Substation time sequence data intelligent verification method based on object model mapping

Abstract

The invention discloses an intelligent verification method for time series data of a transformer substation based on object model mapping, belongs to the technical field of intelligent power grids, and aims to solve the problems of rule lag, poor error tolerance and high data dependence in the prior art. S2, multi-source data mapping normalization is carried out; s3, checking the self-adaptive threshold value of a single measuring point; s4, multi-measuring-point correlation constraint verification is carried out; s5, fusing and outputting multi-source verification results; according to the method, through a dual verification mechanism formed by single-measurement-point adaptive threshold verification and multi-measurement-point correlation constraint verification, the physical law and the real-time operation state are combined, so that the complex anomaly can be identified more accurately, and the false alarm rate and the missing report rate of a single method are remarkably reduced.

Description

Technical Field

[0001] This invention belongs to the field of smart grid technology, specifically relating to an intelligent verification method for substation time-series data based on object model mapping. Background Technology

[0002] With the intelligent development of power systems, substations, as core facilities of the power system, are crucial to the stability of the entire power network due to their safe and reliable operation. In modern substations, various intelligent devices and sensors are widely used to collect various operational data, including parameters such as equipment status, temperature, pressure, current, and voltage. These data are usually recorded and stored in the form of time-series data for purposes such as real-time monitoring, fault diagnosis, and operation optimization. However, during the collection and transmission of substation time-series data, errors, loss, or abnormal fluctuations may occur due to factors such as equipment failure, transmission anomalies, data loss, and collection errors. These abnormal data not only affect the real-time monitoring and scheduling decisions of the system but may also lead to misjudgments of equipment status, thereby affecting the safe operation of the power system. Therefore, how to detect and verify these time-series data anomalies in a timely and effective manner has become a key technology in the intelligent management of substations.

[0003] Currently, research on substation time-series data verification mainly focuses on traditional data cleaning and anomaly detection methods. Common techniques include threshold-based anomaly detection, statistical methods, and rule matching. However, these methods still have the following shortcomings and limitations:

[0004] 1. Outdated rules: Traditional threshold settings and rules are based on historical experience and are difficult to adapt to the complex and ever-changing substation operating environment. Furthermore, the rules have poor adaptability when facing different equipment and different operating states.

[0005] Second, poor error tolerance: Traditional methods have poor tolerance for data errors and are prone to misjudging small errors as anomalies, resulting in a large number of false alarms.

[0006] Third, strong data dependence: Most methods rely on a large amount of historical data for modeling. However, after new equipment is put into use in substations, there is a lack of sufficient historical data for verification, making it difficult for these methods to effectively adapt to new equipment and new environments.

[0007] Therefore, there is a need for an intelligent verification method for substation time-series data based on object model mapping to solve the problems of rule lag, poor error tolerance, and strong data dependence in existing technologies. Summary of the Invention

[0008] The purpose of this invention is to provide an intelligent verification method for substation time-series data based on object model mapping, so as to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: an intelligent verification method for substation time-series data based on object model mapping, comprising the following steps:

[0010] S1. Construction of the object model mapping library: Construct an object model mapping library for the substation. The object model mapping library defines a digital twin for each physical device or logical component in the substation and configures its corresponding object model for each digital twin. The object model includes at least an object identifier, a set of static attributes, a set of dynamic measurement points, and a set of association relationships.

[0011] S2. Multi-source data mapping and normalization: Real-time access to the original time-series data stream of the substation, and according to the object model mapping library, the original time-series data is mapped and normalized to the corresponding object model dynamic measurement points to form structured data with unified semantics.

[0012] S3, Single Measurement Point Adaptive Threshold Verification: Based on the dynamic measurement point attributes of the object model, an adaptive dynamic threshold is configured for each measurement point, and the data of a single measurement point is checked for the first level of anomalies based on the threshold.

[0013] S4. Multi-measurement point association constraint verification: Based on the association relationship set of the object model, construct physical or logical constraint rules between associated measurement points, and perform a second-level association verification on the data of multiple associated measurement points based on the constraint rules.

