CN-121072362-B - Method, device and storage medium for analyzing influence factors of battery carbon emission

CN121072362BCN 121072362 BCN121072362 BCN 121072362BCN-121072362-B

Abstract

The application discloses a method, equipment and a storage medium for analyzing influence factors of battery carbon emission, and belongs to the technical field of data processing. Acquiring life cycle data of a battery at the current moment, constructing a model through a map, constructing a knowledge fusion map library of the battery based on the life cycle data, generating a carbon emission prediction model according to the knowledge fusion map library, calculating the prediction contribution degree of each factor variable in the carbon emission prediction model through a characteristic contribution degree analysis algorithm, determining a target factor variable according to the prediction contribution degree, constructing a directed acyclic graph of the target factor variable based on a preset electrochemical constraint condition, calculating the conditional probability distribution of the electrochemical constraint condition in the directed acyclic graph, forming a coupling relation model of the target factor variable, and calculating the sensitivity coefficient of the factor variable in the coupling relation model to the carbon emission. The application improves the accuracy of analysis results by quantifying the sensitivity coefficient of each carbon emission factor under the coupling relation.

Inventors

FENG WEI
LIU JIE
LIU LUJING
XUE DENGGAO
LEI HANG

Assignees

深圳先进技术研究院

Dates

Publication Date: 20260508
Application Date: 20251110

Claims (9)

1. A method for analyzing influence factors of carbon emission of a battery, which is characterized by comprising the following steps: Acquiring life cycle data of a battery at the current moment from a data source, constructing a model through a map, and constructing a knowledge fusion map base of the battery based on the life cycle data; generating a carbon emission prediction model through deep learning according to the knowledge fusion map library; calculating the predicted contribution degree of each factor variable in the carbon emission prediction model through a characteristic contribution degree analysis algorithm, and determining a target factor variable according to the predicted contribution degree; constructing a directed acyclic graph of the target factor variable based on a preset electrochemical constraint condition, and calculating conditional probability distribution of the electrochemical constraint condition in the directed acyclic graph to form a coupling relation model of the target factor variable; Calculating a sensitivity coefficient of the target factor variable in the coupling relation model to the carbon emission through a preset sensitivity analysis model, wherein the sensitivity analysis model is used for evaluating the sensitivity degree of model output to input parameter change, and the sensitivity coefficient is used for measuring the change of an output value caused by unit change of the input parameter; The step of calculating the sensitivity coefficient of the target factor variable to the carbon emission in the coupling relation model through a preset sensitivity analysis model comprises the following steps: The method comprises the steps of extracting the conditional probability distribution of each target factor variable in the coupling relation model, sampling the target factor variables based on the conditional probability distribution to generate at least two groups of factor variable combinations, calculating probability distribution change of the carbon emission for each group of factor variable combinations, and determining the sensitivity coefficient of the target factor variable to the carbon emission according to the probability distribution change corresponding to all the factor variable combinations.
2. The method for analyzing influence factors of carbon emission of a battery according to claim 1, wherein the step of acquiring life cycle data of a current time of the battery from a data source and constructing a model by a map, and constructing a knowledge fusion map base of the battery based on the life cycle data comprises: crawling the life cycle data of the electrochemical energy storage battery at the current moment from a data source through a distributed crawler; The entity information of the life cycle data is identified through the map construction model, and entity disambiguation is carried out on the entity information to obtain entity data; And determining relation data between the entity data based on the life cycle data and the entity data through the map construction model to form the knowledge fusion map base.
3. The method for analyzing influence factors of carbon emission of a battery according to claim 1, wherein the step of generating the carbon emission prediction model through deep learning according to the knowledge fusion map library comprises: acquiring factor variables and predicted variables from the knowledge fusion map library; And training a predictive model to be trained based on the factor variable and the predictive variable to generate the carbon emission predictive model.
4. The method for analyzing influence factors of carbon emission of a battery according to claim 1, wherein the step of calculating a predicted contribution of each factor variable in the carbon emission prediction model by a characteristic contribution analysis algorithm, and determining a target factor variable according to the predicted contribution comprises: calculating the prediction contribution degree of each factor variable in the carbon emission prediction model to a prediction result through a characteristic contribution degree analysis algorithm; and sequencing the prediction results according to the prediction contribution degree, and determining the target factor variable according to the sequencing result.
5. The method for analyzing influence factors of carbon emission of a battery according to claim 1, wherein the steps of constructing a directed acyclic graph of the target factor variable based on a preset electrochemical constraint condition, and calculating a conditional probability distribution of the electrochemical constraint condition in the directed acyclic graph, and forming a coupling relation model of the target factor variable include: Establishing the directed acyclic graph of the target factor variable based on the electrochemical constraints; Calculating the conditional probability distribution of the electrochemical constraint conditions by constructing a correlation function to obtain a conditional probability table; and embedding the conditional probability table into the directed acyclic graph to form a coupling relation model of the target factor variable.
6. The method for analyzing influence factors of carbon emission of a battery according to claim 5, wherein the step of calculating a conditional probability distribution of the electrochemical constraint condition by constructing a correlation function, the conditional probability table comprising: performing pairwise association measurement on node variables in the directed acyclic graph based on a Gaussian function to form a target association function; Calculating expected values of model parameters in the target association function through a maximum expected algorithm, and iterating the target association function based on the expected values to obtain an association function; and calculating conditional probability distribution information according to the association function and the node variable, and integrating the conditional probability table based on the conditional probability distribution information.
7. The method for analyzing influence factors of carbon emission of a battery according to claim 6, wherein the steps of calculating expected values of model parameters in the target correlation function by a maximum expected algorithm, and iterating the target correlation function based on the expected values, to obtain the correlation function include: Constructing a likelihood function of model parameters in the target association function based on the maximum expectation algorithm; Determining an electrochemical regular constraint term according to the electrochemical constraint condition and the model parameter, and summing the product of the electrochemical regular constraint term and the weight coefficient with the likelihood function to form an objective function; Calculating gradient information of the objective function on model coefficients through a back propagation algorithm based on expected values corresponding to the objective function; According to the gradient information, iterating the weight coefficient and the target correlation function, and calculating a convergence value of the target function after iteration; And when the convergence value is smaller than a preset convergence threshold value, determining the association function based on the updated model parameters.
8. An influence factor analysis device of a battery carbon emission, characterized in that the device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program being configured to implement the steps of the influence factor analysis method of a battery carbon emission as claimed in any of claims 1 to 7.
9. A storage medium, characterized in that the storage medium is a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, implements the steps of the method for analyzing influence factors of carbon emission of a battery according to any one of claims 1 to 7.

