Method and system for processing electric power project data

Abstract

The invention provides an electric power project data processing method and system, and relates to the technical field of computers.The method comprises the steps that firstly, structured measurement data and unstructured knowledge data are obtained; secondly, performing feature extraction on the unstructured data, generating a text coding vector and an image feature vector, and calculating a sliding window time sequence correlation coefficient of the structured data; and then, fusing the features to construct a cross-modal dynamic alignment factor, inputting the cross-modal dynamic alignment factor, a dynamic line loss rate and an equipment health degree vector into a space-time knowledge fusion model, and outputting an equipment risk matrix and a line loss anomaly probability. And then, inputting the technical document into the electric power cognition large model, analyzing a model result in combination with a professional prompt template, and retrieving associated data to generate a governance scheme. According to the scheme, after finite element simulation verification of the digital twin module, if the line loss variable quantity is lower than a dynamic threshold value, the line loss variable quantity is pushed to a power management system to be executed. According to the method, accurate risk assessment and intelligent decision making of the power system can be realized.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and system for processing power project data. Background Art

[0002] With the rapid development of smart grids and the power Internet of Things (IoT), data generated by power systems is characterized by multi-source heterogeneity. Currently, power data management primarily relies on independent systems to process different types of data, such as analyzing real-time measurement data through SCADA systems or storing unstructured technical information using document management systems. However, the lack of deep semantic connections between structured and unstructured data leads to severe information fragmentation in scenarios such as equipment risk assessment and line loss analysis, making it difficult to form a comprehensive decision-making basis. Summary of the Invention

[0003] The technical problem to be solved by this application is to provide a method and system for processing power project data, which can realize accurate assessment of power system risks and intelligent decision-making. The specific solution is as follows:

[0004] A method for processing power project data, comprising:

[0005] Obtaining power project data to be processed, the power project data including structured measurement data acquired in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from the power information system, wherein the structured measurement data includes distribution network load curves, equipment operating parameters, and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images, and technical documents;

[0006] Perform feature extraction on the unstructured knowledge data to generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculate the time series correlation coefficient within a sliding time window for the structured measurement data;

[0007] Constructing a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, engineering drawing image feature vector and time series correlation coefficient;

[0008] Inputting the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data;

[0009] Input the technical documents in the unstructured knowledge data into the power cognitive model, parse the output of the spatiotemporal knowledge fusion model in combination with the preset power professional knowledge prompt template, and link the natural language to database query engine to retrieve the related structured measurement data records to generate a governance plan;

[0010] Conduct finite element simulation verification of the governance plan through the digital twin verification module to obtain simulation results;

[0011] When the line loss rate change of the simulation result does not exceed the dynamic threshold, the governance plan is sent to the power management system, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

[0012] Optionally, the above method includes constructing a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, the engineering drawing image feature vector, and the time series correlation coefficient, including:

[0013] A contrastive learning algorithm is used to align the text encoding vector and the image feature vector in semantic space;

[0014] Calculating the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, where the basic alignment weight represents the strength of the semantic association between the aligned text encoding vector and the image feature vector;

[0015] Calculating a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by the autocorrelation coefficient of the measurement data within the sliding time window;

[0016] The basic alignment weight and the dynamically weighted temporal correlation coefficient are feature fused to obtain the cross-modal dynamic alignment factor.

[0017] In the above method, optionally, the training process of the spatiotemporal knowledge fusion model includes:

[0018] Obtaining an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label for each training sample, each training sample being obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data;

[0019] Select the target training sample currently used for training from the training sample set;

[0020] Inputting the target training sample into the initial model to obtain the recognition result of the initial model;

[0021] Inputting the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary;

[0022] Updating the model parameters of the initial model according to the loss function value;

[0023] If the initial model after the model parameters are updated does not meet the training completion condition, returning to the step of selecting the target training sample currently used for training from the training sample set;

[0024] When the initial model after the model parameters are updated meets the training completion condition, the initial model after the model parameters are updated is determined as the trained spatiotemporal knowledge fusion model.

