Power grid real-time data management method and system based on service platform

Abstract

The invention discloses a power grid real-time data management method and system based on a service platform, and relates to the technical field of power grid data transmission. Historical power grid data are collected through a technical architecture based on the service platform, a collection period is set, and real-time power grid data are continuously collected; meanwhile, a data processing method is adopted to process historical power grid data, then an equipment state evaluation model and a power grid operation trend prediction mechanism are constructed respectively, the two mechanisms are integrated to form a power grid data analysis model, and then the equipment state is evaluated through real-time power grid data based on the constructed power grid data analysis model. The power grid real-time data management method comprises the steps of establishing a real-time data management monitoring rule, predicting a power grid operation trend, recording power grid real-time data management information, finally establishing a real-time data management monitoring rule, optimizing a power grid real-time data management flow according to the recorded power grid real-time data management information and business requirements, and effectively improving power grid data management efficiency and flow adaptation capability.

Description

Technical Field

[0001] This invention relates to the field of power grid data transmission technology, specifically to a method and system for real-time power grid data governance based on a business middleware platform. Background Technology

[0002] With the rapid development of smart grids, the number of power grid equipment terminals has surged, resulting in an explosive growth in real-time operating parameters, condition records, and other data. The effectiveness of data governance directly impacts the accuracy of equipment operation and maintenance and power grid dispatch. Currently, power grid data governance largely adopts a decentralized architecture, which has significant shortcomings: inconsistent data interfaces from multiple equipment sources lead to chaotic data collection standards and difficulties in data integration; data processing and modeling are isolated, and equipment status assessment models and operational trend prediction mechanisms lack linkage, making it difficult to form collaborative decision-making capabilities.

[0003] Existing technology, such as the invention patent application with publication number CN120049610A, discloses a power grid data governance method and system based on a business middleware platform. The method includes: acquiring power grid production management data based on a power production management system; the business middleware platform receiving the power grid production management data; and obtaining a first transmission priority index for the power grid production management data under the data feature dimension transmission priority index based on the power grid production management data and data feature dimensions; calculating a second transmission priority index for the power grid production management data under the business demand feature dimension based on the power grid production management data; calculating a third transmission priority index for the power grid production management data under the waiting time feature dimension based on the power grid production management data; and transmitting information in order of the information transmission priority index. This invention can ensure the timely transmission of important data, improve the reliability and stability of the power grid system, and support the refined management of smart grids.

[0004] As can be seen from the above solutions, the current methods only guarantee transmission efficiency through priority, which cannot meet the power grid's in-depth needs for equipment operation and maintenance and risk early warning. In addition, the lack of data governance monitoring and iterative optimization mechanisms makes it impossible to trace the root cause of data problems and dynamically adjust governance strategies, making it difficult to adapt to real-time governance scenarios with changing power grid operating conditions and upgraded business needs. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time power grid data governance method and system based on a business middleware platform, which solves the problems existing in the background technology.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a real-time power grid data governance method based on a business middleware platform, specifically including the following steps:

[0007] S1 collects historical power grid data based on the technical architecture of the business middle platform, and sets the collection period to continuously collect real-time power grid data;

[0008] S2, The historical power grid data is processed using data processing methods to obtain the processed historical power grid data;

[0009] S3. Based on the processed historical power grid data, an equipment status assessment model and a power grid operation trend prediction mechanism are constructed respectively, and the constructed equipment status assessment model and power grid operation trend prediction mechanism are integrated to form a power grid data analysis model.

[0010] S4, based on the constructed power grid data analysis model, evaluates the status of equipment through real-time power grid data, predicts the power grid operation trend, and records real-time power grid data governance information;

[0011] S5 establishes real-time data governance monitoring rules and optimizes the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs.

[0012] Preferably, the technical architecture based on the business middle platform for collecting historical power grid data and setting a collection period to continuously collect real-time power grid data includes the following steps:

[0013] Based on the distributed architecture of the business middle platform, it connects to the terminals of multi-source equipment in the power grid, extracts historical power grid operating parameters, equipment operating condition records and fault information data through standardized interfaces, and integrates the extracted data to form a historical power grid dataset.

[0014] The power grid data acquisition cycle is set according to the resource scheduling capabilities of the business middle platform. Real-time power grid data is obtained according to the set acquisition cycle, and the power grid data acquisition node and time information are recorded synchronously.

