Intelligent verification system for enterprise carbon emission data

Abstract

This invention provides an intelligent verification system for enterprise carbon emission data. By integrating multiple verification mechanisms and introducing adaptive learning capabilities, it effectively overcomes the shortcomings of poor adaptability of general rules in existing technologies. It can dynamically generate personalized verification rules based on the historical data of each power plant, achieving accurate adaptation to the data characteristics of different power plants and significantly improving the accuracy of anomaly data identification. Simultaneously, by integrating real-time calculation and multi-rule collaborative verification, it enhances the timeliness and coverage depth of verification. The system also provides highly flexible custom configuration capabilities and closed-loop anomaly analysis and management functions, making the carbon emission data quality control process more comprehensive, accurate, flexible, and traceable, thereby greatly improving the reliability, efficiency, and intelligence level of carbon emission data management in thermal power enterprises.

Description

Technical Field

[0001] This invention relates to the field of carbon emission data management technology, and in particular to an intelligent verification system for enterprise carbon emission data. Background Technology

[0002] In practical applications, carbon emission data reporting by thermal power plants involves multiple dimensions, including fuel characteristics, unit operation, and pollutant emissions. Data sources cover various channels such as manual reporting, real-time monitoring, and ledger records. Different thermal power plants exhibit significant differences in unit models, fuel structures, operating conditions, and management levels, placing higher demands on the adaptability, accuracy, and dynamic adjustment capabilities of data verification.

[0003] Traditional data verification methods primarily rely on fixed threshold rules based on industry-standard experience to determine the boundaries of reported data. However, thermal power plants exhibit significant individualization and dynamism in their actual production data due to factors such as fuel sources, unit equipment characteristics, operating conditions, and regional differences. General fixed threshold verification methods are ill-suited to the diverse historical data distributions and operational realities of different power plants, resulting in poor adaptability. This often leads to misjudgments or omissions, failing to accurately identify genuine data anomalies and potentially issuing incorrect warnings for reasonable data fluctuations. Consequently, the accuracy and reliability of data quality control are severely compromised, making it difficult to meet the demands of refined and intelligent carbon management. Summary of the Invention

[0004] In view of this, the present invention provides an intelligent verification system for enterprise carbon emission data to address the technical deficiencies in the prior art.

[0005] Specifically, the present invention provides an intelligent verification system for enterprise carbon emission data, comprising:

[0006] It includes a data access layer, a one-factory-one-policy learning and training layer, a multi-method verification layer, an anomaly warning and cause statistics layer, and a result output layer, which are connected in sequence, as well as a custom rule configuration layer connected to the multi-method verification layer; The data access layer is used to collect multi-source data, including data entered, real-time data, and historical data, perform standardization processing to obtain standardized data, and then output it. The "One Factory, One Policy" learning and training layer is used to receive historical data from standardized data, perform learning and training, and generate and output a dynamic verification range containing dynamic verification rules. The custom rule configuration layer is used to receive configuration instructions from the administrator, generate and output configuration rules for controlling the verification process and alert methods; The multi-mode verification layer is used to receive the currently entered data, dynamic verification rules and configuration rules in the standardized data, and to perform collaborative verification through threshold verification, historical data verification and real-time carbon emission calculation verification, and generate and output the verification results. The anomaly warning and cause statistics layer is used to receive the verification results, issue warnings when abnormal data appears in the verification results, determine the causes of the anomalies associated with the abnormal data, and generate and output warning information, cause statistics results and deviation analysis reports. The results output layer is used to receive early warning information, cause statistics, and deviation analysis reports, and to visualize and export them.

[0007] In some implementations, the data access layer includes a multi-source data integration module for collecting reported data, real-time data, and historical data through at least one of the following methods: application programming interface, database synchronization, direct connection to IoT devices, or cloud platform data subscription.

[0008] In some implementations, the "one factory, one policy" learning and training layer includes: The power plant data learning unit is used to learn from the historical data of all coal-fired power plants and build a data model of coal-fired power plants. The single-plant data learning unit is used to independently learn from the historical data of a single coal-fired power plant and build a dedicated data model for that plant. The dynamic range generation unit connects the power plant data learning unit and the single plant data learning unit. It is used to calculate the confidence interval based on the coal-fired power plant data model and the single plant-specific data model, according to the preset confidence level, so as to generate dynamic verification rules.

[0009] In some implementations, the multi-mode verification layer includes: The threshold verification unit is used to perform threshold verification on the core fuel indicators according to the configuration rules and dynamic verification rules, and generate the first verification result; The historical data verification unit is used to compare and verify the currently entered data with historical data according to the configured rules and by calling the dynamic verification rules, and to generate a second verification result. The real-time carbon emission calculation and verification unit is used to calculate and verify the emissions based on real-time data and an associated prediction model according to the configuration rules, and generate a third verification result. Among them, the threshold verification unit, the historical data verification unit, and the real-time carbon emission calculation and verification unit work together, and their outputs together constitute the verification result.

[0010] In some implementations, the threshold used by the threshold verification unit includes a static basic threshold and a dynamic optimized threshold, with the dynamic optimized threshold obtained according to the dynamic verification rules.

[0011] In some implementations, the core fuel indicators include at least one of the following: lower heating value of coal fed into the furnace, total water content of coal fed into the furnace, internal water content, total sulfur content on an air-dried basis, carbon content on a comprehensive sample basis, and hydrogen content on an air-dried basis.

[0012] In some implementations, the real-time carbon emission measurement and verification unit includes: The prediction module is used to calculate carbon emission prediction data based on real-time data through a correlation prediction model that links emissions, quotas, coal consumption for power generation, and coal consumption for heating. The comparison module, connected to the prediction module, is used to compare the carbon emission prediction data with the monthly carbon emission data in the currently submitted data, and to determine whether there is any abnormality based on the deviation ratio range set in the configuration rules.

[0013] In some implementations, the custom rule configuration layer includes a rule management unit, which performs at least one of the following operations to generate configuration rules based on configuration instructions: enabling or disabling threshold verification, historical data verification, or real-time carbon emission calculation verification; modifying the threshold range, confidence interval confidence level, or deviation ratio range; adding special verification rules for specific power plants or indicators, or deleting existing rules; and setting the method for pushing early warning information.

[0014] In some implementations, the anomaly warning and cause statistics layer includes: The automatic early warning unit is used to generate and push early warning information based on the verification results and configuration rules, including the abnormal data item identifier, the power plant identifier, the trigger verification rule identifier, and the degree of deviation. The anomaly cause statistics unit is used to match or receive manually added anomaly causes from the preset error cause categories, count the anomaly causes, and form a cause statistics ledger. The deviation analysis unit, connected to the anomaly cause statistics unit, is used to calculate the reporting deviation rate and anomaly frequency of core fuel indicators, analyze the distribution ratio of anomaly causes, and generate a reporting data deviation analysis report containing the reporting deviation rate, anomaly frequency, and distribution ratio.

[0015] In some implementations, the results output layer includes a visualization unit and a report generation unit, which are used to display early warning information, cause statistics and deviation analysis reports in a graphical interface or report form, and support data export.

[0016] At least one embodiment of the present invention effectively overcomes the shortcomings of poor adaptability of general rules in existing technologies by integrating multiple verification mechanisms and introducing adaptive learning capabilities. It can dynamically generate personalized verification rules based on the historical data of each power plant, achieving accurate adaptation to the data characteristics of different power plants and significantly improving the accuracy of anomaly data identification. Simultaneously, by integrating real-time calculation and multi-rule collaborative verification, the timeliness and coverage depth of the verification are enhanced. The system also provides highly flexible custom configuration capabilities and closed-loop anomaly analysis and management functions, making the carbon emission data quality control process more comprehensive, accurate, flexible, and traceable, thereby greatly improving the reliability, efficiency, and intelligence level of carbon emission data management in thermal power plants. Attached Figure Description

[0017] Figure 1 This is a structural block diagram of an intelligent verification system for enterprise carbon emission data provided by the present invention. Detailed Implementation

[0018] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0019] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the one or more embodiments of this specification. The singular forms “a” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items. The modifications “a” and “a plurality” as used in this disclosure are illustrative and not restrictive, and those skilled in the art will understand that they should be understood as “one or more” unless the context clearly indicates otherwise.

