Carbon emission prediction method, system, device and storage medium based on artificial intelligence

Abstract

The present invention discloses an artificial intelligence-based carbon emissions prediction method, system, device and storage medium, which specifically relates to the field of carbon emissions; by establishing an interface with a trusted data source, the latest carbon emissions data is acquired and integrated in real time to ensure that the model is corrected and adjusted on the most timely basis, and at the same time, by monitoring the correction time and analyzing the data distribution, the real-time performance of the model and the degree of data anomaly are accurately evaluated, thereby improving the overall prediction accuracy of the model; finally, through comprehensive analysis and continuous monitoring, the accuracy of the model output can be predicted and warned to ensure the reliability of decision-making, which not only enhances the adaptability and stability of the model, but also effectively prevents the generation of feedback loops, so that the model can maintain efficient carbon emissions prediction capabilities in a dynamically changing environment, and provide solid data support for the company's emission reduction decisions.

Description

Technical Field

[0001] The present invention relates to the field of carbon emissions, and in particular to an artificial intelligence-based carbon emissions prediction method, system, device and storage medium. Background Art

[0002] AI-based carbon emissions forecasting refers to the use of AI technology to analyze and predict future carbon emissions trends. These systems typically use machine learning algorithms to identify patterns and factors related to carbon emissions, such as energy consumption, industrial production, and transportation, from large amounts of historical data. Through these patterns, AI can predict carbon emissions in a specific time period in the future, helping companies, governments, and organizations better plan emission reduction measures. For example, AI can identify which industries or regions have faster carbon emissions growth, so that the corresponding emission reduction measures can be prioritized. The application of this technology helps promote the achievement of global emission reduction goals and address the challenges of climate change. However, in some cases, the prediction results of AI models may be directly used to make policies or corporate decisions, which in turn will affect future carbon emissions data. If decision makers rely too much on these predictions, the model may continue to strengthen its own prediction results, forming a feedback loop. For example, if AI predicts that carbon emissions in a certain industry will increase significantly, the government may take measures in advance, such as increasing taxes or setting strict emission standards. These measures may change the behavior of the industry, which in turn affects future carbon emissions data, making the original predictions and adjustments of the model inaccurate. This feedback loop may not only lead to invalid predictions, but also make the decision-making process rigid and dependent on unreliable model outputs. Summary of the invention

[0003] The purpose of the present invention is to provide a carbon emissions prediction method, system, device and storage medium based on artificial intelligence to address the deficiencies in the background technology.

[0004] In order to achieve the above object, the present invention provides the following technical solution: a carbon emissions prediction method based on artificial intelligence, comprising the following steps:

[0005] S1: Establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the obtained carbon emission data into the model in real time;

[0006] S2: Input the carbon emission data obtained in real time into the model for correction, record the time from receiving the new data to completing the correction of the monitoring model, determine the degree to which the model correction time deviates from the standard correction time, and evaluate the real-time performance of the model correction;

[0007] S3: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of abnormality of data distribution;

[0008] S4: Comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results;

[0009] S5: Further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make a prediction and early warning on the accuracy of the model output results based on the analysis results.

[0010] Preferably, in S2, the carbon emission data acquired in real time is input into the model for correction, the time from receiving the new data to completing the correction of the monitoring model is recorded, the degree of deviation of the model correction time from the standard correction time is determined, and the real-time performance of the model correction is evaluated, specifically:

[0011] Calculate the average time it takes for the model to complete the calibration, which is defined as the standard calibration time Tstandard. When new carbon emission data is input into the model, the timestamp is recorded immediately, which is defined as Tstart. When the model completes the calibration and generates a new prediction result, the timestamp is recorded immediately, which is defined as Tend.

[0012] The actual time Tactual used by the calculation model to complete the correction is expressed as: Compare the actual calibration time with the standard calibration time and calculate the deviation degree D of the calibration time. The expression is: If D > 0, it means that the actual calibration time is greater than or equal to the standard time, that is, the model calibration is slow; if D < 0, it means that the actual calibration time is less than the standard time, that is, the model calibration is fast.