[0015] Preferably, the step of processing the collected historical power grid data using a data processing method to obtain processed historical power grid data includes the following steps:

[0016] Based on the business middle platform verification rules, abnormal data values ​​and duplicate records that are outside the reasonable operating range in the historical power grid data are identified and removed, and the source and cause of invalid data are marked to obtain the preliminary processed historical power grid data;

[0017] The pre-processed historical power grid data is classified according to power grid equipment type, operating period, and business scenario to form three-dimensional historical power grid data. The three-dimensional historical power grid data is then processed using the Z-score standardization method to obtain the processed historical power grid data.

[0018] The Z-score standardization formula is shown below:

[0019] ;

[0020] in For standardized historical power grid data, These are three-dimensional historical power grid data values. These are the mean and standard deviation of similar historical power grid data in the three-dimensional historical power grid data;

[0021] Preferably, the steps of constructing an equipment condition assessment model and a power grid operation trend prediction mechanism based on processed historical power grid data, and integrating the constructed equipment condition assessment model and power grid operation trend prediction mechanism to form a power grid data analysis model include the following steps:

[0022] S31, Long Short-Term Memory Network is used to extract features from the processed historical power grid data to obtain the features of the historical power grid data;

[0023] S32, Based on the characteristics of the obtained historical power grid data, a gradient boosting decision tree algorithm is used to construct an equipment status assessment model;

[0024] S33, establish a power grid operation trend prediction mechanism based on historical power grid data and real-time power grid data, and construct the linkage relationship between the equipment status assessment model and the power grid operation trend prediction mechanism according to the service architecture of the business middle platform;

[0025] S34 integrates the equipment status assessment model with the power grid operation trend prediction mechanism to form a power grid data analysis model. Combined with the business needs of the middle platform and historical power grid governance case data, the power grid data analysis model is verified and optimized to obtain the final power grid data analysis model.

[0026] Preferably, the step of using a long short-term memory network to extract features from the processed historical power grid data to obtain historical power grid data features includes the following steps:

[0027] Based on the timestamp standard of the business middle platform, the processed historical power grid data is reorganized into continuous time series data, and the dimensions of the continuous time series data are adjusted according to the input characteristics of the long short-term memory network to form a matrix structure of time-feature dimension.

[0028] A long short-term memory network architecture is established based on the adjusted matrix structure. The long and short-term dependencies in the historical power grid data are extracted according to the gating mechanism of the long short-term memory network and processed through a regularization layer to ensure the generalization ability of feature extraction. The feature vector of the historical power grid data is output through the feature output layer.

[0029] The Long Short-Term Memory (LSTM) network architecture includes: an input layer, a hidden layer, a regularization layer, and a feature output layer;

[0030] Based on the feature invocation specification of the business middle platform, the feature vectors of historical power grid data are standardized, and each feature is labeled with equipment association attributes, timeliness level and business meaning according to the data management standard of the business middle platform to obtain the features of historical power grid data.

[0031] Preferably, the step of constructing an equipment condition assessment model based on the obtained historical power grid data characteristics using a gradient boosting decision tree algorithm includes the following steps:

[0032] Set the target value of the historical power grid data characteristics and calculate the initial predicted value of the historical power grid data characteristics. Calculate the initial residual based on the initial predicted value and the target value of the historical power grid data characteristics. Set the number of decision trees and the maximum depth of a single tree in the gradient boosting decision tree according to the historical power grid data characteristics to complete the initialization of the gradient boosting decision tree model.

[0033] The formula for calculating the initial predicted value is as follows:

[0034] ;

[0035] The formula for calculating residuals is as follows:

[0036] ;

[0037] in These are predicted values ​​based on historical power grid data characteristics. Let j be the sample set corresponding to the j-th leaf node of the m-th decision tree. This represents the number of samples under that leaf node. Let be the residual corresponding to the i-th power grid data sample in the m-th iteration. Let i be the target value of the device state for the i-th sample. This is the total predicted value of the i-th sample after the (m-1)-th iteration;

[0038] Based on the initial residual, in each iteration, the current residual is used as the new target value, and samples are selected from the historical power grid data features to construct a decision tree for that round. During the splitting process at the root node of the tree, the historical power grid data features that contribute the most to the reduction of residual are selected to participate in the splitting.

[0039] During the splitting process, the reduction of the residual for each historical power grid data feature is calculated, and the historical power grid data feature with the largest reduction is taken as the splitting node until the set tree depth is met. After the splitting is completed, the predicted value of the residual is calculated for each leaf node.