Description

Method, device and storage medium for analyzing influence factors of battery carbon emission Technical Field The present application relates to the field of data processing technologies, and in particular, to a method, an apparatus, and a storage medium for analyzing influencing factors of carbon emission of a battery. Background In the related art, the analysis method of the carbon emission of the battery mainly depends on a life cycle evaluation (LCA, life Cycle Assessment) framework, and the total carbon emission is calculated by collecting data of each link of the battery and combining the emission factors. Among them, in terms of factor analysis, most studies have employed independent modeling methods to identify key influencing factors, such as calculating the correlation of factor variables with emissions by linear regression, or evaluating the sensitivity of different input parameters based on variance decomposition. However, in the process of sensitivity calculation, the related technology generally analyzes the independent influence of each factor on the carbon emission in an isolated manner, and neglects the complex coupling relation of multiple factors in a battery system under the electrochemical constraint. In this case, it is difficult to accurately quantify the comprehensive influence of the interaction between factors on the carbon emission as a result of calculation of the sensitivity of each factor, resulting in lower accuracy of the analysis result of the carbon emission of the battery. The foregoing is provided merely for the purpose of facilitating understanding of the technical solutions of the present application and is not intended to represent an admission that the foregoing is prior art. Disclosure of Invention The application mainly aims to provide a method, equipment and a storage medium for analyzing influence factors of battery carbon emission, and aims to solve the technical problem of low accuracy of analysis results of the influence factors of battery carbon emission. In order to achieve the above object, the present application provides a method for analyzing influencing factors of carbon emission of a battery, the method comprising the steps of: Acquiring life cycle data of a battery at the current moment from a data source, constructing a model through a map, and constructing a knowledge fusion map base of the battery based on the life cycle data; generating a carbon emission prediction model through deep learning according to the knowledge fusion map library; calculating the predicted contribution degree of each factor variable in the carbon emission prediction model through a characteristic contribution degree analysis algorithm, and determining a target factor variable according to the predicted contribution degree; constructing a directed acyclic graph of the target factor variable based on a preset electrochemical constraint condition, and calculating conditional probability distribution of the electrochemical constraint condition in the directed acyclic graph to form a coupling relation model of the target factor variable; and calculating the sensitivity coefficient of the factor variable in the coupling relation model to the carbon emission through a preset sensitivity analysis model. In an embodiment, the step of acquiring life cycle data of the current time of the battery from the data source, and constructing a knowledge fusion map base of the battery based on the life cycle data through a map construction model includes: crawling the life cycle data of the electrochemical energy storage battery at the current moment from a data source through a distributed crawler; The entity information of the life cycle data is identified through the map construction model, and entity disambiguation is carried out on the entity information to obtain entity data; and determining relation data among the entity data based on the life cycle data and the entity data through the map construction model to form a knowledge fusion map base of the life cycle data. In an embodiment, the step of generating the carbon emission prediction model through deep learning according to the knowledge fusion map library includes: acquiring factor variables and predicted variables from the knowledge fusion map library; And training a predictive model to be trained based on the factor variable and the predictive variable to generate the carbon emission predictive model. In one embodiment, the step of calculating the predicted contribution of each factor variable in the carbon emission prediction model by using a feature contribution analysis algorithm, and determining the target factor variable according to the predicted contribution includes: calculating the prediction contribution degree of each factor variable in the carbon emission prediction model to a prediction result through a characteristic contribution degree analysis algorithm; and sequencing the prediction results according to the predic