[0014] S5. Multi-source verification result fusion output: Based on the results of the first-level anomaly verification and the second-level correlation verification, the comprehensive confidence of each data point is calculated, and the final verification conclusion and data quality label are output based on the comprehensive confidence.

[0015] It should be noted in the scheme that the specific process of constructing the object model mapping library in step S1 includes:

[0016] S11: Decompose the equipment entities and logical functions of the substation to form a list of equipment objects;

[0017] S12: Create a unique device identifier for each device object in the list and define its static attribute set, which includes device model, rated parameters, installation location and health status baseline;

[0018] S13: Define a dynamic measurement point set for each device object. The dynamic measurement point set includes measurement point ID, measurement point name, physical quantity, data unit, acquisition frequency, and data type.

[0019] S14: Define the set of relationships within a device object and between different device objects. The set of relationships includes topological connection relationships, electrical coupling relationships, and physical law constraint relationships.

[0020] It is further worth noting that, in step S3, the method for configuring and calculating the adaptive dynamic threshold includes:

[0021] Based on the object model mapping library, historical operating data of the target measuring point and its associated load condition data are obtained;

[0022] Based on the historical data, the statistical characteristics of the data distribution of the measuring point within the typical load range were calculated.

[0023] Based on the statistical characteristics of the data distribution and real-time load data, the reasonable threshold range for the measuring point is dynamically calculated. The upper and lower limits of the dynamic threshold are determined by the following model:

[0024]

[0025] In the formula, L and U are the lower and upper limits of the dynamic threshold, respectively. and These are the mean and standard deviation calculated based on the historical data, respectively, where k is a configurable sensitivity coefficient. To associate load parameters The operating condition correction function is used to adaptively adjust the threshold width according to the system operating status.

[0026] Furthermore, it should be noted that in step S4, the second-level association verification specifically involves: for the set of measurement points with strong associations { }, construct its constraint relationship function based on physical laws or business logic. During verification, the deviation between the actual data and the constraint relationship is calculated. :

[0027]

[0028] when deviation Exceeding the preset fault tolerance range If so, it is determined that there is an anomaly in the associated data set.

[0029] In a preferred embodiment, the constraint relationship function F is constructed by establishing a mathematical correlation model among multiple measurement points based on the electrical connection relationships or physical principles defined in the object model correlation set; the mathematical correlation model includes, but is not limited to:

[0030] A summation constraint model is established for the branch currents of the same electrical node based on Kirchhoff's current law.

[0031] A power balance constraint model is established for the same circuit or equipment based on the principle of power conservation.

[0032] A thermal circuit model is established based on the thermodynamic characteristics of the equipment, considering the equipment temperature, ambient temperature, and load current.

[0033] In a preferred embodiment, the calculation model for the comprehensive confidence level C in step S5 is as follows:

[0034]

[0035] In the formula, The score for the first level of anomaly detection is negatively correlated with the degree to which the data point deviates from its adaptive dynamic threshold. The score for the second-level association check is the deviation of the data point from its association constraint group. Negative correlation; and They are respectively and The weights, and + =1, the weight can be configured according to the measurement point type and correlation strength.

[0036] In a preferred embodiment, the specific process of outputting the verification conclusion based on the comprehensive confidence level C in step S5 is as follows: a high confidence threshold is preset. and a low confidence threshold ;

[0037] like If the data is normal, the label "Normal" will be output.

[0038] like If the data is deemed suspicious, a "warning" label will be output and a manual review will be requested.

[0039] like If the data is abnormal, an "Abnormal" label will be output and an alarm will be triggered.

[0040] In a preferred embodiment, the method further includes a model self-learning and optimization step S6:

[0041] The conclusions of the first-level anomaly check and the second-level correlation check, as well as the results of manual review from external input, are used as feedback data.

[0042] Using the feedback data, the key parameters in the object model mapping library are optimized through incremental learning. The parameter update mechanism is represented by the following general formula:

[0043]

[0044] In the formula, Represents the historical parameters before optimization. This represents the new parameter estimate calculated from the feedback data. This represents the optimized update parameters. The learning rate is used to control the magnitude and speed of parameter updates.