[0040] The process involves iteratively constructing and splitting residuals. In each round, the prediction results of the new decision tree are scaled according to the learning rate and superimposed on the total prediction result. The residuals are then updated. All decision trees are weighted and integrated according to the learning rate to output the final prediction result, forming a gradient boosting decision tree model. The resulting gradient boosting decision tree model is used as the equipment status assessment model.

[0041] Preferably, the step of establishing a power grid operation trend prediction mechanism based on historical and real-time power grid data, and constructing the linkage between the equipment status assessment model and the power grid operation trend prediction mechanism according to the service-oriented architecture of the business middleware, includes the following steps:

[0042] The processed historical power grid data and real-time collected power grid operation data are hierarchically correlated according to time series, equipment type and operating conditions. The dynamic change characteristics of the power grid data are captured by the exponential weighted moving average method formula, and the power grid operation trend prediction results are output to form a power grid operation trend prediction mechanism.

[0043] The standard formula for the exponentially weighted moving average method is shown below:

[0044] ;

[0045] in Let t be the dynamic characteristic value in the prediction of the power grid operation trend. For smoothing coefficients, The characteristic value of the power grid operation trend at time t-1;

[0046] Based on the service-oriented interface system of the business middle platform, a data communication channel is established between the equipment status assessment model and the power grid operation trend prediction mechanism. The output of the equipment status assessment model is synchronized to the power grid operation trend prediction mechanism as a constraint condition for power grid operation trend prediction, and a linkage relationship between the two is established based on business scenarios.

[0047] Preferably, the process of evaluating equipment status and predicting power grid operation trends based on the constructed power grid data analysis model, and recording real-time power grid data governance information, includes the following steps:

[0048] Based on the real-time power grid data collected at a set period, the data processing method in step S2 is used to process the data, and the processed real-time data is then input into the power grid data analysis model.

[0049] The equipment status assessment results are output through the equipment status assessment model in the power grid data analysis model, and the operation trend is generated according to the power grid operation trend prediction mechanism.

[0050] Record assessment results, predict operational trends, collect time and model call log governance information, and store them in categories according to device type and time dimension to form governance information records.

[0051] Preferably, the establishment of real-time data governance monitoring rules and the optimization of the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs include the following steps:

[0052] Establish real-time data governance and monitoring rules to monitor the integrity of power grid data and the prediction accuracy of power grid data analysis models, set qualified thresholds for data missing rate and prediction deviation value, and push abnormal notifications to relevant responsible persons when the qualified thresholds are exceeded.

[0053] Based on the established real-time data governance monitoring rules, combined with the recorded power grid data governance information, we trace the specific scenarios of missing data and the reasons for prediction deviations, and assess the adaptation shortcomings of the existing power grid data governance process according to actual business needs.

[0054] Based on the assessment results, the data acquisition strategy was adjusted, the latest power grid operation data was added to retrain the model to optimize parameters, the anomaly verification logic in data processing was improved, and the monitoring qualification threshold was adjusted in sync with changes in business needs.

[0055] This embodiment also discloses a system for a real-time power grid data governance method based on a business middle platform, including: a data acquisition module, a data preprocessing module, a model building module, an evaluation and recording module, and a monitoring and optimization module;

[0056] The data acquisition module is used to collect historical power grid data according to the distributed architecture and standardized interface of the business middle platform, and to continuously collect power grid data according to the set collection period, and record the power grid data collection node and time information.

[0057] The data preprocessing module is used to preprocess and standardize the power grid data based on the business middle platform verification rules, obtain the processed power grid data, and form three-dimensional historical power grid data according to three-dimensional classification.

[0058] The model building module is used to construct an equipment status assessment model and a power grid operation trend prediction mechanism based on processed historical and real-time power grid data, through long short-term memory network and gradient boosting decision tree algorithm, and integrate the two to form a power grid data analysis model.

[0059] The assessment and recording module is used to output equipment status assessment results based on the power grid data analysis model and the equipment status assessment model, and output prediction information through the operation trend prediction mechanism, while simultaneously recording power grid data governance information.

[0060] The monitoring and optimization module is used to establish real-time monitoring rules, set qualified thresholds for data missing rate and prediction deviation value, and optimize the power grid data model and power grid governance process by combining the recorded governance information and the shortcomings of the business demand assessment process.