[0013] Preferably, the actual correction time greater than or equal to the standard time within the S time period is collected, and a data set is established to calculate the correction time anomaly index to evaluate the real-time performance of the model correction. The correction time anomaly index is obtained as follows:

[0014] In the time period S, the actual time Tactual of each model correction is collected and arranged in chronological order to form a time series data set {Tactual,t}, where t represents the time point. The ARIMA model is constructed, and the time data {Tactual,t} is corrected by historical data. The ARIMA model is trained and the model parameters (p, d, q) are fitted. Among them, p is the order of the autoregressive term, that is, the correction time at the past moment affects the current correction time, d is the number of differences, which ensures the stability of the time series, and q is the order of the moving average term, that is, the number of lags of the error term. The trained ARIMA model is used to predict the correction time Tpred,t at future time points. The expression is: Calculate the prediction error et at each time point, that is, the difference between the actual correction time and the predicted correction time: ; Calculate the correction time anomaly index, the expression is: ,in, , is the indicator function, when hour, =1, otherwise 0, θ is the abnormal threshold, Tstandard is the standard correction time, is the corrected time anomaly index.

[0015] Preferably, in S3, the statistical distribution characteristics of the carbon emission data acquired in real time are analyzed to determine the deviation from the historical data distribution and to evaluate the abnormality of the data distribution, specifically:

[0016] Collect historical carbon emission data within the S time period to form a historical data set H={H1,H2,…,Hn}, where n is the number of historical data; obtain real-time carbon emission data within the current period to form a real-time data set R={R1,R2,…,Rm}, where m is the number of real-time data; calculate the mean of the historical data set and standard deviation , and the mean of the real-time dataset and standard deviation , calculate the difference between the real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data , the expressions are: ; ; In the formula, is the mean of the real-time data set, is the mean of the historical data set, is the standard deviation of the real-time data set, is the standard deviation of the historical data set; the difference between the calculated real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data Calculate the data distribution deviation index, the expression is: ; Where DI is the data distribution deviation index.

[0017] Preferably, in S4, the corrected time anomaly index and the data distribution deviation index are normalized, and the overall prediction accuracy coefficient of the model is calculated by the normalized corrected time anomaly index and the data distribution deviation index.

[0018] Preferably, the acquired overall prediction accuracy coefficient of the model is compared with a preset overall prediction accuracy coefficient reference threshold. If the overall prediction accuracy coefficient of the model is greater than or equal to the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is high. In this case, a prediction accuracy signal is generated and the model structure is not adjusted. If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is low. In this case, a prediction inaccuracy signal is generated and the model structure is adjusted.

[0019] Preferably, in S5, the overall prediction accuracy of the adjusted model within a fixed time period is further analyzed, and a prediction warning is performed on the accuracy of the model output results based on the analysis results;

[0020] If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, the model structure needs to be adjusted, and the overall prediction accuracy coefficient of the model that is less than the overall prediction accuracy coefficient reference threshold at each moment in a fixed time period after the adjustment is collected, and the difference between the coefficient and the actual value is calculated. The calculated differences are collected to form a difference time series, and the standard deviation of the difference time series is calculated, and the abnormal coefficient of the model output result is calculated.

[0021] Preferably, the overall prediction accuracy coefficient of the model and the standard deviation of the difference time series are converted into a first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the abnormal coefficient value label of each group of first eigenvector prediction model output results as the prediction target, and minimizes the sum of the prediction errors of the abnormal coefficient value labels of all model output results as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The abnormal coefficient value of the model output result is determined according to the model output result, wherein the machine learning model is a polynomial regression model;

[0022] If the abnormal coefficient value of the model output result is less than the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is high, and no warning signal is generated at this time;

[0023] If the abnormal coefficient value of the model output result is greater than or equal to the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is low, and a warning signal is generated at this time.

[0024] The present invention also provides a carbon emission prediction system based on artificial intelligence, including a data acquisition module, a model correction module, a data distribution analysis module, a model structure adjustment module and a prediction and early warning module;

[0025] Data acquisition module: establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the acquired carbon emission data into the model in real time;

[0026] Model correction module: inputs the carbon emission data obtained in real time into the model for correction, records the time from receiving new data to completing the correction of the monitoring model, determines the degree to which the model correction time deviates from the standard correction time, and evaluates the real-time performance of the model correction;

[0027] Data distribution analysis module: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of data distribution anomaly;

[0028] Model structure adjustment module: comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results;

[0029] Prediction and warning module: further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make predictions and warnings on the accuracy of the model output results based on the analysis results.

[0030] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the above-mentioned carbon emission prediction method based on artificial intelligence when executing the computer program.

[0031] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the above-mentioned carbon emission prediction method based on artificial intelligence is implemented.