[0025] Optionally, the above method includes determining the dynamic threshold value based on the dynamic line loss rate of the distribution network, the device health vector, and the environmental monitoring index, including:

[0026] The dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index are input into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

[0027] The above method, optionally, after sending the control solution to the power management system, further includes:

[0028] Real-time monitoring of the distribution network dynamic line loss rate, equipment operating parameters and environmental monitoring indicators after the implementation of the control plan, and generation of execution feedback data stream;

[0029] Performing a deviation analysis on the execution feedback data stream and the simulation results of the digital twin verification module, and calculating a dynamic error coefficient between the actual line loss rate change and the simulation results;

[0030] When the dynamic error coefficient exceeds a preset error tolerance threshold, the execution feedback data stream is input into the spatiotemporal knowledge fusion model for incremental training to update the weight parameters of the device risk scoring matrix.

[0031] A power project data processing system, comprising:

[0032] An acquisition unit is configured to acquire power project data to be processed, wherein the power project data includes structured measurement data acquired in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from a power information system, wherein the structured measurement data includes distribution network load curves, equipment operating parameters, and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images, and technical documents;

[0033] A feature extraction unit is used to extract features from the unstructured knowledge data, generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculate a time series correlation coefficient within a sliding time window for the structured measurement data;

[0034] A construction unit, configured to construct a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, the engineering drawing image feature vector, and the time series correlation coefficient;

[0035] A first execution unit is configured to input the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; and the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data;

[0036] The second execution unit is configured to input the technical documents in the unstructured knowledge data into the power cognitive model, parse the output of the spatiotemporal knowledge fusion model in combination with a preset power professional knowledge prompt template, and use a natural language to database query engine to retrieve related structured measurement data records to generate a governance plan;

[0037] A simulation unit, configured to perform finite element simulation verification on the treatment plan through a digital twin verification module to obtain simulation results;

[0038] The third execution unit is used to send the governance plan to the power management system when the change in the line loss rate of the simulation result does not exceed a dynamic threshold, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

[0039] In the above system, optionally, the construction unit includes:

[0040] An alignment subunit, used to align the text encoding vector and the image feature vector in semantic space using a contrastive learning algorithm;

[0041] A first calculation subunit is configured to calculate the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, wherein the basic alignment weight represents the semantic association strength between the aligned text encoding vector and the image feature vector;

[0042] a second calculation subunit, configured to calculate a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by an autocorrelation coefficient of the measurement data within the sliding time window;

[0043] The feature fusion subunit is used to perform feature fusion on the basic alignment weight and the dynamically weighted temporal correlation coefficient to obtain the cross-modal dynamic alignment factor.

[0044] In the above system, optionally, the first execution unit includes:

[0045] An acquisition subunit is used to acquire an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label of each training sample, and each training sample is obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data;

[0046] A selection subunit is used to select a target training sample currently used for training from the training sample set;

[0047] A first execution subunit is used to input the target training sample into the initial model to obtain the recognition result of the initial model;

[0048] A second execution subunit is configured to input the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary;

[0049] An updating subunit, configured to update the model parameters of the initial model according to the loss function value;

[0050] A third execution subunit is configured to, if the initial model after updating the model parameters does not meet the training completion condition, return to the step of selecting a target training sample currently used for training from the training sample set;

[0051] The fourth execution subunit is used to determine the initial model after the model parameters are updated as the trained spatiotemporal knowledge fusion model when the initial model after the model parameters are updated meets the training completion condition.