[0061] The beneficial effects of this invention are as follows:

[0062] (1) This invention collects historical power grid data through a technical architecture based on a business middle platform, sets a collection period, continuously collects real-time power grid data, and uses data processing methods to process the historical power grid data to obtain processed historical power grid data. Based on the processed historical power grid data, an equipment status assessment model and a power grid operation trend prediction mechanism are constructed respectively. The constructed equipment status assessment model and power grid operation trend prediction mechanism are integrated to form a power grid data analysis model. Based on the constructed power grid data analysis model, the equipment status is assessed through real-time power grid data, and the power grid operation trend is predicted. Real-time power grid data governance information is recorded. Finally, real-time data governance monitoring rules are established, and the real-time power grid data governance process is optimized according to the recorded real-time power grid data governance information and business needs, effectively improving the efficiency and process adaptability of power grid data governance.

[0063] (2) This invention achieves unified standardization of multi-source power grid data through three-dimensional classification and standardization processing. Combining the temporal feature extraction capability of long short-term memory network with the evaluation advantage of gradient boosting decision tree, it significantly improves data quality and the reliability of model analysis, providing a high-quality data foundation for accurate power grid decision-making.

[0064] (3) This invention realizes the linkage and collaboration of equipment status assessment and operation trend prediction through the service architecture of the business middle platform. Combined with the full-process monitoring and dynamic optimization mechanism, it effectively reduces the governance risks caused by data loss and prediction deviation, and helps the power grid achieve refined operation and maintenance and efficient scheduling. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the power grid real-time data governance method based on a business middleware according to the present invention. Detailed Implementation

[0066] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0067] Example 1

[0068] Please see Figure 1 This embodiment of the power grid real-time data governance method based on the business middle platform specifically includes the following steps:

[0069] S1 collects historical power grid data based on the technical architecture of the business middle platform, and sets the collection period to continuously collect real-time power grid data;

[0070] S2, The historical power grid data is processed using data processing methods to obtain the processed historical power grid data;

[0071] S3. Based on the processed historical power grid data, an equipment status assessment model and a power grid operation trend prediction mechanism are constructed respectively, and the constructed equipment status assessment model and power grid operation trend prediction mechanism are integrated to form a power grid data analysis model.

[0072] S4, based on the constructed power grid data analysis model, evaluates the status of equipment through real-time power grid data, predicts the power grid operation trend, and records real-time power grid data governance information;

[0073] S5 establishes real-time data governance monitoring rules and optimizes the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs.

[0074] Further, please refer to Figure 1 The process of collecting historical power grid data based on the technical architecture of the business middle platform, and setting a collection cycle, and continuously collecting real-time power grid data includes the following steps:

[0075] Based on the distributed architecture of the business middle platform, it connects to the terminals of multi-source equipment in the power grid, extracts historical power grid operating parameters, equipment operating condition records and fault information data through standardized interfaces, and integrates the extracted data to form a historical power grid dataset.

[0076] The power grid data acquisition cycle is set according to the resource scheduling capabilities of the business middle platform. Real-time power grid data is obtained according to the set acquisition cycle, and the power grid data acquisition node and time information are recorded synchronously.

[0077] Further, please refer to Figure 1 The process of processing the collected historical power grid data using data processing methods to obtain processed historical power grid data includes the following steps:

[0078] Based on the business middle platform verification rules, abnormal data values ​​and duplicate records that are outside the reasonable operating range in the historical power grid data are identified and removed, and the source and cause of invalid data are marked to obtain the preliminary processed historical power grid data;

[0079] The pre-processed historical power grid data is classified according to power grid equipment type, operating period, and business scenario to form three-dimensional historical power grid data. The three-dimensional historical power grid data is then processed using the Z-score standardization method to obtain the processed historical power grid data.

[0080] The Z-score standardization formula is shown below:

[0081] ;

[0082] in For standardized historical power grid data, These are three-dimensional historical power grid data values. These are the mean and standard deviation of similar historical power grid data in the three-dimensional historical power grid data;

[0083] Further, please refer to Figure 1 Based on processed historical power grid data, an equipment condition assessment model and a power grid operation trend prediction mechanism are constructed respectively. The integration of the constructed equipment condition assessment model and power grid operation trend prediction mechanism into a power grid data analysis model includes the following steps:

[0084] S31, Long Short-Term Memory Network is used to extract features from the processed historical power grid data to obtain the features of the historical power grid data;

[0085] S32, Based on the characteristics of the obtained historical power grid data, a gradient boosting decision tree algorithm is used to construct an equipment status assessment model;

[0086] S33, establish a power grid operation trend prediction mechanism based on historical power grid data and real-time power grid data, and construct the linkage relationship between the equipment status assessment model and the power grid operation trend prediction mechanism according to the service architecture of the business middle platform;

[0087] S34 integrates the equipment status assessment model with the power grid operation trend prediction mechanism to form a power grid data analysis model. Combined with the business needs of the middle platform and historical power grid governance case data, the power grid data analysis model is verified and optimized to obtain the final power grid data analysis model.