[0032] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0033] 1. The present invention establishes an interface with a trusted data source and integrates data in real time, so that the model can obtain the latest carbon emission data in a timely manner to ensure the timeliness of the prediction results. Based on a comprehensive analysis of data distribution characteristics and model correction time, the scheme further evaluates and adjusts the model structure to improve the adaptability and prediction accuracy of the model. In particular, through the application of normalization processing and machine learning algorithms, the overall prediction accuracy coefficient and abnormal coefficient of the model can be accurately calculated and predicted, thereby realizing an automated early warning mechanism to help timely discover and correct potential prediction deviations.

[0034] 2. This invention significantly improves the overall performance of the carbon emission prediction model, allowing the model to maintain a high degree of prediction accuracy in a dynamically changing environment. Through continuous monitoring and intelligent adjustment of the model output results, enterprises can more effectively respond to challenges in carbon emission management and make scientific and accurate decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0036] Figure 1 The figure is a flow chart of the method of the present invention.

[0037] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Example 1, please refer to Figure 1 As shown, the carbon emission prediction method based on artificial intelligence described in this embodiment includes the following steps:

[0040] S1: Establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the obtained carbon emission data into the model in real time;

[0041] S2: Input the carbon emission data obtained in real time into the model for correction, record the time from receiving the new data to completing the correction of the monitoring model, determine the degree to which the model correction time deviates from the standard correction time, and evaluate the real-time performance of the model correction;

[0042] S3: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of data distribution abnormality;

[0043] S4: Comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results;

[0044] S5: Further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make a prediction and early warning on the accuracy of the model output results based on the analysis results.

[0045] Among them, in S1, an interface with a trusted data source is established to regularly obtain the latest carbon emission data, and the obtained carbon emission data is integrated into the model in real time.

[0046] Identify trusted data sources: Identify the main sources of carbon emissions data, such as government environmental monitoring agencies (EPA), international organizations (IPCC), energy monitoring companies (such as the IEA), satellite data providers, or industry reports.

[0047] According to the interface provided by the data source, select appropriate technology to obtain data, such as API call, FTP download, database connection, or Web Scraping.

[0048] API call: If the data source provides an API, use HTTP requests to periodically call the API to obtain data. FTP download: If the data is provided through an FTP server, write a script to periodically download the latest data from the FTP server. Database connection: If the data source provides database access, use SQL queries to obtain data from the remote database. Web Scraping: If the data is publicly available on the website but there is no API interface, use crawler technology to extract data periodically.

[0049] Write data parsing scripts according to the formats provided by the data source (such as JSON, CSV, XML, etc.) to convert raw data into standardized formats to facilitate subsequent processing and integration.

[0050] Check the data for completeness, including missing values, duplicate records, or inconsistent data entries. Supplement or delete as needed.

[0051] Convert data from different sources to a common unit of measurement and time scale (e.g. tonnes of CO2 / month) and ensure that it matches the format required by the model.

[0052] Store the cleaned data in an intermediate database or data warehouse for subsequent integration and use.

[0053] Set up an automated push mechanism. Whenever new data is acquired and preprocessed, it is immediately pushed to the database used by the model or directly loaded into memory for model calls.

[0054] If the amount of data is large, use a batch processing method to push new data to the model all at once.

[0055] If the data source provides real-time streaming data (such as IoT sensor data), use stream processing technologies (such as Apache Kafka and Apache Flink) to achieve real-time data integration.

[0056] Before integrating data into the model, real-time validation is performed to ensure data quality and consistency. Validation includes timestamp verification, data range checking, and outlier identification.

[0057] S2: Input the carbon emission data obtained in real time into the model for correction, record the time from receiving the new data to completing the correction of the monitoring model, determine the degree to which the model correction time deviates from the standard correction time, and evaluate the real-time performance of the model correction.

[0058] Based on historical data calibration records or through multiple experiments, the average time it takes for a model to complete calibration is calculated and defined as the standard calibration time (Tstandard). For example, if the average time it takes for a model to complete calibration is 2 minutes, then Tstandard = 2 minutes.

[0059] When new carbon emission data is input into the model, a timestamp is recorded immediately, defined as Tstart.

[0060] When the model completes the calibration and generates a new prediction result, the timestamp is recorded immediately, which is defined as Tend.

[0061] The actual time Tactual used by the calculation model to complete the correction is expressed as: For example, if T_start = 14:00:00, Tend = 14:02:30, then Tactual = 2.5 minutes.