[0020] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0021] See Figure 1 , Figure 1 This diagram illustrates a structural block diagram of an intelligent verification system for enterprise carbon emission data according to some embodiments of this specification. The intelligent verification system for enterprise carbon emission data includes: In some embodiments, an intelligent verification system for enterprise carbon emission data includes, in sequence, a data access layer, a "one factory, one policy" learning and training layer, a multi-method verification layer, an anomaly warning and cause statistics layer, and a result output layer, as well as a custom rule configuration layer connected to the multi-method verification layer. The data access layer collects multi-source data including reported data, real-time data, and historical data, performs standardization processing to obtain standardized data, and outputs it. The "one factory, one policy" learning and training layer receives historical data from the standardized data, performs learning and training, generates and outputs a dynamic verification range containing dynamic verification rules. The custom rule configuration layer receives configuration instructions from the administrator and generates... The system generates and outputs configuration rules for controlling the verification process and early warning methods; the multi-mode verification layer receives the currently entered data, dynamic verification rules, and configuration rules from the standardized data, and performs collaborative verification through threshold verification, historical data verification, and real-time carbon emission calculation verification to generate and output verification results; the anomaly early warning and cause statistics layer receives the verification results, issues early warnings when abnormal data appears in the verification results, determines the causes of the anomalies associated with the abnormal data, and generates and outputs early warning information, cause statistics results, and deviation analysis reports; the result output layer receives early warning information, cause statistics results, and deviation analysis reports, and performs visualization and export.

[0022] The data access layer integrates data from multiple channels and performs standardized processing. Its core function is to provide a unified, complete, and effective data input for the entire system, ensuring consistency in data processing across subsequent stages. The "One Plant, One Policy" learning and training layer focuses on specific historical data from individual power plants to generate personalized dynamic verification rules. Its core objective is to achieve accurate adaptation between verification standards and the power plant's own data characteristics. The multi-method verification layer integrates various verification logics to collaboratively complete data quality judgment. Through the complementarity of different verification methods, it improves the comprehensiveness and accuracy of abnormal data identification. The custom rule configuration layer allows administrators to flexibly adjust verification rules according to enterprise needs, supporting the system's adaptation to different scenarios and dynamic response to management requirements. The anomaly warning and cause statistics layer provides warning pushes, cause analysis, and deviation statistics for anomaly data discovered during verification. Its core function is to achieve traceability and analyzability of anomaly data. The results output layer presents verification results, warning information, and analysis reports in an intuitive format and supports export, facilitating user viewing, use, and data retention. Multi-source data can come from different collection channels and different types of datasets, including indicator data manually filled in by enterprises, operational data transmitted in real time from the data platform, and historical statistical data stored in the system. Standardized data can be organized data after format unification, unit standardization, missing value handling, and outlier preprocessing, ensuring consistency and usability of data when flowing between different levels. Dynamic verification rules can be dynamically adjusted verification standards generated based on historical data of the power plant, which are different from the fixed traditional thresholds and can adapt to changes in the operating conditions of the power plant. Configuration rules can be a set of rules set by the administrator according to the enterprise management needs to control the verification process and early warning methods, including rule activation status, parameter range, early warning form, etc. Verification results can be the judgment results output by the multi-method verification layer after verifying the currently filled data, including whether the data is normal and the abnormal trigger conditions. Early warning information can be the prompt information generated for abnormal data, clearly informing users of the location of the abnormality and related details, facilitating quick problem location. Cause statistics results can be the results formed by classifying and statistically analyzing the error causes of abnormal data, including the frequency of occurrence and scope of each cause. Deviation analysis reports are comprehensive reports that analyze the characteristics of data entry deviations and weak points in quality based on abnormal data and statistical analysis of causes, providing a basis for data quality optimization.

[0023] As an example, the components and connections of the system in the foregoing embodiment can be shown in the following table:

[0024] The following detailed embodiment further illustrates the invention: A large thermal power group owns multiple coal-fired power plants with different unit models and significantly different fuel structures. To meet the national carbon accounting data quality control requirements and adapt to the personalized management needs of each power plant, the group deployed the aforementioned intelligent verification system. The overall operation process of the system revolves around a closed loop of "data access - learning and training - rule configuration - multi-method verification - anomaly handling - result output," with each level working collaboratively: First, in the data access phase, the data access layer collects multi-source data through multiple channels: receiving data submitted by each power plant through designated reporting ports, including core indicators such as the lower heating value of coal fed into the furnace, carbon content, and power generation; accessing real-time data from the group's data platform through a dedicated interface, covering dynamic operating data such as unit output and real-time fuel consumption; and synchronizing historical data from each power plant for the past 5 years from the system's historical database, including annual carbon emission accounting data and fuel index statistics. After data collection, the data access layer standardizes the data, unifying data formats, standardizing indicator units, and handling missing items and obvious formatting errors. Standardized data is then transmitted to both the "One Plant, One Policy" learning and training layer and the multi-method verification layer. Next, the "One Plant, One Policy" learning and training phase begins. This layer receives historical data from the standardized data and conducts learning and training according to a pre-defined logic: first, it performs a comprehensive study of historical data from all coal-fired power plants to grasp the general distribution characteristics and correlation patterns of key indicators for similar units within the industry; then, for each independent power plant, it focuses on its own historical data over the past five years, deeply analyzing the fluctuation range, trends, and internal correlations of key indicators to construct a plant-specific data model. Based on the constructed data analysis model, statistical methods are used to calculate reasonable confidence intervals, forming a dynamic verification range that includes dynamic verification rules. This dynamic verification range is then output to the multi-method verification layer, providing personalized standards for accurate verification. Simultaneously, administrators set rules through the custom rule configuration layer: issuing configuration instructions based on the group's overall management requirements and the specific circumstances of each power plant. After receiving the instruction, the custom rule configuration layer generates the corresponding configuration rules, clearly enabling three methods: threshold verification, historical data verification, and real-time carbon emission calculation verification. It sets the parameter range for each verification method, such as the historical data comparison period and the allowable deviation ratio for real-time verification. It determines that the early warning information will be pushed through a system pop-up window and SMS notification, and sends these configuration rules to the multi-method verification layer.The multi-method verification layer receives the currently submitted data from the data access layer, the dynamic verification rules output by the "one plant, one policy" learning and training layer, and the configuration rules issued by the custom rule configuration layer. It then initiates collaborative work across three verification methods: threshold verification, which assesses the reasonableness of the currently submitted core fuel indicators based on the configuration and dynamic verification rules; historical data verification, which compares the currently submitted data with historical data for a set period by calling the confidence interval in the dynamic verification rules; and real-time carbon emission calculation verification, which calculates predicted values ​​based on real-time data using a correlation prediction model and compares them with the currently submitted monthly accounting data. Each of the three verification methods outputs its own judgment result, which is then combined to form the final verification result, which is transmitted to the anomaly warning and cause statistics layer. After receiving the verification results, the anomaly warning and cause statistics layer processes the data marked as abnormal: The automatic warning unit generates warning information containing the name of the abnormal data item, the power plant to which it belongs, the triggered verification rule, and the degree of deviation, according to the push method in the configuration rules, and pushes it to the administrator of the corresponding power plant and the group's environmental management department; The anomaly cause statistics unit matches possible causes of abnormal data from preset error cause categories (such as data entry errors, fuel quality fluctuations, etc.), and also supports administrators to manually supplement special causes not preset, and counts the frequency of occurrence and related indicators of each cause to form cause statistics results; The deviation analysis unit calculates the reporting deviation rate and frequency of anomalies of each core indicator based on the abnormal data and cause statistics results, analyzes the distribution ratio of different error causes, and generates a detailed data deviation analysis report. Finally, the result output layer receives the warning information, cause statistics results, and deviation analysis report output by the anomaly warning and cause statistics layer, and displays the distribution of abnormal data, cause statistics ledger, etc. in the form of charts through a visual interface, while also providing report generation and data export functions, supporting users to export relevant results to common document formats for easy archiving and subsequent analysis.