[0088] Further, please refer to Figure 1 The process of extracting features from the processed historical power grid data using a long short-term memory network includes the following steps:

[0089] Based on the timestamp standard of the business middle platform, the processed historical power grid data is reorganized into continuous time series data, and the dimensions of the continuous time series data are adjusted according to the input characteristics of the long short-term memory network to form a matrix structure of time-feature dimension.

[0090] A long short-term memory network architecture is established based on the adjusted matrix structure. The long and short-term dependencies in the historical power grid data are extracted according to the gating mechanism of the long short-term memory network and processed through a regularization layer to ensure the generalization ability of feature extraction. The feature vector of the historical power grid data is output through the feature output layer.

[0091] The Long Short-Term Memory (LSTM) network architecture includes: an input layer, a hidden layer, a regularization layer, and a feature output layer;

[0092] Based on the feature invocation specification of the business middle platform, the feature vectors of historical power grid data are standardized, and each feature is labeled with equipment association attributes, timeliness level and business meaning according to the data management standard of the business middle platform to obtain the features of historical power grid data.

[0093] Further, please refer to Figure 1 Based on the characteristics of the obtained historical power grid data, the gradient boosting decision tree algorithm is used to construct an equipment condition assessment model, which includes the following steps:

[0094] Set the target value of the historical power grid data characteristics and calculate the initial predicted value of the historical power grid data characteristics. Calculate the initial residual based on the initial predicted value and the target value of the historical power grid data characteristics. Set the number of decision trees and the maximum depth of a single tree in the gradient boosting decision tree according to the historical power grid data characteristics to complete the initialization of the gradient boosting decision tree model.

[0095] The formula for calculating the initial predicted value is as follows:

[0096] ;

[0097] The formula for calculating residuals is as follows:

[0098] ;

[0099] in These are predicted values ​​based on historical power grid data characteristics. Let j be the sample set corresponding to the j-th leaf node of the m-th decision tree. This represents the number of samples under that leaf node. Let be the residual corresponding to the i-th power grid data sample in the m-th iteration. Let i be the target value of the device state for the i-th sample. This is the total predicted value of the i-th sample after the (m-1)-th iteration;

[0100] Based on the initial residual, in each iteration, the current residual is used as the new target value, and samples are selected from the historical power grid data features to construct a decision tree for that round. During the splitting process at the root node of the tree, the historical power grid data features that contribute the most to the reduction of residual are selected to participate in the splitting.

[0101] During the splitting process, the reduction of the residual for each historical power grid data feature is calculated, and the historical power grid data feature with the largest reduction is taken as the splitting node until the set tree depth is met. After the splitting is completed, the predicted value of the residual is calculated for each leaf node.

[0102] The process involves iteratively constructing and splitting residuals. In each round, the prediction results of the new decision tree are scaled according to the learning rate and superimposed on the total prediction result. The residuals are then updated. All decision trees are weighted and integrated according to the learning rate to output the final prediction result, forming a gradient boosting decision tree model. The resulting gradient boosting decision tree model is used as the equipment status assessment model.

[0103] Further, please refer to Figure 1 Based on historical and real-time power grid data, a power grid operation trend prediction mechanism is established. Furthermore, according to the service-oriented architecture of the business platform, the linkage between the equipment status assessment model and the power grid operation trend prediction mechanism is constructed, including the following steps:

[0104] The processed historical power grid data and real-time collected power grid operation data are hierarchically correlated according to time series, equipment type and operating conditions. The dynamic change characteristics of the power grid data are captured by the exponential weighted moving average method formula, and the power grid operation trend prediction results are output to form a power grid operation trend prediction mechanism.

[0105] The standard formula for the exponentially weighted moving average method is shown below:

[0106] ;

[0107] in Let t be the dynamic characteristic value in the prediction of the power grid operation trend. For smoothing coefficients, The characteristic value of the power grid operation trend at time t-1;

[0108] Based on the service-oriented interface system of the business middle platform, a data communication channel is established between the equipment status assessment model and the power grid operation trend prediction mechanism. The output of the equipment status assessment model is synchronized to the power grid operation trend prediction mechanism as a constraint condition for power grid operation trend prediction, and a linkage relationship between the two is established based on business scenarios.