[0062] Compare the actual calibration time with the standard calibration time and calculate the deviation degree D of the calibration time. The expression is: If D > 0, it means that the actual calibration time is greater than or equal to the standard time, that is, the model calibration is slow.

[0063] If D < 0, it means that the actual calibration time is less than the standard time, that is, the model calibration is fast.

[0064] For example, if Tstandard = 2 minutes and Tactual = 2.5 minutes, then: minute.

[0065] Define an allowable time deviation range (for example, ±0.2 minutes), indicating that the model correction time is acceptable within this range.

[0066] Based on the calculated D value, determine whether the correction time deviation is within the allowable range. If |D| ≤ 0.2 minutes, the correction time deviation is within the acceptable range and no adjustment is required.

[0067] If |D| > 0.2 minutes, the correction time deviation is beyond the acceptable range and the data processing pipeline or model performance may need to be reviewed.

[0068] The actual correction time greater than or equal to the standard time in the S time period is collected and a data set is established. The correction time anomaly index is calculated to evaluate the real-time performance of the model correction. The correction time anomaly index is obtained as follows:

[0069] In the time period S, the actual time Tactual of each model correction is collected and arranged in chronological order to form a time series data set {Tactual,t}, where t represents the time point. The ARIMA model is constructed, and the time data {Tactual,t} is corrected by historical data. The ARIMA model is trained and the model parameters (p, d, q) are fitted. Among them, p is the order of the autoregressive term, that is, the correction time at the past moment affects the current correction time, d is the number of differences, which ensures the stability of the time series, and q is the order of the moving average term, that is, the number of lags of the error term. The trained ARIMA model is used to predict the correction time Tpred,t at future time points. The expression is: Calculate the prediction error et at each time point, that is, the difference between the actual correction time and the predicted correction time: ; Calculate the correction time anomaly index, the expression is: ,in, , is the indicator function, when hour, =1, otherwise 0, θ is the abnormal threshold, which is used to identify abnormal correction time, usually calculated by the mean and standard deviation of the prediction error, Tstandard standard correction time, is the corrected time anomaly index.

[0070] The larger the correction time anomaly index is, the worse the real-time performance of the model correction is. This means that when the model processes new carbon emission data, its correction time often significantly exceeds the standard correction time, and more anomalies occur. This deviation indicates that the model is slow to respond to changes in real-time data, which may cause the prediction results to fail to reflect the latest carbon emissions in a timely manner, thereby affecting the accuracy and timeliness of decision-making. Therefore, a higher AE is an indicator of poor real-time performance of the model, and the performance of the model and the data processing process need to be further optimized.

[0071] S3: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of abnormality of data distribution.

[0072] Collect historical carbon emission data within the S time period to form a historical data set H={H1,H2,…,Hn}, where n is the number of historical data; obtain real-time carbon emission data within the current period to form a real-time data set R={R1,R2,…,Rm}, where m is the number of real-time data; calculate the mean of the historical data set and standard deviation , and the mean of the real-time dataset and standard deviation , calculate the difference between the real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data , the expressions are: ; ; In the formula, is the mean of the real-time data set, is the mean of the historical data set, is the standard deviation of the real-time data set, is the standard deviation of the historical data set; the difference between the calculated real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data Calculate the data distribution deviation index, the expression is: ; Where DI is the data distribution deviation index.

[0073] The larger the data distribution deviation index is, the more it means that the real-time carbon emission data deviates significantly from the historical data in terms of mean and volatility, indicating that the model may not be able to accurately capture and adapt to the current trend of change. If this deviation persists, the correction results of the model may lag behind the actual situation, resulting in reduced prediction accuracy, which in turn affects the effectiveness of decision-making.

[0074] S4: Comprehensively analyze the real-time performance of model correction and the degree of data distribution abnormality, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results.

[0075] The corrected time anomaly index and the data distribution deviation index are normalized, and the overall prediction accuracy coefficient of the model is calculated by the normalized corrected time anomaly index and the data distribution deviation index.

[0076] For example, the present invention can use the following formula to calculate the overall prediction accuracy coefficient of the model, and the calculation expression is: ; In the formula, is the overall prediction accuracy coefficient of the model, is the time anomaly correction index, DI is the data distribution deviation index, is the proportional coefficient for correcting the time anomaly index and the data distribution deviation index, and 0.