[0025] The beneficial effects of one of the embodiments in this specification include at least the following: by integrating multiple verification mechanisms and introducing adaptive learning capabilities, the poor adaptability of general rules in the prior art is effectively overcome. It can dynamically generate personalized verification rules based on the historical data of each power plant, achieving accurate adaptation to the data characteristics of different power plants and significantly improving the accuracy of abnormal data identification. Simultaneously, by integrating real-time calculation and multi-rule collaborative verification, the timeliness and coverage depth of the verification are enhanced. The system also provides highly flexible custom configuration capabilities and closed-loop anomaly analysis and management functions, making the carbon emission data quality control process more comprehensive, accurate, flexible, and traceable, thereby greatly improving the reliability, efficiency, and intelligence level of carbon emission data management in thermal power plants.

[0026] In some embodiments, the data access layer includes a multi-source data integration module, which is used to collect data, real-time data and historical data through at least one of the following methods: application programming interface, database synchronization, direct connection to IoT devices or cloud platform data subscription.

[0027] The multi-source data integration module is a core functional module in the data access layer specifically responsible for collecting various types of data. It is adaptable to multiple data acquisition methods and is a key unit ensuring the comprehensiveness of the system's data sources. The Application Programming Interface (API) provides a standardized interface for data interaction between different systems. Through predefined rules and protocols, it enables fast and secure data transmission, suitable for cross-platform and cross-system data acquisition. Database synchronization allows data scattered across different databases to be synchronized and updated according to preset frequencies or trigger conditions, ensuring that the data acquired by the system is consistent with the source database data. Direct IoT device connection allows direct connection to IoT sensors and monitoring devices installed on the production site, enabling real-time acquisition of device operating data, environmental parameters, etc., with low data transmission latency and strong real-time performance. Cloud platform data subscription allows users to subscribe to data services provided by a cloud service platform and periodically obtain data stored and processed in the cloud platform, suitable for enterprises that have deployed cloud-based management systems.

[0028] The following detailed embodiment further illustrates the invention: In the aforementioned intelligent verification system for thermal power groups, the multi-source data integration module of the data access layer employs a combination of various acquisition methods to ensure comprehensive acquisition and efficient transmission of multi-source data. For reported data, each power plant inputs relevant indicators through the group's unified carbon emission data reporting platform. The multi-source data integration module establishes a connection with this reporting platform through an application programming interface (API), setting the data synchronization frequency to once per hour. It automatically captures the reported data submitted by each power plant, including indicators such as the lower heating value of coal fed into the furnace, carbon content, power generation, and pollutant emissions, ensuring timely acquisition of the latest reported information. For real-time data, the group's data platform aggregates dynamic data such as unit operation and fuel consumption from each power plant. The multi-source data integration module establishes a connection with the core database of the data platform through database synchronization, adopting an incremental synchronization mechanism to collect only real-time data that has changed since the last synchronization, including real-time unit output, real-time coal consumption, and boiler operating parameters. This ensures data real-time performance while reducing data transmission bandwidth usage. For real-time data from key production processes, such as coal sampling and testing data, the multi-source data integration module directly connects to IoT sensors and data acquisition terminals installed at sampling points and testing equipment via IoT devices. Data is transmitted using 5G networks, enabling real-time uploading of testing data and ensuring zero-delay data acquisition, providing accurate data source support for real-time verification. For historical data resources deployed on the group's cloud platform, such as annual carbon emission accounting reports and industry benchmarking data, the multi-source data integration module subscribes to relevant data topics through the cloud platform's data subscription service. The cloud platform automatically pushes updated data at set intervals, which the multi-source data integration module receives and stores, enriching the system's historical data reserves. The multi-source data integration module summarizes the reported data, real-time data, and historical data collected through the above methods, then performs standardization processing, unifying data formats, standardizing indicator names and units, and eliminating obviously invalid data. After forming standardized data, it outputs it to the "one plant, one policy" learning and training layer and the multi-method verification layer according to system requirements, providing high-quality data input for subsequent learning, training, and data verification.

[0029] The beneficial effects of this embodiment include at least the following: by combining multiple data acquisition methods, the system can comprehensively cover carbon emission-related data of different types and sources, solving the problems of incomplete data coverage and insufficient real-time performance of traditional single acquisition methods; the flexible adaptation of application programming interfaces, database synchronization, direct connection to IoT devices, cloud platform data subscription, and other methods meets the acquisition needs of different data scenarios, improving the flexibility and adaptability of data acquisition; the adoption of incremental synchronization, real-time transmission, and other mechanisms ensures data timeliness while reducing system resource consumption, laying a solid data foundation for the efficient operation of the entire intelligent verification system.

[0030] In some embodiments, the one-plant-one-policy learning training layer includes: a power plant data learning unit, used to learn from the historical data of all coal-fired power plants and construct a coal-fired power plant data model; a single-plant data learning unit, used to independently learn from the historical data of a single coal-fired power plant and construct a single-plant-specific data model; and a dynamic range generation unit, connecting the power plant data learning unit and the single-plant data learning unit, used to calculate a confidence interval based on the coal-fired power plant data model and the single-plant-specific data model according to a preset confidence level, so as to generate dynamic verification rules.

[0031] In some implementations, the dynamic range generation unit is preset with a confidence level of 90%, 95%, or 99%.

[0032] In some implementations, the dynamic range generation unit generates a dynamic verification rule that is a verification range determined by adding or subtracting N times the standard deviation of the historical data, where N is a preset positive integer.

[0033] The power plant data learning unit, within the "one plant, one policy" learning and training layer, is a functional unit responsible for analyzing historical data from all coal-fired power plants in the industry. Its core function is to grasp the general patterns in data from similar power plants, providing an industry reference benchmark for single-plant data learning. The single-plant data learning unit, within the "one plant, one policy" learning and training layer, is a functional unit that performs specialized analysis of historical data from a single power plant, focusing on the plant's own data characteristics and building a personalized data model. The dynamic range generation unit connects the two learning units, calculating confidence intervals and generating dynamic verification rules based on the data analysis model. It is the core of achieving dynamic adaptation for "one plant, one policy." The coal-fired power plant data model is a model built based on historical data from all coal-fired power plants, reflecting the general data characteristics and relationships in the industry, including the industry distribution range and typical fluctuation patterns of key indicators. The single-plant-specific data model is a model built based on historical data from a single power plant, reflecting the unique data characteristics of that power plant, including the personalized fluctuation range and trends of its key indicators. Confidence level is a statistical parameter used to determine confidence intervals, representing the probability that data falls within that interval. Commonly used values ​​include 90%, 95%, and 99%. A higher confidence level results in a wider interval and greater tolerance for error. Confidence intervals can be derived from historical data statistical analysis, encompassing the range of data's high probability distribution, and serve as a dynamic standard for judging the reasonableness of currently submitted data. Standard deviation is a statistical measure describing the dispersion of historical data, reflecting the magnitude of data fluctuation. The calculation method of adding or subtracting N times the standard deviation from the mean can quickly determine the reasonable distribution range of the data.