[0109] Further, please refer to Figure 1 Based on the constructed power grid data analysis model, the system assesses equipment status using real-time power grid data, predicts power grid operation trends, and records real-time power grid data governance information, including the following steps:

[0110] Based on the real-time power grid data collected at a set period, the data processing method in step S2 is used to process the data, and the processed real-time data is then input into the power grid data analysis model.

[0111] The equipment status assessment results are output through the equipment status assessment model in the power grid data analysis model, and the operation trend is generated according to the power grid operation trend prediction mechanism.

[0112] Record assessment results, predict operational trends, collect time and model call log governance information, and store them in categories according to device type and time dimension to form governance information records.

[0113] Further, please refer to Figure 1 Establish real-time data governance monitoring rules, and optimize the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs, including the following steps:

[0114] Establish real-time data governance and monitoring rules to monitor the integrity of power grid data and the prediction accuracy of power grid data analysis models, set qualified thresholds for data missing rate and prediction deviation value, and push abnormal notifications to relevant responsible persons when the qualified thresholds are exceeded.

[0115] Based on the established real-time data governance monitoring rules, combined with the recorded power grid data governance information, we trace the specific scenarios of missing data and the reasons for prediction deviations, and assess the adaptation shortcomings of the existing power grid data governance process according to actual business needs.

[0116] Based on the assessment results, the data acquisition strategy was adjusted, the latest power grid operation data was added to retrain the model to optimize parameters, the anomaly verification logic in data processing was improved, and the monitoring qualification threshold was adjusted in sync with changes in business needs.

[0117] Example 2

[0118] This embodiment also discloses a system for a real-time power grid data governance method based on a business middle platform, including: a data acquisition module, a data preprocessing module, a model building module, an evaluation and recording module, and a monitoring and optimization module;

[0119] The data acquisition module is used to collect historical power grid data according to the distributed architecture and standardized interface of the business middle platform, and to continuously collect power grid data according to the set collection period, and record the power grid data collection node and time information.

[0120] The data preprocessing module is used to preprocess and standardize the power grid data based on the business middle platform verification rules, obtain the processed power grid data, and form three-dimensional historical power grid data according to three-dimensional classification.

[0121] The model building module is used to construct an equipment status assessment model and a power grid operation trend prediction mechanism based on processed historical and real-time power grid data, through long short-term memory network and gradient boosting decision tree algorithm, and integrate the two to form a power grid data analysis model.

[0122] The assessment and recording module is used to output equipment status assessment results based on the power grid data analysis model and the equipment status assessment model, and output prediction information through the operation trend prediction mechanism, while simultaneously recording power grid data governance information.

[0123] The monitoring and optimization module is used to establish real-time monitoring rules, set qualified thresholds for data missing rate and prediction deviation value, and optimize the power grid data model and power grid governance process by combining the recorded governance information and the shortcomings of the business demand assessment process.

[0124] It should be noted that,

[0125] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for real-time power grid data governance based on a business middleware platform, characterized in that: Includes the following steps: S1 collects historical power grid data based on the technical architecture of the business middle platform, and sets the collection period to continuously collect real-time power grid data; S2, The historical power grid data is processed using data processing methods to obtain the processed historical power grid data; S3. Based on the processed historical power grid data, an equipment status assessment model and a power grid operation trend prediction mechanism are constructed respectively, and the constructed equipment status assessment model and power grid operation trend prediction mechanism are integrated to form a power grid data analysis model. S4, based on the constructed power grid data analysis model, evaluates the status of equipment through real-time power grid data, predicts the power grid operation trend, and records real-time power grid data governance information; S5 establishes real-time data governance monitoring rules and optimizes the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs.

2. The power grid real-time data governance method based on a business middleware platform according to claim 1, characterized in that, The technical architecture based on the business middle platform collects historical power grid data and sets a collection period to continuously collect real-time power grid data, including the following steps: Based on the distributed architecture of the business middle platform, it connects to the terminals of multi-source equipment in the power grid, extracts historical power grid operating parameters, equipment operating condition records and fault information data through standardized interfaces, and integrates the extracted data to form a historical power grid dataset. The power grid data acquisition cycle is set according to the resource scheduling capabilities of the business middle platform. Real-time power grid data is obtained according to the set acquisition cycle, and the power grid data acquisition node and time information are recorded synchronously.