[0077] The obtained overall prediction accuracy coefficient of the model is compared with the preset overall prediction accuracy coefficient reference threshold. If the overall prediction accuracy coefficient of the model is greater than or equal to the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is high. At this time, a prediction accuracy signal is generated and the model structure is not adjusted. If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is low. At this time, a prediction inaccuracy signal is generated and the model structure is adjusted.

[0078] The model structure adjustment specifically includes:

[0079] Introducing new features: Analyzing the deviation between real-time data and historical data may reveal some important features that have not been captured. These new features can be introduced through feature engineering to improve the model's adaptability to new data.

[0080] Model complexity adjustment: Adjust the complexity of the model according to the complexity of real-time data. For example, increase the number of layers or nodes of the neural network to improve the expressiveness of the model, or introduce more decision trees to improve the robustness of the random forest model.

[0081] S5: Further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make predictions and early warnings on the accuracy of the model output results based on the analysis results.

[0082] If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, the model structure needs to be adjusted, and the overall prediction accuracy coefficient of the model that is less than the overall prediction accuracy coefficient reference threshold at each moment in a fixed time period after the adjustment is collected, and the difference between the coefficient and the actual value is calculated. The calculated differences are collected to form a difference time series, and the standard deviation of the difference time series is calculated, and the abnormal coefficient of the model output result is calculated.

[0083] The overall prediction accuracy coefficient of the model and the standard deviation of the difference time series are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the abnormal coefficient value label of each group of first eigenvector prediction model output results as the prediction target, and minimizes the sum of prediction errors of the abnormal coefficient value labels of all model output results as the training target. The machine learning model is trained until the sum of prediction errors converges and the model training is stopped. The abnormal coefficient value of the model output result is determined according to the model output result, wherein the machine learning model is a polynomial regression model.

[0084] The method for obtaining the abnormal coefficient value of the model output result is: from the first feature vector training data of the trained machine learning model, obtain the corresponding function expression: ; In the formula, is the output function of the model, is the overall prediction accuracy coefficient of the model, is the standard deviation of the difference time series, The abnormal coefficient value of the model output result.

[0085] If the abnormal coefficient value of the model output result is less than the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is high, and no warning signal is generated at this time;

[0086] If the abnormal coefficient value of the model output result is greater than or equal to the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is low. At this time, a warning signal is generated, prompting relevant personnel that they may need to further adjust or pay attention to the prediction results.

[0087] In this embodiment, by establishing an interface with a trusted data source, the latest carbon emission data is regularly obtained, and the obtained carbon emission data is integrated into the model in real time; the carbon emission data obtained in real time is input into the model for correction, the time from receiving new data to completing correction of the monitoring model is recorded, the degree of deviation of the model correction time from the standard correction time is judged, and the real-time performance of the model correction is evaluated; the statistical distribution characteristics of the carbon emission data obtained in real time are analyzed, the deviation state from the historical data distribution is judged, and the degree of data distribution abnormality is evaluated; the real-time performance of the model correction and the degree of data distribution abnormality are comprehensively analyzed, the overall prediction accuracy of the model is evaluated, and the model structure is adjusted according to the evaluation results; the overall prediction accuracy of the adjusted model within a fixed time period is further analyzed, and the accuracy of the model output results is predicted and warned according to the analysis results. Ensure that the model always maintains efficient and reliable prediction capabilities. Not only does it improve the adaptability and stability of the model, but it can also effectively reduce prediction errors and enhance the precise management and decision support of carbon emissions.

[0088] Example 2, please refer to Figure 2 As shown, the carbon emission prediction system based on artificial intelligence described in this embodiment includes a data acquisition module, a model correction module, a data distribution analysis module, a model structure adjustment module and a prediction and early warning module;

[0089] Data acquisition module: establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the acquired carbon emission data into the model in real time;

[0090] Model correction module: inputs the carbon emission data obtained in real time into the model for correction, records the time from receiving new data to completing the correction of the monitoring model, determines the degree to which the model correction time deviates from the standard correction time, and evaluates the real-time performance of the model correction;

[0091] Data distribution analysis module: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of data distribution anomaly;

[0092] Model structure adjustment module: comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results;

[0093] Prediction and warning module: further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make predictions and warnings on the accuracy of the model output results based on the analysis results.