[0034] The following detailed embodiment further illustrates the invention: In the aforementioned intelligent verification system for thermal power groups, the three units of the "one plant, one policy" learning and training layer work collaboratively to generate dynamic verification rules. First, the power plant data learning unit initiates its work, collecting historical data from all coal-fired power plants under the group for the past year, covering key indicators such as the lower heating value of coal fed into the furnace, carbon content, coal consumption for power generation, and total carbon emissions. Through statistical analysis methods, the industry distribution characteristics of these indicators are analyzed, such as the average carbon content of coal fed into the furnace for different types of power plants and the overall fluctuation range of coal consumption for power generation within the industry. The correlation between various indicators is analyzed, such as the correspondence between power generation and fuel consumption, ultimately constructing a coal-fired power plant data model that reflects the general laws of the industry, providing an industry reference for subsequent single-plant data learning. Subsequently, the single-plant data learning unit conducts independent learning for a specific power plant under the group. This unit acquires historical data from the power plant for the past five years, including core fuel indicators, operating parameters, and carbon emission accounting data for each month of each year, focusing on analyzing the data fluctuation patterns of the power plant itself. For example, by analyzing the changing trends of the lower heating value of coal fed into the power plant in different seasons and the stable range of carbon content during full-load operation of the units, and combining personalized factors such as the power plant's unit model, fuel procurement channels, and operation management mode, unique characteristics behind the data can be extracted to construct a single-plant-specific data model that is only applicable to this power plant, accurately matching the power plant's own operating characteristics. The dynamic range generation unit calls both the coal-fired power plant data model and the single-plant-specific data model, and performs calculations based on preset confidence levels. The administrator sets the confidence level to 95% through the custom rule configuration layer. The dynamic range generation unit calculates the average and standard deviation of each key indicator based on historical data in the single-plant-specific data model, and determines the initial confidence interval using the method of "average plus or minus 2 times the standard deviation". At the same time, referring to the industry general range in the coal-fired power plant data model, the initial interval is reasonably corrected to ensure that the interval conforms to the power plant's own characteristics and does not deviate from the reasonable range of the industry. The final result is a set of dynamic verification rules that include reasonable ranges for each key indicator, such as the dynamic verification range for the lower heating value of the coal fed into the power plant and the dynamic judgment range for carbon content. These dynamic verification rules are then output to the multi-method verification layer to provide a standard for the power plant's personalized data verification. In other implementations, depending on the group's different requirements for the stringency of data verification, the administrator can adjust the confidence level to 90% or 99%. When the confidence level is set to 90%, the confidence interval calculated by the dynamic range generation unit is relatively narrow, and the verification standard is more stringent, suitable for power plants with high data management levels and small data fluctuations. When the confidence level is set to 99%, the confidence interval is relatively wide, with stronger fault tolerance, suitable for power plants with unstable fuel sources and large data fluctuations. In addition, the dynamic range generation unit can also use other calculation methods based on standard deviation, such as the average value plus or minus 1.5 times or 3 times the standard deviation, which can be flexibly adjusted according to the actual data fluctuations of the power plant.

[0035] The beneficial effects of this embodiment include at least the following: Through a two-layer learning mechanism of power plant data learning units and single-plant data learning units, it not only grasps the general laws of the industry but also accurately captures the personalized characteristics of individual power plants, providing a comprehensive basis for the generation of dynamic verification rules; the confidence interval calculated based on confidence level and standard deviation replaces the traditional fixed threshold, achieving dynamic adaptation of "one policy per plant," significantly improving the matching degree between verification rules and the actual situation of power plants; the configurability of confidence level and standard deviation multiples allows dynamic verification rules to flexibly adapt to the operating characteristics of different power plants and the management requirements of the group, effectively solving the problem of poor adaptability of traditional verification rules and improving the accuracy and rationality of data verification.

[0036] In some embodiments, the multi-mode verification layer includes: a threshold verification unit, used to perform threshold verification on core fuel indicators according to configuration rules and dynamic verification rules, and generate a first verification result; a historical data verification unit, used to compare and verify the currently reported data with historical data according to configuration rules and by calling dynamic verification rules, and generate a second verification result; and a real-time carbon emission calculation and verification unit, used to perform calculation and verification based on real-time data through an association prediction model according to configuration rules, and generate a third verification result; wherein, the threshold verification unit, the historical data verification unit, and the real-time carbon emission calculation and verification unit work together, and their outputs together constitute the verification result.

[0037] In some implementations, the historical data verification unit compares data for a period of M that is an integer between 1 and 5 years, and the value of M is set according to the configuration rules from the custom rule configuration layer.

[0038] The threshold verification unit, a functional unit within the multi-method verification layer, judges the reasonableness of core fuel indicators based on preset thresholds. It represents the most basic verification method, quickly filtering out data that clearly exceeds reasonable limits. The historical data verification unit, another functional unit within the multi-method verification layer, judges data consistency and reasonableness by comparing current and historical data, leveraging the regularity of historical data to improve verification accuracy. The real-time carbon emission calculation verification unit, a functional unit within the multi-method verification layer, calculates carbon emission data based on real-time operational data and a correlation prediction model, comparing it with the reported data. It possesses both real-time and predictive capabilities. The first verification result, output by the threshold verification unit after threshold verification of core fuel indicators, indicates whether each indicator exceeds the threshold range. The second verification result, output by the historical data verification unit after comparing current reported data with historical data, indicates data consistency and reasonableness. The third verification result, output by the real-time carbon emission calculation verification unit after calculating using a correlation prediction model and comparing with the reported data, indicates whether the data deviation is within the allowable range. The correlation prediction model is a prediction model built based on the correlation between multiple indicators. It can use real-time operational data to predict reasonable values ​​for carbon emission-related indicators, providing a basis for real-time verification. The historical data period M can be the time span of historical data selected when the historical data verification unit compares data, and can be flexibly set to 1-5 years according to verification needs.

[0039] The invention is further illustrated by a detailed embodiment: In the aforementioned intelligent verification system for thermal power groups, the three verification units of the multi-method verification layer work collaboratively according to configuration rules to complete a comprehensive verification of the currently submitted data. First, the threshold verification unit receives configuration rules from the custom rule configuration layer and dynamic verification rules output by the plant-specific learning and training layer, clarifying the list of core fuel indicators to be verified and the corresponding threshold standards. The thresholds include static basic thresholds and dynamic optimized thresholds. The static basic thresholds are set based on national carbon accounting technical specifications, while the dynamic optimized thresholds are derived from the confidence intervals in the dynamic verification rules. The threshold verification unit checks each submitted core fuel indicator, such as the lower heating value of the coal entering the furnace, carbon content, and air-dry basis total sulfur, to determine whether the submitted value of each indicator is within a reasonable range formed by the combination of the corresponding static basic threshold and dynamic optimized threshold. If an indicator value exceeds any threshold range, it is marked as abnormal, ultimately forming a first verification result containing the verification status of all core fuel indicators. Secondly, the historical data verification unit receives the configuration rules and, according to the historical data period M=3 years set in the rules, retrieves the historical data for the same period of the past 3 years for the corresponding power plant from the system's historical database. Simultaneously, it calls the confidence interval in the dynamic verification rules to perform a double comparison between the currently submitted indicator data and the historical data for the same period of the past 3 years, as well as the corresponding confidence interval: on the one hand, it determines whether the difference between the current data and the historical data for the same period is within a reasonable fluctuation range; on the other hand, it determines whether the current data falls within the dynamic confidence interval. If either comparison result does not meet the requirements, it is marked as abnormal, and a second verification result containing the historical comparison information for each indicator is generated.

[0040] In other implementations, administrators can adjust the historical data period M to 1 year, 2 years, 4 years, or 5 years through a custom rule configuration layer, based on the power plant's data stability and management needs. For example, for newly commissioned power plants with limited historical data, M can be set to 1 year; for power plants with stable operation and abundant historical data, M can be set to 5 years to improve the accuracy of the comparison. Then, the real-time carbon emission calculation and verification unit receives the configuration rules and obtains real-time operating data from the data access layer, including real-time unit output, real-time fuel consumption, and boiler operating parameters. Based on a preset correlation prediction model, which has been trained with a large amount of historical data, a correlation relationship has been established between indicators such as emissions, quotas, power generation coal consumption, and heating coal consumption and real-time operating data. The model is used to predict the carbon emissions for the current month, obtaining predicted carbon emission data. The predicted data is compared with the currently reported monthly carbon emission data, and the deviation ratio is calculated to determine whether the deviation ratio is within the allowable range set by the configuration rules. If the deviation ratio exceeds the allowable range, it is marked as abnormal, and a third verification result is generated. Finally, the multi-method verification layer comprehensively summarizes the results of the first, second, and third verifications. If any verification unit marks a piece of data as abnormal, then that data is ultimately determined to be abnormal; if all verification units determine that the data is normal, then that data is ultimately determined to be normal. The final verification result, encompassing the verification status of all submitted data, is then transmitted to the anomaly warning and cause statistics layer.