[0052] In the above system, optionally, the third execution unit includes:

[0053] The fifth execution sub-unit is used to input the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

[0054] The above system may optionally further include:

[0055] A monitoring unit, configured to monitor in real time the dynamic line loss rate of the distribution network, equipment operating parameters, and environmental monitoring indicators after the execution of the control plan, and generate an execution feedback data stream;

[0056] a calculation unit, configured to perform deviation analysis on the execution feedback data stream and the simulation result of the digital twin verification module, and calculate a dynamic error coefficient between the actual line loss rate change and the simulation result;

[0057] A training unit is configured to input the execution feedback data stream into the spatiotemporal knowledge fusion model for incremental training when the dynamic error coefficient exceeds a preset error tolerance threshold, so as to update the weight parameters of the device risk scoring matrix.

[0058] Based on the above-mentioned method and system for processing power project data provided by the implementation of this application, power project data to be processed can be obtained, and the power project data includes structured measurement data obtained in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from the power information system, wherein the structured measurement data includes distribution network load curve, equipment operating parameters and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images and technical documents; feature extraction is performed on the unstructured knowledge data to generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and the time series correlation coefficient within the sliding time window of the structured measurement data is calculated; a cross-modal dynamic alignment factor is constructed based on the equipment ledger text encoding vector, engineering drawing image feature vector and time series correlation coefficient; the cross-modal dynamic alignment factor, distribution network dynamic line loss rate and equipment health are combined The vector input is pre-trained into the spatiotemporal knowledge fusion model to obtain the output result of the spatiotemporal knowledge fusion model; wherein, the output result includes the equipment risk scoring matrix and the line loss abnormality probability; the dynamic line loss rate of the distribution network and the equipment health vector are determined according to the structured measurement data; the technical documents in the unstructured knowledge data are input into the power cognitive model, and the output result of the spatiotemporal knowledge fusion model is parsed in combination with the preset power professional knowledge prompt template, and the natural language to database query engine is linked to retrieve the related structured measurement data records to generate a governance plan; the governance plan is verified by finite element simulation through the digital twin verification module to obtain the simulation result; when the line loss rate change of the simulation result does not exceed the dynamic threshold, the governance plan is sent to the power management system, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index. The method provided in the embodiment of the present application is applied to realize accurate risk assessment and intelligent decision-making of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without any creative work.

[0060] Figure 1 A flow chart of a method for processing power project data provided in this application;

[0061] Figure 2 A flowchart of a process for constructing a cross-modal dynamic alignment factor provided in this application;

[0062] Figure 3A flowchart of the training process of a spatiotemporal knowledge fusion model provided in this application;

[0063] Figure 4 A schematic diagram of the structure of a power project data processing system provided in this application. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0065] In this application, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0066] An embodiment of the present invention provides a method for processing power project data, which is applied to electronic equipment. The flowchart of the method is as follows: Figure 1 As shown, specifically including:

[0067] S101: Obtain power project data to be processed, wherein the power project data includes structured measurement data obtained in real time through power Internet of Things acquisition equipment, and unstructured knowledge data extracted from the power information system, wherein the structured measurement data includes distribution network load curves, equipment operating parameters and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images and technical documents.

[0068] S102: extracting features from the unstructured knowledge data to generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculating a time series correlation coefficient within a sliding time window for the structured measurement data;

[0069] S103: Constructing a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, the engineering drawing image feature vector, and the time series correlation coefficient;

[0070] S104: Inputting the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data;

[0071] S105: Inputting the technical documents in the unstructured knowledge data into the power cognitive model, parsing the output of the spatiotemporal knowledge fusion model in combination with the preset power professional knowledge prompt template, and linking the natural language to database query engine to retrieve related structured measurement data records to generate a governance plan;

[0072] S106: Performing finite element simulation verification on the governance solution through a digital twin verification module to obtain simulation results;

[0073] S107: When the line loss rate change of the simulation result does not exceed a dynamic threshold, the governance plan is sent to the power management system, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

[0074] The method provided in the embodiments of this application is applied to achieve accurate assessment of power system risks and intelligent decision-making.