[0094] Embodiment 3, an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method for predicting carbon emissions based on artificial intelligence in embodiment 1 is implemented. Embodiment 3 of the present disclosure is only an example and should not impose any restrictions on the functions and scope of use of the embodiments of the present disclosure. The electronic device can be expressed in the form of a general-purpose computing device, for example, it can be a server device. The components of the device may include, but are not limited to: at least one processor, at least one memory, and a bus connecting different system components (including memory and processor. The bus includes a data bus, an address bus, and a control bus. The memory may include a volatile memory, such as a random access memory (RAM) and / or a cache memory, and may further include a read-only memory (ROM). The memory may also include a program tool having a set (at least one) of program modules, such program modules including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each of these examples or some combination thereof may include the implementation of a network environment.

[0095] The processor executes various functional applications and data processing by running a computer program stored in the memory. The device can also communicate with one or more external devices (such as a keyboard, a pointing device, etc.). This communication can be performed through an input / output (I / O) interface. In addition, the device can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN) and / or a public network, such as the Internet) through a network adapter. The network adapter communicates with other modules of the device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the device, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID disk arrays) systems, tape drives, and data backup storage systems. It should be noted that although several units / modules or subunits / modules of the device are mentioned in the above detailed description, this division is only exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.

[0096] Embodiment 4, a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the program, when executed by a processor, implements the steps of the carbon emission prediction method based on artificial intelligence in Embodiment 1, wherein the computer-readable storage medium may more specifically include but is not limited to: a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0097] In a possible implementation, the present disclosure may also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to perform the steps of implementing the carbon emission prediction method based on artificial intelligence described in Example 1. The program code for executing the present disclosure may be written in any combination of one or more programming languages. The program code may be executed completely on the user device, partially on the user device, as an independent software package, partially on the user device, partially on a remote device, or completely on a remote device.

[0098] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0099] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0100] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0101] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A carbon emissions prediction method based on artificial intelligence, characterized by: The steps include: S1: Establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the obtained carbon emission data into the model in real time; S2: Input the carbon emission data obtained in real time into the model for correction, record the time from receiving the new data to completing the correction of the monitoring model, determine the degree to which the model correction time deviates from the standard correction time, and evaluate the real-time performance of the model correction; The step S2 specifically includes: calculating the average time for the model to complete the calibration, which is defined as the standard calibration time Tstandard; when new carbon emission data is input into the model, the timestamp is recorded immediately, which is defined as Tstart; when the model completes the calibration and generates a new prediction result, the timestamp is recorded immediately, which is defined as Tend; calculating the actual time Tactual used by the model to complete the calibration, which is expressed as: Compare the actual calibration time with the standard calibration time and calculate the deviation degree D of the calibration time. The expression is: If D > 0, it means that the actual calibration time is greater than or equal to the standard time, that is, the model calibration is slow; if D < 0, it means that the actual calibration time is less than the standard time, that is, the model calibration is fast; The actual correction time greater than or equal to the standard time in the S time period is collected, and a data set is established to calculate the correction time anomaly index. The real-time performance of the model correction is evaluated. The correction time anomaly index is obtained as follows: in the time period S, the actual time Tactual of each model correction is collected and arranged in chronological order to form a time series data set {Tactual, t}, where t represents the time point, and an ARIMA model is constructed. The time data {Tactual, t} is corrected by historical data, and the ARIMA model is trained to fit the model parameters (p, d, q), where p is the order of the autoregressive term, that is, the correction time at the past moment affects the current correction time, d is the number of differences, to ensure the stability of the time series, and q is the order of the moving average term, that is, the number of lags of the error term. The trained ARIMA model is used to predict the correction time Tpred, t at future time points. The expression is: Calculate the prediction error et at each time point, that is, the difference between the actual correction time and the predicted correction time: ; Calculate the correction time anomaly index, the expression is: ,in, , N is the total number of time points in time period S, is the indicator function, when hour, =1, otherwise 0, θ is the abnormal threshold, Tstandard is the standard correction time, To correct the time anomaly index; S3: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of abnormality of data distribution; The step S3 specifically includes: collecting historical carbon emission data within a time period S to form a historical data set H={H1, H2, ..., Hn}, where n is the number of historical data; obtaining real-time carbon emission data within the current time period to form a real-time data set R={R1, R2, ..., Rm}, where m is the number of real-time data; calculating the mean of the historical data set and standard deviation , and the mean of the real-time dataset and standard deviation , calculate the difference between the real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data , the expressions are: ; ; In the formula, is the mean of the real-time data set, is the mean of the historical data set, is the standard deviation of the real-time data set, is the standard deviation of the historical data set; the difference between the calculated real-time data mean and the historical data mean and the difference between the standard deviation of real-time data and the standard deviation of historical data Calculate the data distribution deviation index, the expression is: ; In the formula, DI is the data distribution deviation index; S4: Comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results; The correction time anomaly index and the data distribution deviation index are normalized, and the overall prediction accuracy coefficient of the model is calculated by the normalized correction time anomaly index and the data distribution deviation index; The following formula is used to calculate the overall prediction accuracy coefficient of the model, and the calculation expression is: ; In the formula, is the overall prediction accuracy coefficient of the model, is the time anomaly correction index, DI is the data distribution deviation index, is the proportional coefficient for correcting the time anomaly index and the data distribution deviation index, and 0; S5: Further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make predictions and early warnings on the accuracy of the model output results based on the analysis results.