[0041] The beneficial effects of this embodiment include at least the following: by working together through threshold verification, historical data verification, and real-time carbon emission calculation verification, multi-dimensional and full-scenario verification of carbon emission data is achieved, solving the problems of incomplete coverage and low accuracy of traditional single verification methods; the configurability of the historical data period M enables historical data verification to adapt to the historical data accumulation and management needs of different power plants; the comprehensive judgment mechanism of the three verification results effectively reduces the possibility of misjudgment and omission that may occur with single verification methods, greatly improves the accuracy and reliability of abnormal data identification, and provides core support for ensuring the quality of carbon emission data.

[0042] In some embodiments, the threshold used by the threshold verification unit includes a static basic threshold and a dynamic optimization threshold, wherein the dynamic optimization threshold is obtained according to the dynamic verification rules.

[0043] Static baseline thresholds, which are fixed thresholds set based on national carbon emission accounting technical specifications, industry standards, or unified management requirements of the group, serve as the basic standard for ensuring data compliance and possess universality and stability. Dynamic optimization thresholds, on the other hand, are dynamic thresholds determined based on dynamic verification rules generated by the plant-specific learning and training layer. They can reflect the personalized data characteristics of individual power plants and are targeted and flexible.

[0044] The following detailed embodiment further illustrates the invention: In the aforementioned intelligent verification system for thermal power groups, the threshold verification unit employs a combination of static basic thresholds and dynamic optimized thresholds to accurately verify core fuel indicators. The static basic thresholds are set based on current national carbon emission accounting technical specifications and relevant industry standards, while also considering the group's unified management requirements, setting a fixed and reasonable range for each core fuel indicator. For example, according to national regulations on the lower heating value of coal fed into the furnace, and considering the overall fuel types of the group's power plants, the static basic threshold for the lower heating value of coal fed into the furnace is set at 18MJ / kg-26MJ / kg. This threshold applies to all power plants in the group and serves as the basic standard for judging data compliance, and will not be adjusted due to changes in the situation of individual power plants. The dynamic optimized thresholds are derived from the dynamic verification rules output by the "one plant, one policy" learning and training layer, and are set individually for each core fuel indicator of each power plant. Taking the carbon content of coal fed into a power plant as an example, the dynamic range generation unit calculates a confidence interval of 65%-72% based on the power plant's historical carbon content data over the past five years at a 95% confidence level. This confidence interval is the dynamic optimization threshold for the carbon content of coal fed into the power plant. This threshold fully reflects the power plant's unique factors such as fuel procurement channels and unit combustion characteristics, and can accurately adapt to the actual data characteristics of the power plant. The threshold verification unit refers to both the static basic threshold and the dynamic optimization threshold during verification. For the current reported data of a certain core fuel indicator, if the data is within both the static basic threshold range and the dynamic optimization threshold range, the indicator data is considered normal. If the data exceeds the static basic threshold range, regardless of whether it is within the dynamic optimization threshold range, it is directly judged as abnormal because this violates national regulations and group-wide requirements. If the data is within the static basic threshold range but exceeds the dynamic optimization threshold range, it is marked as potentially abnormal and requires further judgment based on the results of historical data verification and real-time carbon emission calculation verification. For example, a power plant reported that the lower heating value of the coal fed into the furnace was 25 MJ / kg. This value is within the static basic threshold range of 18 MJ / kg-26 MJ / kg, but exceeds the dynamic optimization threshold of 20 MJ / kg-24 MJ / kg corresponding to the power plant. The threshold verification unit marked it as a suspected anomaly, and the other two verification units further verified the rationality of the data.

[0045] The beneficial effects of this embodiment include at least the following: static basic thresholds ensure data compliance and industry uniformity, preventing enterprise data from deviating from national standards and group management requirements; dynamically optimized thresholds adapt to the personalized characteristics of individual power plants, improving the pertinence and accuracy of threshold verification; the combined verification method ensures that the data conforms to unified standards while taking into account the individual differences of power plants, effectively solving the problems of poor adaptability and insufficient accuracy of traditional single fixed thresholds, and significantly improving the comprehensiveness and reliability of threshold verification.

[0046] In some embodiments, the core fuel indicators include at least one of the following: lower heating value of coal fed into the furnace, total water content of coal fed into the furnace, internal water content, total sulfur content on an air-dried basis, carbon content on a comprehensive sample basis, and hydrogen content on an air-dried basis.

[0047] The lower heating value of coal fed into the furnace is the heat value after deducting the latent heat of condensation of water vapor from the heat released after the complete combustion of a unit mass of coal fed into the furnace. It is a key energy indicator affecting carbon emission accounting. Total water content in coal fed into the furnace is the total content of free and combined water in the coal, directly affecting the combustion efficiency and calorific value calculation, and thus the carbon emission accounting results. Internal water content is the moisture content adsorbed internally by the coal under certain conditions, an important indicator reflecting coal quality characteristics, and has a significant impact on the combustion process and carbon emission calculations. Air-dried total sulfur content is the total sulfur content of coal under air-dried conditions. Sulfur combustion produces pollutants and affects the accuracy of carbon emission-related indicators. As-received carbon content is the mass fraction of carbon in coal under as-received conditions, a core indicator for calculating carbon emissions, directly determining the accuracy of carbon emission accounting results. Air-dried hydrogen content is the mass fraction of hydrogen in coal under air-dried conditions. Hydrogen combustion affects heat release and moisture generation, indirectly affecting carbon emission calculations.

[0048] The invention is further illustrated by a detailed embodiment below: In the aforementioned intelligent verification system for thermal power groups, core fuel indicators cover key indicators such as the lower heating value of coal fed into the furnace, total water content, internal water content, air-dried total sulfur content, comprehensive sample-based carbon content, and air-dried hydrogen content. A multi-method verification layer conducts comprehensive verification of these indicators: The system first clarifies the collection requirements and verification standards for each core fuel indicator to ensure the completeness of data collection and the relevance of verification. For the lower heating value of coal fed into the furnace, power plants are required to report monthly average values ​​based on daily sampling and testing data. Verification requires combining static basic thresholds and dynamic optimized thresholds, comparing with historical data from the same period over the past three years, and predicting a reasonable range using real-time fuel consumption and unit output data to comprehensively judge the reasonableness of the reported data. For the total water content and internal water content of coal fed into the furnace, the correlation between their effects and calorific value is emphasized, as excessive moisture content reduces the actual calorific value of the coal. During verification, the threshold verification unit determines whether the reported value is within the corresponding dual threshold range; the historical data verification unit analyzes the historical fluctuation pattern of moisture content to determine whether the current data conforms to the power plant's own fluctuation characteristics; the real-time carbon emission calculation verification unit combines real-time combustion efficiency data to predict a reasonable range of moisture content and compares it with the reported data. For the air-dry basis total sulfur index, on the one hand, it verifies whether it meets national environmental protection standards and power plant fuel procurement requirements, and on the other hand, it analyzes its correlation with pollutant emission data. If the reported total sulfur content data is abnormal, it needs to be cross-validated with the reported pollutant emission data to ensure data consistency. The comprehensive sample-based carbon element content is the core indicator for carbon emission accounting, and the most stringent standards are adopted during verification: the threshold verification unit strictly compares the static basic threshold and the dynamic optimized threshold; the historical data verification unit selects historical data from the past 5 years for in-depth comparison and analyzes the annual, quarterly, and monthly trends; the real-time carbon emission calculation verification unit accurately predicts the reasonable range of carbon element content based on real-time fuel consumption, unit output, combustion efficiency, and other multi-dimensional data. If the deviation between the reported data and the predicted data exceeds the allowable range, it is immediately marked as abnormal. For the hydrogen index based on air-dried fuels, the focus is on analyzing its correlation with calorific value and carbon content. A multi-indicator collaborative verification method is used to determine the rationality of the reported data. For example, if the reported hydrogen content is abnormally high, while the calorific value and carbon content are low, a comprehensive assessment is needed to determine if the data contains problems. The verification results of all core fuel indicators are incorporated into the first, second, and third verification results. Finally, a comprehensive verification result is formed through the combined summary of multiple verification methods, ensuring no key indicators are omitted and no data quality issues arise.