[0075] In one embodiment provided in the present application, based on the above solution, optionally, the process of constructing a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, engineering drawing image feature vector and time series correlation coefficient is as follows: Figure 2 Shown, including:

[0076] S201: Use contrastive learning algorithm to align the text encoding vector and image feature vector in semantic space;

[0077] S202: Calculating the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, where the basic alignment weight represents the semantic association strength between the aligned text encoding vector and the image feature vector;

[0078] S203: Calculating a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by an autocorrelation coefficient of the measurement data within the sliding time window;

[0079] S204: Perform feature fusion on the basic alignment weight and the dynamically weighted temporal correlation coefficient to obtain the cross-modal dynamic alignment factor.

[0080] In one embodiment provided in this application, based on the above solution, optionally, the training process of the spatiotemporal knowledge fusion model is as follows: Figure 3 Shown, including:

[0081] S301: Obtain an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label for each training sample, and each training sample is obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data;

[0082] S302: Select a target training sample currently used for training from the training sample set.

[0083] S303: Inputting the target training sample into the initial model to obtain the recognition result of the initial model;

[0084] S304: Inputting the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary;

[0085] S305: updating the model parameters of the initial model according to the loss function value;

[0086] S306: Determine whether the initial model after updating the model parameters meets the training completion condition; if not, return to execute S302; if so, execute S307.

[0087] S307: Determine the initial model after updating the model parameters as the trained spatiotemporal knowledge fusion model.

[0088] In one embodiment provided in the present application, based on the above solution, optionally, the process of determining the dynamic threshold value according to the dynamic line loss rate of the distribution network, the device health vector, and the environmental monitoring index includes:

[0089] The dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index are input into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

[0090] In an embodiment provided in the present application, based on the above solution, optionally, after sending the governance solution to the power management system, the method further includes:

[0091] Real-time monitoring of the distribution network dynamic line loss rate, equipment operating parameters and environmental monitoring indicators after the implementation of the control plan, and generation of execution feedback data stream;

[0092] Performing a deviation analysis on the execution feedback data stream and the simulation results of the digital twin verification module, and calculating a dynamic error coefficient between the actual line loss rate change and the simulation results;

[0093] When the dynamic error coefficient exceeds a preset error tolerance threshold, the execution feedback data stream is input into the spatiotemporal knowledge fusion model for incremental training to update the weight parameters of the device risk scoring matrix.

[0094] See also Figure 4 , is a structural diagram of a power project data processing system provided in this application, the system comprising:

[0095] An acquisition unit 401 is configured to acquire power project data to be processed, wherein the power project data includes structured measurement data acquired in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from a power information system. The structured measurement data includes distribution network load curves, equipment operating parameters, and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images, and technical documents.

[0096] A feature extraction unit 402 is configured to extract features from the unstructured knowledge data, generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculate a temporal correlation coefficient within a sliding time window for the structured measurement data;

[0097] A construction unit 403 is configured to construct a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, the engineering drawing image feature vector, and the time series correlation coefficient;

[0098] A first execution unit 404 is configured to input the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; and the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data.

[0099] The second execution unit 405 is configured to input the technical documents in the unstructured knowledge data into the power cognitive model, parse the output of the spatiotemporal knowledge fusion model using a preset power professional knowledge prompt template, and use a natural language to database query engine to retrieve related structured measurement data records to generate a governance plan.

[0100] The simulation unit 406 is used to perform finite element simulation verification on the treatment plan through the digital twin verification module to obtain simulation results;

[0101] The third execution unit 407 is used to send the governance plan to the power management system when the change in the line loss rate of the simulation result does not exceed the dynamic threshold, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

[0102] In an embodiment provided in the present application, based on the above solution, optionally, the construction unit 403 includes:

[0103] An alignment subunit, used to align the text encoding vector and the image feature vector in semantic space using a contrastive learning algorithm;

[0104] A first calculation subunit is configured to calculate the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, wherein the basic alignment weight represents the semantic association strength between the aligned text encoding vector and the image feature vector;

[0105] a second calculation subunit, configured to calculate a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by an autocorrelation coefficient of the measurement data within the sliding time window;

[0106] The feature fusion subunit is used to perform feature fusion on the basic alignment weight and the dynamically weighted temporal correlation coefficient to obtain the cross-modal dynamic alignment factor.