2. The method for predicting carbon emissions based on artificial intelligence according to claim 1, characterized in that: The obtained overall prediction accuracy coefficient of the model is compared with the preset overall prediction accuracy coefficient reference threshold. If the overall prediction accuracy coefficient of the model is greater than or equal to the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is high. At this time, a prediction accuracy signal is generated and the model structure is not adjusted. If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, it means that the overall prediction accuracy of the model is low. At this time, a prediction inaccuracy signal is generated and the model structure is adjusted.

3. The method for predicting carbon emissions based on artificial intelligence according to claim 1, characterized in that: In S5, the overall prediction accuracy of the adjusted model within a fixed time period is further analyzed, and a prediction warning is made on the accuracy of the model output results based on the analysis results; If the overall prediction accuracy coefficient of the model is less than the preset overall prediction accuracy coefficient reference threshold, the model structure needs to be adjusted, and the overall prediction accuracy coefficient of the model that is less than the overall prediction accuracy coefficient reference threshold at each moment in a fixed time period after the adjustment is collected, and the difference between the coefficient and the actual value is calculated. The calculated differences are collected to form a difference time series, and the standard deviation of the difference time series is calculated, and the abnormal coefficient of the model output result is calculated.

4. The method for predicting carbon emissions based on artificial intelligence according to claim 3, characterized in that: The overall prediction accuracy coefficient of the model and the standard deviation of the difference time series are converted into the first eigenvector, and the first eigenvector is used as the input of the machine learning model. The machine learning model uses the abnormal coefficient value label of each group of first eigenvector prediction model output results as the prediction target, and minimizes the sum of the prediction errors of the abnormal coefficient value labels of all model output results as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The abnormal coefficient value of the model output result is determined according to the model output result, wherein the machine learning model is a polynomial regression model; If the abnormal coefficient value of the model output result is less than the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is high, and no warning signal is generated at this time; If the abnormal coefficient value of the model output result is greater than or equal to the reference threshold of the abnormal coefficient value: it means that the overall prediction accuracy of the model within a fixed time period is low, and a warning signal is generated at this time.

5. An artificial intelligence-based carbon emissions prediction system, used to implement the artificial intelligence-based carbon emissions prediction method according to any one of claims 1 to 4, characterized in that: It includes data acquisition module, model correction module, data distribution analysis module, model structure adjustment module and prediction and early warning module; Data acquisition module: establish an interface with a trusted data source, regularly obtain the latest carbon emission data, and integrate the acquired carbon emission data into the model in real time; Model correction module: inputs the carbon emission data obtained in real time into the model for correction, records the time from receiving new data to completing the correction of the monitoring model, determines the degree to which the model correction time deviates from the standard correction time, and evaluates the real-time performance of the model correction; Data distribution analysis module: Analyze the statistical distribution characteristics of carbon emission data obtained in real time, determine its deviation from the historical data distribution, and evaluate the degree of data distribution anomaly; Model structure adjustment module: comprehensively analyze the real-time performance of model correction and the degree of abnormal data distribution, evaluate the overall prediction accuracy of the model, and adjust the model structure based on the evaluation results; Prediction and warning module: further analyze the overall prediction accuracy of the adjusted model within a fixed time period, and make predictions and warnings on the accuracy of the model output results based on the analysis results.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, it implements the artificial intelligence-based carbon emission prediction method described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the artificial intelligence-based carbon emission prediction method described in any one of claims 1 to 4 is implemented.

Images