3. The power grid real-time data governance method based on a business middleware platform according to claim 1, characterized in that, The process of processing the collected historical power grid data using data processing methods to obtain processed historical power grid data includes the following steps: Based on the business middle platform verification rules, abnormal data values ​​and duplicate records that are outside the reasonable operating range in the historical power grid data are identified and removed, and the source and cause of invalid data are marked to obtain the preliminary processed historical power grid data; The pre-processed historical power grid data is classified according to power grid equipment type, operating period, and business scenario to form three-dimensional historical power grid data. The three-dimensional historical power grid data is then processed using the Z-score standardization method to obtain the processed historical power grid data. The Z-score standardization formula is shown below: ； in For standardized historical power grid data, These are three-dimensional historical power grid data values. These represent the mean and standard deviation of similar historical power grid data in the three-dimensional historical power grid data set.

4. The power grid real-time data governance method based on a business middleware platform according to claim 1, characterized in that, The process of constructing an equipment condition assessment model and a power grid operation trend prediction mechanism based on processed historical power grid data, and then integrating the constructed equipment condition assessment model and power grid operation trend prediction mechanism to form a power grid data analysis model includes the following steps: S31, Long Short-Term Memory Network is used to extract features from the processed historical power grid data to obtain the features of the historical power grid data; S32, Based on the characteristics of the obtained historical power grid data, a gradient boosting decision tree algorithm is used to construct an equipment status assessment model; S33, establish a power grid operation trend prediction mechanism based on historical power grid data and real-time power grid data, and construct the linkage relationship between the equipment status assessment model and the power grid operation trend prediction mechanism according to the service architecture of the business middle platform; S34 integrates the equipment status assessment model with the power grid operation trend prediction mechanism to form a power grid data analysis model. Combined with the business needs of the middle platform and historical power grid governance case data, the power grid data analysis model is verified and optimized to obtain the final power grid data analysis model.

5. The power grid real-time data governance method based on a business middleware platform according to claim 4, characterized in that, The process of using a Long Short-Term Memory (LSTM) network to extract features from the processed historical power grid data includes the following steps: Based on the timestamp standard of the business middle platform, the processed historical power grid data is reorganized into continuous time series data, and the dimensions of the continuous time series data are adjusted according to the input characteristics of the long short-term memory network to form a matrix structure of time-feature dimension. A long short-term memory network architecture is established based on the adjusted matrix structure. The long and short-term dependencies in the historical power grid data are extracted according to the gating mechanism of the long short-term memory network and processed through a regularization layer to ensure the generalization ability of feature extraction. The feature vector of the historical power grid data is output through the feature output layer. The Long Short-Term Memory (LSTM) network architecture includes: an input layer, a hidden layer, a regularization layer, and a feature output layer; Based on the feature invocation specification of the business middle platform, the feature vectors of historical power grid data are standardized, and each feature is labeled with equipment association attributes, timeliness level and business meaning according to the data management standard of the business middle platform to obtain the features of historical power grid data.

6. The power grid real-time data governance method based on a business middleware platform according to claim 4, characterized in that, The process of constructing an equipment condition assessment model based on the obtained historical power grid data features using the gradient boosting decision tree algorithm includes the following steps: Set the target value of the historical power grid data characteristics and calculate the initial predicted value of the historical power grid data characteristics. Calculate the initial residual based on the initial predicted value and the target value of the historical power grid data characteristics. Set the number of decision trees and the maximum depth of a single tree in the gradient boosting decision tree according to the historical power grid data characteristics to complete the initialization of the gradient boosting decision tree model. The formula for calculating the initial predicted value is as follows: ； The formula for calculating residuals is as follows: ； in These are predicted values ​​based on historical power grid data characteristics. Let j be the sample set corresponding to the j-th leaf node of the m-th decision tree. This represents the number of samples under that leaf node. Let be the residual corresponding to the i-th power grid data sample in the m-th iteration. Let i be the target value of the device state for the i-th sample. This is the total predicted value of the i-th sample after the (m-1)-th iteration; Based on the initial residual, in each iteration, the current residual is used as the new target value, and samples are selected from the historical power grid data features to construct a decision tree for that round. During the splitting process at the root node of the tree, the historical power grid data features that contribute the most to the reduction of residual are selected to participate in the splitting. During the splitting process, the reduction of the residual for each historical power grid data feature is calculated, and the historical power grid data feature with the largest reduction is taken as the splitting node until the set tree depth is met. After the splitting is completed, the predicted value of the residual is calculated for each leaf node. The process involves iteratively constructing and splitting residuals. In each round, the prediction results of the new decision tree are scaled according to the learning rate and superimposed on the total prediction result. The residuals are then updated. All decision trees are weighted and integrated according to the learning rate to output the final prediction result, forming a gradient boosting decision tree model. The resulting gradient boosting decision tree model is used as the equipment status assessment model.