[0107] In an embodiment provided in the present application, based on the above solution, optionally, the first execution unit 404 includes:

[0108] An acquisition subunit is used to acquire an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label of each training sample, and each training sample is obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data;

[0109] A selection subunit is used to select a target training sample currently used for training from the training sample set;

[0110] A first execution subunit is used to input the target training sample into the initial model to obtain the recognition result of the initial model;

[0111] A second execution subunit is configured to input the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary;

[0112] An updating subunit, configured to update the model parameters of the initial model according to the loss function value;

[0113] A third execution subunit is configured to, if the initial model after updating the model parameters does not meet the training completion condition, return to the step of selecting a target training sample currently used for training from the training sample set;

[0114] The fourth execution subunit is used to determine the initial model after the model parameters are updated as the trained spatiotemporal knowledge fusion model when the initial model after the model parameters are updated meets the training completion condition.

[0115] In an embodiment provided in the present application, based on the above solution, optionally, the third execution unit 407 includes:

[0116] The fifth execution sub-unit is used to input the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

[0117] In an embodiment provided in this application, based on the above solution, optionally, the following is further included:

[0118] A monitoring unit, configured to monitor in real time the dynamic line loss rate of the distribution network, equipment operating parameters, and environmental monitoring indicators after the execution of the control plan, and generate an execution feedback data stream;

[0119] a calculation unit, configured to perform deviation analysis on the execution feedback data stream and the simulation result of the digital twin verification module, and calculate a dynamic error coefficient between the actual line loss rate change and the simulation result;

[0120] A training unit is configured to input the execution feedback data stream into the spatiotemporal knowledge fusion model for incremental training when the dynamic error coefficient exceeds a preset error tolerance threshold, so as to update the weight parameters of the device risk scoring matrix.

[0121] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referenced to each other.

[0122] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0123] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0124] Through the description of the above embodiments, it can be seen that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present application or certain parts of the embodiments.

[0125] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for processing power project data, characterized in that: include: Obtaining power project data to be processed, the power project data including structured measurement data acquired in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from the power information system, wherein the structured measurement data includes distribution network load curves, equipment operating parameters, and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images, and technical documents; Perform feature extraction on the unstructured knowledge data to generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculate the time series correlation coefficient within a sliding time window for the structured measurement data; Constructing a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, engineering drawing image feature vector and time series correlation coefficient; Inputting the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data; Input the technical documents in the unstructured knowledge data into the power cognitive model, parse the output of the spatiotemporal knowledge fusion model in combination with the preset power professional knowledge prompt template, and link the natural language to database query engine to retrieve the related structured measurement data records to generate a governance plan; Conduct finite element simulation verification of the governance plan through the digital twin verification module to obtain simulation results; When the line loss rate change of the simulation result does not exceed the dynamic threshold, the governance plan is sent to the power management system, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

2. The method according to claim 1, characterized in that The cross-modal dynamic alignment factor is constructed based on the equipment ledger text encoding vector, the engineering drawing image feature vector and the time series correlation coefficient, including: A contrastive learning algorithm is used to align the text encoding vector and the image feature vector in semantic space; Calculating the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, where the basic alignment weight represents the strength of the semantic association between the aligned text encoding vector and the image feature vector; Calculating a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by the autocorrelation coefficient of the measurement data within the sliding time window; The basic alignment weight and the dynamically weighted temporal correlation coefficient are feature fused to obtain the cross-modal dynamic alignment factor.