[0049] The beneficial effects of this embodiment include at least the following: it accurately covers the core fuel indicators that affect the carbon emission accounting of thermal power plants, ensuring that the verification work directly addresses the key issues and avoids data quality problems caused by incomplete indicator coverage; it formulates differentiated verification logic based on the characteristics of each core fuel indicator and its impact on carbon emission accounting, improving the relevance and accuracy of the verification; and it ensures the consistency and rationality of the data through multi-indicator collaborative verification and correlation analysis, providing key support for the accuracy of carbon emission accounting results, while meeting the national requirements for the control of core carbon data indicators.

[0050] In some embodiments, the real-time carbon emission calculation and verification unit includes: a prediction module, used to calculate carbon emission prediction data based on real-time data through a correlation prediction model that associates emissions, quotas, power generation coal consumption and heating coal consumption; and a comparison module, connected to the prediction module, used to compare the carbon emission prediction data with the monthly carbon emission data in the currently reported data, and determine whether it is abnormal according to the deviation ratio range set in the configuration rules.

[0051] The prediction module, within the real-time carbon emission calculation and verification unit, is responsible for calculating predicted carbon emission data using real-time data and associated prediction models. It is the core of real-time calculation and verification. The comparison module, also within the real-time carbon emission calculation and verification unit, is responsible for comparing predicted carbon emission data with reported data and determining whether deviations are compliant. It is crucial for outputting the third verification result. Predicted carbon emission data refers to the predicted values ​​calculated by the prediction module based on real-time operational data and associated prediction models, reflecting the reasonable level of carbon emissions for the current period. Monthly carbon emission accounting data refers to the accounting values ​​reported by the power plant after summarizing relevant monthly data according to national carbon accounting methods, reflecting the actual carbon emissions for the month. The deviation ratio range, set by the administrator through configuration rules, represents the maximum allowable deviation ratio between predicted carbon emission data and monthly carbon emission accounting data, serving as an important criterion for determining whether the data is abnormal.

[0052] The invention is further illustrated by a detailed embodiment: In the aforementioned intelligent verification system for thermal power groups, the prediction module and comparison module of the real-time carbon emission calculation and verification unit work together to complete real-time calculation and verification. First, the prediction module receives real-time data transmitted from the data access layer, including key operational data such as real-time unit output, real-time fuel consumption, boiler combustion efficiency, and heating load. This data is collected in real-time through direct connection to IoT devices and database synchronization to ensure the timeliness and accuracy of the data. The prediction module has a built-in correlation prediction model, which has been trained and optimized using nearly three years of historical data from all power plants under the group, establishing a complex correlation between four major indicators—emissions, quotas, power generation coal consumption, and heating coal consumption—and real-time operational data. The model can predict power generation coal consumption and heating coal consumption based on real-time fuel consumption, and, combined with unit output and combustion efficiency data, further predict the total carbon emissions and quotas for the month, ultimately generating complete carbon emission prediction data, including key data such as monthly carbon emission prediction values ​​and quota prediction values. Subsequently, the comparison module establishes a data connection with the prediction module and receives the carbon emission prediction data output by the prediction module. Simultaneously, the comparison module extracts monthly carbon emission data from the currently submitted data, including the total monthly carbon emissions reported by the power plant and the calculated quota value. The comparison module calculates the deviation ratio between the predicted carbon emissions data and the monthly carbon emission data according to the deviation ratio range set in the configuration rules (e.g., ±5%). The calculation method is: (Monthly carbon emission data - Predicted carbon emissions data) / Predicted carbon emissions data × 100%. If the calculated deviation ratio is within the set ±5% range, the monthly carbon emission data is considered normal; if the deviation ratio exceeds ±5%, the direction and possible causes of the deviation are further analyzed, and the data is marked as abnormal. For example, if a power plant reports a total monthly carbon emission of 1.2 million tons, and the predicted carbon emissions data calculated by the prediction module is 1.12 million tons, the deviation ratio is 7.14%, exceeding the allowable range of ±5%. The comparison module marks this data as abnormal, records the deviation value and ratio, generates a third verification result, and transmits it to the multi-method verification layer for comprehensive judgment.

[0053] The beneficial effects of this embodiment include at least the following: the correlation prediction model is built based on the correlation of multiple indicators and trained and optimized with a large amount of historical data, which can accurately predict carbon emission-related data and provide a reliable basis for real-time verification; the collaborative work of the prediction module and the comparison module realizes the full-process automation of "real-time prediction - data comparison - anomaly judgment", improving the real-time performance and efficiency of verification; the configurability of the deviation ratio range enables the verification standard to adapt to the operating characteristics of different power plants and the management requirements of the group, effectively identifying the reported data that exceeds the reasonable fluctuation range, ensuring the accuracy and rationality of monthly carbon emission data, and providing reliable data support for corporate carbon trading and quota management.

[0054] In some embodiments, the custom rule configuration layer includes a rule management unit, which performs at least one of the following operations to generate configuration rules according to configuration instructions: enabling or disabling threshold verification, historical data verification, or real-time carbon emission calculation verification; modifying the threshold range, confidence interval confidence level, or deviation ratio range; adding special verification rules for specific power plants or indicators or deleting existing rules; and setting the method for pushing early warning information.

[0055] The rule management unit is a core functional unit in the custom rule configuration layer responsible for receiving administrator configuration commands, executing rule adjustment operations, and generating configuration rules. It is key to realizing custom rule configuration within the system. Configuration commands are instructions issued by administrators through the system interface to adjust verification rules based on enterprise management needs, including rule start / stop, parameter modification, rule addition / deletion, and alert method settings. Specialized verification rules are personalized verification rules tailored to specific power plants or indicators, suitable for handling data verification needs under special scenarios or requirements. The method of pushing alert information allows the system to push abnormal alert information to users in various forms, including system pop-ups, SMS notifications, and email alerts.

[0056] The following detailed embodiment further illustrates the present invention: In the aforementioned intelligent verification system for thermal power groups, the rule management unit of the custom rule configuration layer provides administrators with full-dimensional rule configuration functions. The specific operation process is as follows: The administrator logs into the custom rule configuration layer through the system's visual configuration interface and issues various configuration commands according to the group's management requirements and the actual situation of each power plant. The rule management unit receives and executes the corresponding operations. Regarding rule activation and deactivation, for a newly commissioned power plant with limited historical data accumulation and limited reference value for historical data verification, the administrator issues a configuration command to "deactivate historical data verification, and enable threshold verification and real-time carbon emission calculation verification." The rule management unit executes this command, generates the corresponding configuration rules, and clarifies that the power plant only enables two verification methods, ensuring that the verification work conforms to the actual data situation of the power plant. Regarding parameter modifications, the group needed to adjust the static baseline threshold ranges of some indicators based on updates to the national carbon accounting technical specifications. The administrator issued an instruction to "adjust the static baseline threshold for the lower heating value of coal fed into the furnace to 17MJ / kg-27MJ / kg." Simultaneously, for a power plant with significant data fluctuations, an instruction to "adjust the confidence interval to 99%" was issued. For real-time verification, an instruction to "adjust the deviation ratio range to ±8%" was issued. The rule management unit executed these instructions, updating the corresponding parameter settings and integrating them into the configuration rules. Regarding rule additions and deletions, a power plant under the group had a unique fuel procurement channel, resulting in significantly different fluctuation patterns in the carbon content of its coal fed into the furnace compared to other power plants. The administrator issued an instruction to "add a special verification rule for the carbon content of coal fed into the furnace for this power plant, setting a special dynamic optimization threshold calculation logic." Simultaneously, for a certain outdated indicator, an instruction to "delete all verification rules corresponding to this indicator" was issued. The rule management unit executed the instructions, adding special rules and deleting invalid rules, ensuring the relevance and simplicity of the rule system. Regarding the setting of early warning methods, the administrator issues configuration instructions to "push system pop-ups and SMS notifications to power plant environmental protection specialists, and push email reminders and system announcements to the group's environmental management department" based on the responsibilities of different users. The rule management unit executes the instructions, clarifies the early warning information push methods for different users, generates complete configuration rules, and distributes them to the multi-method verification layer and the abnormal early warning and cause statistics layer to guide them in carrying out verification and early warning work.