7. The power grid real-time data governance method based on a business middleware platform according to claim 4, characterized in that, The establishment of a power grid operation trend prediction mechanism based on historical and real-time power grid data, and the construction of the linkage between the equipment status assessment model and the power grid operation trend prediction mechanism according to the service-oriented architecture of the business platform, includes the following steps: The processed historical power grid data and real-time collected power grid operation data are hierarchically correlated according to time series, equipment type and operating conditions. The dynamic change characteristics of the power grid data are captured by the exponential weighted moving average method formula, and the power grid operation trend prediction results are output to form a power grid operation trend prediction mechanism. The standard formula for the exponentially weighted moving average method is shown below: ； in Let t be the dynamic characteristic value in the prediction of the power grid operation trend. For smoothing coefficients, The characteristic value of the power grid operation trend at time t-1; Based on the service-oriented interface system of the business middle platform, a data communication channel is established between the equipment status assessment model and the power grid operation trend prediction mechanism. The output of the equipment status assessment model is synchronized to the power grid operation trend prediction mechanism as a constraint condition for power grid operation trend prediction, and a linkage relationship between the two is established based on business scenarios.

8. The power grid real-time data governance method based on a business middleware platform according to claim 1, characterized in that, The aforementioned power grid data analysis model, which assesses equipment status and predicts power grid operation trends using real-time power grid data, records real-time power grid data governance information, including the following steps: Based on the real-time power grid data collected at a set period, the data processing method in step S2 is used to process the data, and the processed real-time data is then input into the power grid data analysis model. The equipment status assessment results are output through the equipment status assessment model in the power grid data analysis model, and the operation trend is generated according to the power grid operation trend prediction mechanism. Record assessment results, predict operational trends, collect time and model call log governance information, and store them in categories according to device type and time dimension to form governance information records.

9. The power grid real-time data governance method based on a business middleware platform according to claim 1, characterized in that, The establishment of real-time data governance monitoring rules, and the optimization of the real-time data governance process of the power grid based on the recorded real-time data governance information of the power grid and business needs, include the following steps: Establish real-time data governance and monitoring rules to monitor the integrity of power grid data and the prediction accuracy of power grid data analysis models, set qualified thresholds for data missing rate and prediction deviation value, and push abnormal notifications to relevant responsible persons when the qualified thresholds are exceeded. Based on the established real-time data governance monitoring rules, combined with the recorded power grid data governance information, we trace the specific scenarios of missing data and the reasons for prediction deviations, and assess the adaptation shortcomings of the existing power grid data governance process according to actual business needs. Based on the assessment results, the data acquisition strategy was adjusted, the latest power grid operation data was added to retrain the model to optimize parameters, the anomaly verification logic in data processing was improved, and the monitoring qualification threshold was adjusted in sync with changes in business needs.

10. A system for implementing the real-time power grid data governance method based on a business middleware as described in claims 1-9, characterized in that, include: The system includes a data acquisition module, a data preprocessing module, a model building module, an evaluation and recording module, and a monitoring and optimization module. The data acquisition module is used to collect historical power grid data according to the distributed architecture and standardized interface of the business middle platform, and to continuously collect power grid data according to the set collection period, and record the power grid data collection node and time information. The data preprocessing module is used to preprocess and standardize the power grid data based on the business middle platform verification rules, obtain the processed power grid data, and form three-dimensional historical power grid data according to three-dimensional classification. The model building module is used to construct an equipment status assessment model and a power grid operation trend prediction mechanism based on processed historical and real-time power grid data, through long short-term memory network and gradient boosting decision tree algorithm, and integrate the two to form a power grid data analysis model. The assessment and recording module is used to output equipment status assessment results based on the power grid data analysis model and the equipment status assessment model, and output prediction information through the operation trend prediction mechanism, while simultaneously recording power grid data governance information. The monitoring and optimization module is used to establish real-time monitoring rules, set qualified thresholds for data missing rate and prediction deviation value, and optimize the power grid data model and power grid governance process by combining the recorded governance information and the shortcomings of the business demand assessment process.

Images