[0045] Compared with existing technologies, the intelligent verification method for substation time-series data based on object model mapping provided by this invention has at least the following beneficial effects:

[0046] (1) Improve the accuracy and reliability of verification: Through the dual verification mechanism consisting of "single measurement point adaptive threshold verification" and "multi-measure point correlation constraint verification", combined with physical laws and real-time operating status, it can more accurately identify complex anomalies and significantly reduce the false alarm rate and false negative rate of single method.

[0047] (2) Enhance the system’s adaptability and intelligence: By using the adaptive dynamic threshold model and the model self-learning optimization mechanism, the verification system can dynamically adjust the verification parameters according to the real-time load, equipment status and historical feedback, thus getting rid of the absolute dependence on fixed rules and a large amount of historical data, and thus being able to quickly adapt to the complex and ever-changing operating environment of the substation and newly commissioned equipment.

[0048] (3) Achieving unified management and deep understanding of data: By constructing a physical model mapping library, heterogeneous and multi-source raw data is organized into structured information with unified semantics. This not only solves the problems of data silos and semantic ambiguity, but also enables the verification system to "understand" the physical meaning and contextual relationships of the data, laying a solid foundation for subsequent data mining and advanced applications.

[0049] (4) Optimize operation and maintenance efficiency and decision support: By outputting comprehensive confidence level and multi-level quality labels such as "normal, warning, abnormal", it provides operators with clear and quantitative data quality assessment results. This helps to quickly locate problematic data, reduce the workload of manual review, and provide higher quality and more reliable data support for advanced applications such as monitoring, diagnosis and dispatch, thereby ensuring the safe and stable operation of the power system. Attached Figure Description

[0050] Figure 1 This is a flowchart of the object model construction and data regularization process of the present invention;

[0051] Figure 2 This is a flowchart of the intelligent verification and self-learning process of the present invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0053] The present invention will be further described below with reference to embodiments.

[0054] Please see Figure 1-2 This invention provides an intelligent verification method for substation time-series data based on object model mapping, comprising the following steps:

[0055] S1. Construction of the object model mapping library: Construct the object model mapping library for the substation. The object model mapping library defines a digital twin for each physical device or logical component in the substation and configures its corresponding object model for each digital twin. The object model includes at least the device identifier, static attribute set, dynamic measurement point set, and association relationship set.

[0056] S2. Multi-source data mapping and normalization: Real-time access to the original time-series data stream of the substation, and based on the object model mapping library, the original time-series data is mapped and normalized to the corresponding dynamic measurement points of the object model to form structured data with unified semantics.

[0057] S3, Single Measurement Point Adaptive Threshold Verification: Based on the dynamic measurement point attributes of the object model, an adaptive dynamic threshold is configured for each measurement point, and the data of a single measurement point is checked for the first level of anomalies based on the threshold.

[0058] S4. Multi-measurement point association constraint verification: Based on the association relationship set of the object model, construct physical or logical constraint rules between associated measurement points, and perform a second-level association verification on the data of multiple associated measurement points based on the constraint rules.

[0059] S5. Multi-source verification result fusion output: Based on the results of the first-level anomaly verification and the second-level correlation verification, the comprehensive confidence of each data point is calculated, and the final verification conclusion and data quality label are output based on the comprehensive confidence.

[0060] Further as Figure 1 and Figure 2 As shown, it is worth explaining in detail that the specific process of constructing the object model mapping library in step S1 includes:

[0061] S11: Decompose the equipment entities and logical functions of the substation to form a list of equipment objects;

[0062] S12: Create a unique device identifier for each device object in the list and define its static attribute set, which includes device model, rated parameters, installation location and health status baseline;

[0063] S13: Define a dynamic measurement point set for each device object. The dynamic measurement point set includes measurement point ID, measurement point name, physical quantity, data unit, acquisition frequency, and data type.