[0057] The beneficial effects of this embodiment include at least the following: the rule management unit supports full-dimensional rule configuration operations, covering rule start / stop, parameter modification, rule addition / deletion, and alert method settings, greatly improving the system's flexibility and adaptability; administrators can issue configuration commands through a visual interface without modifying system code, reducing operational difficulty and maintenance costs; the specialized rule configuration function for specific power plants or indicators can accurately adapt to the personalized management needs of enterprises, solving the problem of traditional system rules being rigid and unable to respond to special scenario needs; and the personalized setting of alert methods ensures that alert information can be delivered to relevant responsible persons in a timely and accurate manner, improving the efficiency of abnormal data processing.

[0058] In some embodiments, the anomaly warning and cause statistics layer includes: an automatic warning unit, used to generate and push warning information containing anomaly data item identifier, power plant identifier, trigger verification rule identifier, and deviation degree based on verification results and configuration rules; an anomaly cause statistics unit, used to match or receive manually supplemented anomaly causes from preset error cause categories, count anomaly causes, and form a cause statistics ledger; and a deviation analysis unit, connected to the anomaly cause statistics unit, used to calculate the reporting deviation rate and anomaly occurrence frequency of core fuel indicators, analyze the distribution ratio of anomaly causes, and generate a reporting data deviation analysis report containing the reporting deviation rate, anomaly occurrence frequency, and distribution ratio.

[0059] In some implementations, the error cause classification preset by the anomaly cause statistics unit includes data entry errors and fuel quality fluctuations.

[0060] In some implementations, the anomaly cause statistics unit is also used to identify and automatically classify the causes of the input text through natural language processing.

[0061] In some implementations, the anomaly warning and cause statistics layer also includes an AI-assisted attribution unit, which automatically recommends possible causes of anomalies based on anomaly data characteristics, power plant historical problems, fuel and unit operation data.

[0062] The automatic early warning unit, within the anomaly early warning and cause statistics layer, is responsible for generating and pushing early warning information. Its core function is to promptly inform users of abnormal data situations, facilitating rapid response. The anomaly cause statistics unit, also within the anomaly early warning and cause statistics layer, is responsible for matching, supplementing, and statistically analyzing the causes of anomalies, forming the foundation for anomaly tracing. The deviation analysis unit, within the anomaly early warning and cause statistics layer, is responsible for calculating deviation indicators, analyzing cause distribution, and generating deviation analysis reports, providing a basis for data quality optimization. Anomaly data item identifiers are used to clearly identify the specific indicator names of abnormal data, such as "carbon content of coal fed into the furnace." The power plant identifier identifies the name or number of the power plant to which the abnormal data belongs. The trigger verification rule identifier identifies the specific verification rule that caused the data anomaly, such as "threshold verification (dynamically optimized threshold)." The deviation degree indicates the degree of deviation of abnormal data from a reasonable range, including the deviation value and deviation ratio. The preset error cause classification provides a reference for quickly matching anomaly causes, based on common anomaly cause categories pre-defined by the system. The cause statistics ledger is a statistical table that records the frequency of occurrence, involved indicators, and power plant information for various abnormal causes, facilitating traceability and analysis. The reporting deviation rate is the proportion of deviation between the reported data and the reasonable data for core fuel indicators, a key indicator reflecting the quality of data reporting. The frequency of anomalies is the number of times abnormal data occurs for a specific indicator or power plant within a certain period. Natural language processing (NLP) is a technology that uses computer technology to recognize, understand, and process human natural language, enabling automatic classification of causes in unstructured text. The AI-assisted attribution unit is a functional unit that automatically recommends possible causes of abnormal data based on artificial intelligence technology, improving the efficiency and accuracy of anomaly attribution.

[0063] The following detailed embodiment further illustrates the present invention: In the aforementioned intelligent verification system for thermal power groups, multiple units in the anomaly early warning and cause statistics layer work collaboratively to complete the early warning, cause statistics, and deviation analysis of abnormal data. First, the automatic early warning unit receives the verification results and configuration rules output by the multi-mode verification layer and processes the abnormal data in the verification results. For each abnormal data item, the automatic early warning unit extracts the abnormal data item identifier, such as "basic carbon element content of comprehensive sample"; the power plant identifier, such as "XX Power Plant #1 Unit"; the trigger verification rule identifier, such as "historical data verification (comparison of 3 years of historical data)"; and calculates the degree of deviation, including the deviation value and deviation ratio. According to the early warning method set in the configuration rules, system pop-ups and SMS notifications are pushed to the power plant's environmental protection specialists, and email reminders and system announcements are pushed to the group's environmental management department. The early warning information clearly presents all the above details, ensuring that relevant responsible persons can quickly grasp the abnormal situation. Next, the anomaly cause statistics unit presets error cause classifications, including common categories such as data entry errors, fuel quality fluctuations, detection equipment failures, and sudden changes in operating conditions. For abnormal data marked by the automatic early warning unit, the anomaly cause statistics unit matches possible anomaly causes from preset categories based on the characteristics of the abnormal data and the triggered verification rules. For example, if the low calorific value of coal fed into a power plant fluctuates significantly in a short period of time, and other indicators show no obvious abnormalities, the cause is matched as "fuel quality fluctuation"; if the data significantly exceeds the reasonable range and is irregular, the cause is matched as "data entry error". Simultaneously, administrators can manually add special causes not preset, such as "impurities mixed in during fuel transportation". For text causes entered by the administrator, the anomaly cause statistics unit uses natural language processing technology to identify and automatically classify them, ensuring the consistency of cause classification. Finally, the frequency of occurrence of various error causes, the indicators involved, and the power plants are statistically analyzed to form a detailed cause statistics ledger.

[0064] In some implementations, the anomaly warning and cause statistics layer also includes an AI-assisted attribution unit. This unit, based on multi-dimensional information such as the deviation characteristics of abnormal data, historical problem records of the power plant, fuel procurement data, and unit operation data, automatically recommends possible causes of anomalies using a trained AI model. For example, combining recent fuel supplier change records and current fuel indicator anomalies, the AI-assisted attribution unit recommends "fuel quality fluctuation" as the most likely cause and provides a confidence level, which administrators can directly confirm or modify, significantly improving anomaly attribution efficiency. Finally, the deviation analysis unit establishes a data connection with the anomaly cause statistics unit, receiving the cause statistics ledger and detailed anomaly data. It calculates the reporting deviation rate for each core fuel indicator, i.e., (abnormal reported data - reasonable data) / reasonable data × 100%; it also statistically analyzes the frequency of anomalies for each indicator and the distribution percentage of different anomaly causes, such as data entry errors accounting for 30% and fuel quality fluctuations accounting for 45%. Based on these analysis results, a data deviation analysis report is generated, which includes data entry deviation rate ranking, anomaly frequency statistics, and cause distribution pie charts. This identifies weak links in data quality and provides a basis for enterprises to formulate targeted optimization measures.

[0065] The beneficial effects of this embodiment include at least the following: the automatic early warning unit can quickly and accurately push abnormal information, ensuring that relevant responsible persons respond in a timely manner and improving the efficiency of abnormal handling; the abnormal cause statistics unit, through preset classification, manual supplementation and natural language processing technology, realizes comprehensive statistics and standardized management of abnormal causes, and combined with the AI-assisted attribution unit, further improves the accuracy and efficiency of cause matching; the deviation analysis report generated by the deviation analysis unit clearly presents the current status of data quality and weak links, providing a clear direction for enterprise data quality optimization; the entire abnormal early warning and cause statistics process forms a closed loop of "early warning-attribution-analysis", realizing the traceability and analyzability of abnormal data, helping enterprises to solve data quality problems from the root and continuously improve the quality of carbon emission data reporting.

[0066] In some embodiments, the result output layer includes a visualization unit and a report generation unit, which are used to display early warning information, cause statistics and deviation analysis reports in a graphical interface or report form, and support data export.

