causal sensitivity of the influencing factor variable to the carbon emission indicator at different values based on a conditional probability distribution of each parent node of each of the child nodes in a target structure comprises: For a first child node corresponding to the first production process and each first value of a first influencing factor variable of the first production process, the steps of reserving connection between the first child node and a father node corresponding to the first value of the first influencing factor variable of the first child node, reserving connection between the first child node and each father node corresponding to the first influencing factor variable of the first child node, and disconnecting connection between the first child node and each father node corresponding to the first value of the first influencing factor variable of the first child node, so as to obtain a first network corresponding to the first value of the first influencing factor variable of the child node; based on the conditional probability distribution of each father node in the first network corresponding to each first value, respectively calculating the carbon emission index of the first child node under each first value of the first influence factor variable; and determining the causal sensitivity of the first influence factor variable of the first production process corresponding to the first child node to the carbon emission index of the first production process under different values based on the difference between the carbon emission indexes of the first child node under the first influence factor variable under the first value.
7. The method of claim 6, wherein the carbon emission indicator comprises a carbon emission factor, the conditional probability distribution of the parent node comprises a reference probability, a predicted probability, and an uncertainty variance of the parent node pointing to the first child node, and the calculating the carbon emission indicator of the first child node at each of the first values of the first influencing factor variable based on the conditional probability distribution of each parent node in the first network to which each of the first values corresponds comprises: summing products between standardized values and prediction probabilities of all father nodes in the first network with the first value, and adding the summation result and the reference probabilities of all the father nodes to obtain a carbon emission factor of the first child node under the first value of a first influence factor variable; the carbon emission factor is adjusted based on a sum of uncertainty variances of the respective parent nodes.
8. The method of claim 1, wherein the determining a second production process having a carbon reduction value based on causal sensitivity of each of the influencing factor variables of each of the first production processes to the carbon emission index at different values, and a key influencing factor of the second production process having a carbon reduction value comprises: And determining the influence factor variables with the causal sensitivity larger than a preset causal sensitivity threshold as key influence factors with the carbon reduction value of the first production process aiming at the influence factor variables of the first production process, and taking the first production process as a second production process with the carbon reduction value.
9. A bayesian network-based carbon emission assessment device for a building element production process, comprising: The network construction module is used for constructing a Bayesian network by taking each value of each influencing factor variable of each first production procedure of a building component as a father node and taking carbon emission indexes of each first production procedure as child nodes respectively, wherein the influencing factor variables comprise component sizes and external environment factors, and the carbon emission indexes comprise carbon emission factors and procedure activity; The network iteration module is used for carrying out iterative learning on the structure of the Bayesian network based on the historical production data of each first production procedure to obtain the conditional probability distribution of each father node of each child node in the target structure of the Bayesian network; The network inference module is used for performing network inference on causal sensitivity of the influence factor variables to the carbon emission index under different values in the target structure based on conditional probability distribution of each father node of each child node in the target structure for each influence factor variable of each first production process; The target determining module is used for determining a second production process with carbon reduction value and key influencing factors with the carbon reduction value of the second production process based on causal sensitivity of the influencing factor variables of the first production process to the carbon emission index under different values.
10. An electronic device, comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-8.

Description

Carbon emission evaluation method and equipment for building component production procedure based on Bayesian network Technical Field The present disclosure relates to the field of data processing technology. The present disclosure relates specifically to a carbon emission assessment method and apparatus for a bayesian network-based building element production process. Background At present, carbon emission evaluation is carried out on each working procedure in the production process of the building component, so that carbon reduction treatment is carried out on appointed working procedures of the building component, and the green carbon reduction effect of the building component in the production process is improved. In the existing carbon emission evaluation mode of the building member production process, a corresponding unit carbon emission factor is generally determined by looking up a table according to some material parameters or production indexes of the process, and then, the volume or area of the building member is combined to obtain an evaluation value of the carbon emission of the building member production process. However, in a real production environment, the unit carbon emission factor of a building element is not a fixed value, but is dynamically changed by the influence of multi-source variables such as element size, material characteristics, processing strength, energy structure, and climate environment. Thus, it is difficult to accurately evaluate the unit carbon emission factor by means of a look-up table, and further, it is impossible to accurately judge the production process of the building member having the carbon reduction value. Disclosure of Invention The present disclosure provides a carbon emission assessment method and apparatus for a bayesian network-based building element production process. According to an aspect of the present disclosure, there is provided a carbon emission assessment method of a building element production process based on a bayesian network, including: Taking each value of each influencing factor variable of each first production process of a building component as a father node, and taking carbon emission indexes of each first production process as child nodes respectively, so as to construct a Bayesian network, wherein the influencing factor variables comprise component sizes and external environment factors, and the carbon emission indexes comprise carbon emission factors and process activity; Iteratively learning the structure of the Bayesian network based on the historical production data of each first production procedure to obtain the conditional probability distribution of each father node of each child node in the target structure of the Bayesian network; for each influencing factor variable of each first production process, respectively carrying out network inference on causal sensitivity of the influencing factor variable to the carbon emission index under different values in a target structure based on conditional probability distribution of each father node of each child node in the target structure; Determining a second production process having a carbon reduction value and a key influencing factor of the second production process having a carbon reduction value based on causal sensitivity of each influencing factor variable of each first production process to the carbon emission index at different values. According to another aspect of the present disclosure, there is provided a carbon emission assessment device for a building element production process based on a bayesian network, including: The network construction module is used for constructing a Bayesian network by taking each value of each influencing factor variable of each first production procedure of a building component as a father node and taking carbon emission indexes of each first production procedure as child nodes respectively, wherein the influencing factor variables comprise component sizes and external environment factors, and the carbon emission indexes comprise carbon emission factors and procedure activity; The network iteration module is used for carrying out iterative learning on the structure of the Bayesian network based on the historical production data of each first production procedure to obtain the conditional probability distribution of each father node of each child node in the target structure of the Bayesian network; The network inference module is used for performing network inference on causal sensitivity of the influence factor variables to the carbon emission index under different values in the target structure based on conditional probability distribution of each father node of each child node in the target structure for each influence factor variable of each first production process; The target determining module is used for determining a second production process with carbon reduction value and key influencing factors with the carbon reduction value