[0064] S14: Define the set of relationships within and between different equipment objects. The set of relationships includes topological connections, electrical coupling relationships, and physical law constraints. By decomposing the equipment entities and logical functions of the substation and creating a unique identifier and static attribute set for each object, the originally scattered and heterogeneous equipment resources are integrated into a unified digital asset list that can be identified and managed by the system. This eliminates information confusion caused by different equipment models and manufacturers, and provides a clear "household registration" for data traceability and quality governance. By uniformly defining the dynamic measurement point set (including key metadata such as ID, name, and unit), a standardized "translation dictionary" is provided for all accessed raw time-series data. This allows data from different sources and in different formats to be accurately mapped to their corresponding physical meaning, forming structured data with unified semantics. This solves the problem of data ambiguity and allows the system to "understand" the true meaning of each data point.

[0065] Further as Figure 1 and Figure 2 As shown, it is worth noting that the configuration and calculation method of the adaptive dynamic threshold in step S3 includes:

[0066] Based on the object model mapping library, historical operating data of the target measuring point and its associated load condition data are obtained;

[0067] Based on historical data, the statistical characteristics of the data distribution of this measuring point within a typical load range were calculated.

[0068] Based on the statistical characteristics of data distribution and real-time load data, the reasonable threshold range for this measuring point is dynamically calculated. The upper and lower limits of the dynamic threshold are determined by the following model:

[0069]

[0070] In the formula, L and U are the lower and upper limits of the dynamic threshold, respectively. and These are the mean and standard deviation calculated based on historical data, respectively, where k is a configurable sensitivity coefficient. To associate load parameters The operating condition correction function, with the independent variable, is used to adaptively adjust the threshold width according to the system's operating status. Traditional fixed thresholds cannot reflect the normal operating range of equipment under different loads. This method uses the operating condition correction function to dynamically adjust the threshold according to the system load and other operating states. The threshold is reasonably widened under heavy load and automatically tightened under light load, significantly improving the threshold's adaptability to complex and variable operating environments. This fundamentally solves the false alarm or missed alarm problem caused by the traditional "one-size-fits-all" threshold. By using the object model mapping library to obtain historical data and calculate the mean (μ) and standard deviation (σ), the threshold is generated based on the real data distribution pattern, rather than simply relying on human experience. This allows the threshold to reflect the inherent fluctuation characteristics and normal range of the measuring point, making the setting more objective and scientific. The method actively obtains "related load condition data" through the object model, reflecting a systematic approach. The threshold is no longer set in isolation based on the historical data of a single measuring point, but considers the influence of related parameters in the system, making the threshold more reflective of the real system operating status.

[0071] Further as Figure 1 and Figure 2 As shown, it is worth noting that in step S4, the second-level association verification specifically involves: for the set of measurement points with strong associations { }, construct its constraint relationship function based on physical laws or business logic. During verification, the deviation between the actual data and the constraint relationship is calculated. :

[0072]

[0073] when deviation Exceeding the preset fault tolerance range If an anomaly is detected in the associated data set, it is determined that an anomaly exists. Traditional methods only focus on whether a single data point exceeds the limit, while this step elevates the verification dimension from "point" to "surface" by constructing a constraint relationship function. Even if each individual data point is within its threshold range, as long as they are combined to violate known physical laws (such as power conservation and Kirchhoff's laws), the system can identify such hidden, correlated anomalies, greatly improving the ability to detect complex faults and hidden problems. By calculating the deviation value and comparing it with the tolerance range, the method avoids overreacting to small, reasonable fluctuations. This design acknowledges the existence of measurement noise and normal fluctuations, providing a flexible discrimination interval. Anomalies are only detected when the deviation significantly exceeds the reasonable range, thereby greatly reducing false alarms caused by random errors while ensuring the detection rate.

[0074] Further as Figure 1 and Figure 2As shown, it is worth noting that the construction of the constraint relationship function F is specifically based on the electrical connection relationships or physical principles defined in the object model's relationship set, establishing a mathematical relationship model between multiple measurement points; the mathematical relationship model includes, but is not limited to:

[0075] A summation constraint model is established for the branch currents of the same electrical node based on Kirchhoff's current law.

[0076] A power balance constraint model is established for the same circuit or equipment based on the principle of power conservation.