[0067] The visualization unit, located in the results output layer, is responsible for intuitively displaying relevant data and reports in a graphical interface, facilitating quick viewing and understanding by users. The report generation unit, also in the results output layer, is responsible for generating standardized reports from early warning information, cause statistics, and deviation analysis reports, facilitating data retention and sharing. The graphical interface can present data using visual elements such as charts, dashboards, and lists, including bar charts, pie charts, line charts, and data tables. Standardized reports are documents generated in preset formats, containing complete data and analysis results, with a unified structure and style, facilitating easy viewing and archiving. Data export allows users to export the visualization content or generated reports to common document formats, meeting their needs for offline viewing, editing, and retention.

[0068] The following detailed embodiment further illustrates the present invention: In the aforementioned intelligent verification system for thermal power groups, the visualization display unit and report generation unit of the result output layer work together to provide users with intuitive and convenient result presentation and data export services. First, the visualization display unit receives the warning information, cause statistics results, and deviation analysis reports output by the anomaly warning and cause statistics layer, and displays them in a multi-dimensional manner using a graphical interface. For warning information, it is presented in the form of a real-time warning list, clearly listing abnormal data items, the power plant to which they belong, trigger rules, deviation degree, warning time, etc., and supports filtering and sorting by power plant, indicator, warning time, etc. For cause statistics results, a pie chart is used to show the distribution ratio of different abnormal causes, and a bar chart is used to show the frequency of abnormal occurrences for each power plant and indicator, intuitively presenting the concentration trend and distribution characteristics of abnormal causes. For deviation analysis reports, a data dashboard is used to display key indicators such as the ranking of reporting deviation rates of core indicators, the overall data quality pass rate, and the abnormal data processing completion rate, while providing detailed textual analysis content to help users quickly grasp the current status of data quality and optimization directions. Secondly, the report generation unit receives relevant data and analysis results, and generates various reports according to preset standardized formats. These include reports such as the "Abnormal Data Early Warning Report," "Abnormal Cause Statistical Ledger Report," and "Data Deviation Analysis Report." These reports contain complete data details, statistical results, and analysis conclusions, with unified formats and standardized content, meeting the archiving and reporting needs of the group and power plants. Simultaneously, the results output layer supports data export operations. Users can select the content to be exported in the visual interface, or directly select the generated standardized reports, and export them to common document formats such as Excel, PDF, and Word using the system's export function. For example, a power plant administrator can export the power plant's abnormal data early warning report to Excel format for internal data verification and rectification; the group's environmental management department can export the group-wide data deviation analysis report to PDF format for reporting and archiving. The export process is simple and convenient, and data integrity and format consistency are fully guaranteed.

[0069] The beneficial effects of this embodiment include at least the following: the visualization unit uses a graphical interface and multi-dimensional presentation methods, making complex data and analysis results more intuitive and easy to understand, reducing the user's viewing and understanding costs; the standardized reports generated by the report generation unit meet the standardized requirements of enterprise data archiving and reporting; the data export function supports a variety of commonly used document formats, providing users with the convenience of offline viewing, editing, and data retention, improving the system's usability and user experience; the design of the entire result output layer ensures that verification results, early warning information, and analysis reports can be used efficiently, providing strong support for enterprise data quality management and decision-making.

[0070] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this invention. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. An intelligent verification system for enterprise carbon emission data, characterized in that, It includes a data access layer, a one-factory-one-policy learning and training layer, a multi-method verification layer, an anomaly warning and cause statistics layer, and a result output layer connected in sequence, as well as a custom rule configuration layer connected to the multi-method verification layer; The data access layer is used to collect multi-source data including data entered, real-time data and historical data, perform standardization processing to obtain standardized data and output it. The "one factory, one policy" learning and training layer is used to receive historical data from the standardized data, perform learning and training, and generate and output a dynamic verification range containing dynamic verification rules. The custom rule configuration layer is used to receive configuration instructions from the administrator, generate and output configuration rules for controlling the verification process and early warning methods; The multi-mode verification layer is used to receive the currently filled data in the standardized data, the dynamic verification rules and the configuration rules, and to perform collaborative verification through threshold verification, historical data verification and real-time carbon emission calculation verification, and generate and output the verification results. The anomaly warning and cause statistics layer is used to receive the verification results, issue warnings when abnormal data appears in the verification results, determine the abnormal causes associated with the abnormal data, and generate and output warning information, cause statistics results and deviation analysis reports. The result output layer is used to receive the early warning information, cause statistics, and deviation analysis report, and to visualize and export them.

2. The system according to claim 1, characterized in that, The data access layer includes a multi-source data integration module, which is used to collect the reported data, real-time data and historical data through at least one of the following methods: application programming interface, database synchronization, direct connection to IoT devices or cloud platform data subscription.

3. The system according to claim 1, characterized in that, The "one factory, one policy" learning and training layer includes: The power plant data learning unit is used to learn from the historical data of all coal-fired power plants and build a data model of coal-fired power plants. The single-plant data learning unit is used to independently learn from the historical data of a single coal-fired power plant and build a dedicated data model for that plant. The dynamic range generation unit, connected to the power plant data learning unit and the single-plant data learning unit, is used to calculate the confidence interval based on the coal-fired power plant data model and the single-plant dedicated data model, according to a preset confidence level, so as to generate the dynamic verification rule.

4. The system according to claim 1, characterized in that, The multi-mode verification layer includes: The threshold verification unit is used to perform threshold verification on the core fuel index according to the configuration rules and the dynamic verification rules, and generate a first verification result; The historical data verification unit is used to compare and verify the currently entered data with the historical data according to the configuration rules and by calling the dynamic verification rules, and generate a second verification result. The real-time carbon emission calculation and verification unit is used to calculate and verify the emissions based on real-time data and an association prediction model according to the configuration rules, and generate a third verification result. The threshold verification unit, historical data verification unit, and real-time carbon emission calculation verification unit work together, and their outputs together constitute the verification result.

5. The system according to claim 4, characterized in that, The threshold used by the threshold verification unit includes a static basic threshold and a dynamic optimization threshold, and the dynamic optimization threshold is obtained according to the dynamic verification rule.

6. The system according to claim 4, characterized in that, The core fuel indicators include at least one of the following: lower heating value of coal fed into the furnace, total water content of coal fed into the furnace, internal water content, total sulfur content on an air-dried basis, carbon content on a comprehensive sample basis, and hydrogen content on an air-dried basis.

7. The system according to claim 4, characterized in that, The real-time carbon emission calculation and verification unit includes: The prediction module is used to calculate carbon emission prediction data based on the real-time data through a correlation prediction model that links emissions, quotas, coal consumption for power generation, and coal consumption for heating. The comparison module, connected to the prediction module, is used to compare the carbon emission prediction data with the monthly carbon emission data in the currently reported data, and to determine whether there is an anomaly based on the deviation ratio range set in the configuration rules.

8. The system according to claim 1, characterized in that, The custom rule configuration layer includes a rule management unit, which performs at least one of the following operations to generate the configuration rules based on the configuration instructions: Enable or disable threshold verification, historical data verification, or real-time carbon emission calculation verification; Modify the threshold range, confidence interval confidence level, or deviation ratio range; Add new specific verification rules or delete existing rules for specific power plants or indicators; Configure the method for sending early warning information.

9. The system according to claim 1, characterized in that, The anomaly warning and cause statistics layer includes: An automatic early warning unit is used to generate and push early warning information containing an abnormal data item identifier, the power plant identifier, the triggering verification rule identifier, and the degree of deviation based on the verification results and the configuration rules. The anomaly cause statistics unit is used to match or receive manually added anomaly causes from the preset error cause categories, count the anomaly causes, and form a cause statistics ledger. The deviation analysis unit, connected to the anomaly cause statistics unit, is used to calculate the reporting deviation rate and anomaly frequency of core fuel indicators, analyze the distribution ratio of anomaly causes, and generate a reporting data deviation analysis report containing the reporting deviation rate, anomaly frequency, and distribution ratio.

10. The system according to claim 1, characterized in that, The result output layer includes a visualization unit and a report generation unit, which are used to display the warning information, cause statistics and deviation analysis reports in a graphical interface or report form, and support data export.

Images