[0077] Based on the thermodynamic characteristics of the equipment, a thermal circuit model is established for the equipment temperature, ambient temperature, and load current. Traditional rule bases rely on empirical rules summarized manually, while this method directly builds models based on the most fundamental physical laws of power systems (such as Kirchhoff's laws and energy conservation) and the working principles of the equipment. This gives the verification rules a solid scientific foundation, fundamentally avoiding the subjectivity and limitations of empirical rules, and greatly improving the accuracy and reliability of verification. The constructed mathematical models (such as current summation and power balance) can capture the intrinsic relationships between multiple measurement points. This method can identify anomalies that traditional single-point verification cannot detect: for example, when individual current values ​​are not exceeded, but their sum does not satisfy the node current law, the system can accurately locate deep faults at the data quality or equipment level, realizing a leap from the "point" to the "system" verification dimension.

[0078] Further as Figure 1 and Figure 2 As shown, it is worth noting that the calculation model for the comprehensive confidence level C in step S5 is as follows:

[0079]

[0080] In the formula, The score for the first level of anomaly detection is negatively correlated with the degree to which the data point deviates from its adaptive dynamic threshold. The score for the second-level association check is the deviation of the data point from its association constraint group. Negative correlation; and They are respectively and The weights, and + =1, and the weight can be configured according to the type of measurement point and the strength of association. This model unifies the verification results of the first level (single point threshold) and the second level, which are two different dimensions, into a calculable score and merges them through weighted summation. This avoids the limitations of a single verification method and integrates the anomalies of "points" and the contradictions of "surfaces". This makes the final data quality assessment no longer a simple "yes / no" judgment, but a refined and continuous credibility assessment, which significantly improves the accuracy of the conclusion. The comprehensive confidence level C, as a continuous quantitative output, provides a scientific and transparent basis for subsequent data quality classification (such as normal, warning, and abnormal). This allows operation and maintenance personnel not only to know whether the data is "good" or "bad", but also to understand the "degree of good or bad", so as to make more refined decisions: high-confidence normal data can be directly trusted, low-confidence abnormal data should be alerted immediately, and intermediate suspicious data should be manually reviewed, which optimizes the operation and maintenance process and improves efficiency.

[0081] Further as Figure 1 and Figure 2 As shown, it is worth noting that in step S5, the specific process of outputting the verification conclusion based on the comprehensive confidence level C is as follows: a high confidence threshold is preset. and a low confidence threshold ;

[0082] like If the data is normal, the label "Normal" will be output.

[0083] like If the data is deemed suspicious, a "warning" label will be output and a manual review will be requested.

[0084] like If the data is deemed abnormal, an "abnormal" label is output and an alarm is triggered. The three-level labeling system of "normal, warning, and abnormal" corresponds to differentiated handling procedures. "Normal" data can be processed automatically; "abnormal" data triggers an immediate alarm, driving a rapid response; while "warning" data requires human-machine collaborative judgment. This design allows limited operation and maintenance resources to be precisely allocated according to the severity and urgency of data problems, greatly improving the efficiency and targeting of operation and maintenance work. The comprehensive confidence level C and its classification conclusion provide continuous and traceable quantitative indicators for the health of substation data assets. The long-term accumulated label data can be used to analyze the data quality trends of specific equipment or circuits, providing decision support for equipment maintenance and transformation, and realizing a leap from single data verification to long-term data quality management.

[0085] Further as Figure 1 and Figure 2 As shown, it is worth noting that the method also includes a model self-learning and optimization step S6:

[0086] The conclusions of the first-level anomaly check and the second-level correlation check, as well as the results of manual review from external input, are used as feedback data.

[0087] Using feedback data, key parameters in the object model mapping library are optimized through incremental learning. The parameter update mechanism is represented by the following general formula:

[0088]

[0089] In the formula, Represents the historical parameters before optimization. This represents the new parameter estimate calculated from the feedback data. This represents the optimized update parameters. The learning rate controls the magnitude and speed of parameter updates. This step transforms the verification system from a "static expert system" into an "intelligent agent capable of learning from practice." By continuously absorbing new verification conclusions and manual review results, the system can automatically adjust its key parameters, ensuring that the verification rules keep pace with the times and proactively adapt to dynamic environments such as aging substation equipment and changes in operating modes. This completely overcomes the fundamental defects of traditional methods, such as rigid rules that are difficult to adapt to changes. After the system acquires self-learning capabilities, it can start working on newly commissioned equipment without waiting for a long historical data accumulation period. The system can run with initial parameters and quickly use feedback data to complete parameter self-calibration and optimization during this process. This greatly shortens the "maturation" time of the verification model, reduces the strong dependence on historical data, and also reduces the manual intervention costs required for maintaining the rule base later.

[0090] This solution includes the following workflow:

[0091] Step 1: Construct a physical model mapping library for the substation. The physical model mapping library defines a digital twin for each physical device or logical component in the substation and configures its corresponding physical model for each digital twin. The physical model includes at least a device identifier, a set of static attributes, a set of dynamic measurement points, and a set of association relationships.

[0092] Step 2: Real-time access to the original time-series data stream of the substation, and based on the object model mapping library, map and regularize the original time-series data onto the corresponding dynamic measurement points of the object model to form structured data with unified semantics;

[0093] Step 3: Based on the dynamic measurement point attributes of the object model, configure an adaptive dynamic threshold for each measurement point, and perform first-level anomaly verification on the data of a single measurement point based on the threshold.

[0094] Step 4: Based on the set of relationships of the object model, construct physical or logical constraint rules between the associated measurement points, and perform a second-level association verification on the data of multiple associated measurement points based on the constraint rules;

[0095] Step 5: Based on the results of the first-level anomaly check and the second-level correlation check, calculate the comprehensive confidence level of each data point, and output the final check conclusion and data quality label based on the comprehensive confidence level;

[0096] Step Six: Use the conclusions of the first-level anomaly check and the second-level correlation check, as well as the results of manual review from external input, as feedback data; use the feedback data to optimize the key parameters in the object model mapping library through incremental learning.

[0097] In summary: the dual verification mechanism consisting of "single-point adaptive threshold verification" and "multi-point correlation constraint verification," combined with physical laws and real-time operating status, can more accurately identify complex anomalies and significantly reduce the false alarm rate and false negative rate of a single method. Utilizing an adaptive dynamic threshold model and a model self-learning optimization mechanism, the verification system can dynamically adjust verification parameters based on real-time load, equipment status, and historical feedback, eliminating absolute dependence on fixed rules and large amounts of historical data. This allows it to quickly adapt to the complex and ever-changing operating environment of substations and newly commissioned equipment. Furthermore, by constructing a physical model mapping library, heterogeneous and multi-source raw data is organized into structured information with unified semantics. This not only solves the problems of data silos and semantic ambiguity, but also enables the verification system to "understand" the physical meaning and contextual relationships of the data, laying a solid foundation for subsequent data mining and advanced applications. By outputting comprehensive confidence scores and multi-level quality labels such as "normal," "warning," and "abnormal," it provides operators with clear and quantitative data quality assessment results. This helps to quickly locate problematic data, reduce the workload of manual review, and provide higher quality and more reliable data support for advanced applications such as monitoring, diagnosis, and scheduling, thereby ensuring the safe and stable operation of the power system.

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent verification of substation time-series data based on object model mapping, characterized in that, Includes the following steps: S1. Construction of the object model mapping library: Construct an object model mapping library for the substation. The object model mapping library defines a digital twin for each physical device or logical component in the substation and configures its corresponding object model for each digital twin. The object model includes at least an object identifier, a set of static attributes, a set of dynamic measurement points, and a set of association relationships. S2. Multi-source data mapping and normalization: Real-time access to the original time-series data stream of the substation, and according to the object model mapping library, the original time-series data is mapped and normalized to the corresponding object model dynamic measurement points to form structured data with unified semantics. S3, Single Measurement Point Adaptive Threshold Verification: Based on the dynamic measurement point attributes of the object model, an adaptive dynamic threshold is configured for each measurement point, and the data of a single measurement point is verified for the first level of anomaly based on the threshold. S4. Multi-measurement point association constraint verification: Based on the association relationship set of the object model, construct physical or logical constraint rules between associated measurement points, and perform a second-level association verification on the data of multiple associated measurement points based on the constraint rules. S5. Multi-source verification result fusion output: Based on the results of the first-level anomaly verification and the second-level correlation verification, the comprehensive confidence of each data point is calculated, and the final verification conclusion and data quality label are output based on the comprehensive confidence.

2. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: In step S1, the specific process of constructing the object model mapping library includes: S11: Decompose the equipment entities and logical functions of the substation to form a list of equipment objects; S12: Create a unique device identifier for each device object in the list and define its static attribute set, which includes device model, rated parameters, installation location and health status baseline; S13: Define a dynamic measurement point set for each device object. The dynamic measurement point set includes measurement point ID, measurement point name, physical quantity, data unit, acquisition frequency, and data type. S14: Define the set of relationships within a device object and between different device objects. The set of relationships includes topological connection relationships, electrical coupling relationships, and physical law constraint relationships.

3. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: In step S3, the method for configuring and calculating the adaptive dynamic threshold includes: Based on the object model mapping library, historical operating data of the target measuring point and its associated load condition data are obtained; Based on the historical data, the statistical characteristics of the data distribution of the measuring point within the typical load range were calculated. Based on the statistical characteristics of the data distribution and real-time load data, the reasonable threshold range for the measuring point is dynamically calculated. The upper and lower limits of the dynamic threshold are determined by the following model: In the formula, L and U are the lower and upper limits of the dynamic threshold, respectively. and These are the mean and standard deviation calculated based on the historical data, respectively, where k is a configurable sensitivity coefficient. To associate load parameters The operating condition correction function is used to adaptively adjust the threshold width according to the system operating status.

4. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: In step S4, the second-level association verification specifically involves: for a set of measurement points with strong associations { }, construct its constraint relationship function based on physical laws or business logic. During verification, the deviation between the actual data and the constraint relationship is calculated. : when deviation Exceeding the preset fault tolerance range If so, it is determined that there is an anomaly in the associated data set.

5. The intelligent verification method for substation time-series data based on object model mapping according to claim 4, characterized in that: The construction of the constraint relationship function F specifically involves establishing a mathematical correlation model between multiple measurement points based on the electrical connection relationships or physical principles defined in the object model correlation set; the mathematical correlation model includes, but is not limited to: A summation constraint model is established for the branch currents of the same electrical node based on Kirchhoff's current law. A power balance constraint model is established for the same circuit or equipment based on the principle of power conservation. A thermal circuit model is established based on the thermodynamic characteristics of the equipment, considering the equipment temperature, ambient temperature, and load current.

6. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: In step S5, the calculation model for the comprehensive confidence level C is as follows: In the formula, The score for the first level of anomaly detection is negatively correlated with the degree to which the data point deviates from its adaptive dynamic threshold. The score for the second-level association check is the deviation of the data point from its association constraint group. Negative correlation; and They are respectively and The weights, and + =1, the weight can be configured according to the measurement point type and correlation strength.

7. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: In step S5, the specific process of outputting the verification conclusion based on the comprehensive confidence level C is as follows: a high confidence threshold is preset. and a low confidence threshold ; like If the data is normal, the label "Normal" will be output. like If the data is deemed suspicious, a "warning" label will be output and a manual review will be requested. like If the data is abnormal, an "Abnormal" label will be output and an alarm will be triggered.

8. The intelligent verification method for substation time-series data based on object model mapping according to claim 1, characterized in that: The method also includes a model self-learning and optimization step S6: The conclusions of the first-level anomaly check and the second-level correlation check, as well as the results of manual review from external input, are used as feedback data. Using the feedback data, the key parameters in the object model mapping library are optimized through incremental learning. The parameter update mechanism is represented by the following general formula: In the formula, Represents the historical parameters before optimization. This represents the new parameter estimate calculated from the feedback data. This represents the optimized update parameters. The learning rate is used to control the magnitude and speed of parameter updates.

Images