3. The method according to claim 1, characterized in that The training process of the spatiotemporal knowledge fusion model includes: Obtaining an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label for each training sample, each training sample being obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data; Select the target training sample currently used for training from the training sample set; Inputting the target training sample into the initial model to obtain the recognition result of the initial model; Inputting the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary; Updating the model parameters of the initial model according to the loss function value; If the initial model after the model parameters are updated does not meet the training completion condition, returning to the step of selecting the target training sample currently used for training from the training sample set; When the initial model after the model parameters are updated meets the training completion condition, the initial model after the model parameters are updated is determined as the trained spatiotemporal knowledge fusion model.

4. The method according to claim 1, wherein The process of determining the dynamic threshold value according to the dynamic line loss rate of the distribution network, the equipment health vector, and the environmental monitoring index includes: The dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index are input into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

5. The method according to claim 1, characterized in that After sending the control plan to the power management system, the method further includes: Real-time monitoring of the distribution network dynamic line loss rate, equipment operating parameters and environmental monitoring indicators after the implementation of the control plan, and generation of execution feedback data stream; Performing a deviation analysis on the execution feedback data stream and the simulation results of the digital twin verification module, and calculating a dynamic error coefficient between the actual line loss rate change and the simulation results; When the dynamic error coefficient exceeds a preset error tolerance threshold, the execution feedback data stream is input into the spatiotemporal knowledge fusion model for incremental training to update the weight parameters of the device risk scoring matrix.

6. A power project data processing system, characterized in that: include: An acquisition unit is configured to acquire power project data to be processed, wherein the power project data includes structured measurement data acquired in real time by power Internet of Things acquisition equipment, and unstructured knowledge data extracted from a power information system, wherein the structured measurement data includes distribution network load curves, equipment operating parameters, and environmental monitoring indicators, and the unstructured knowledge data includes equipment ledger text, engineering drawing images, and technical documents; A feature extraction unit is used to extract features from the unstructured knowledge data, generate equipment ledger text encoding vectors and engineering drawing image feature vectors, and calculate a time series correlation coefficient within a sliding time window for the structured measurement data; A construction unit, configured to construct a cross-modal dynamic alignment factor based on the equipment ledger text encoding vector, the engineering drawing image feature vector, and the time series correlation coefficient; A first execution unit is configured to input the cross-modal dynamic alignment factor, the distribution network dynamic line loss rate, and the device health vector into a pre-trained spatiotemporal knowledge fusion model to obtain an output result of the spatiotemporal knowledge fusion model; wherein the output result includes an equipment risk score matrix and a line loss abnormality probability; and the distribution network dynamic line loss rate and the device health vector are determined based on the structured measurement data; The second execution unit is configured to input the technical documents in the unstructured knowledge data into the power cognitive model, parse the output of the spatiotemporal knowledge fusion model in combination with a preset power professional knowledge prompt template, and use a natural language to database query engine to retrieve related structured measurement data records to generate a governance plan; A simulation unit, configured to perform finite element simulation verification on the treatment plan through a digital twin verification module to obtain simulation results; The third execution unit is used to send the governance plan to the power management system when the change in the line loss rate of the simulation result does not exceed a dynamic threshold, and the dynamic threshold is determined according to the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index.

7. The system according to claim 6, characterized in that The building block comprises: An alignment subunit, used to align the text encoding vector and the image feature vector in semantic space using a contrastive learning algorithm; A first calculation subunit is configured to calculate the cosine similarity of the aligned text encoding vector and the image feature vector to generate a basic alignment weight, wherein the basic alignment weight represents the semantic association strength between the aligned text encoding vector and the image feature vector; a second calculation subunit, configured to calculate a dynamic weighting coefficient of the time series correlation coefficient based on the structured measurement data within the sliding time window, wherein the dynamic weighting coefficient is determined by an autocorrelation coefficient of the measurement data within the sliding time window; The feature fusion subunit is used to perform feature fusion on the basic alignment weight and the dynamically weighted temporal correlation coefficient to obtain the cross-modal dynamic alignment factor.

8. The system according to claim 6, characterized in that The first execution unit includes: An acquisition subunit is used to acquire an initial model to be trained and a training sample set; the training sample set includes multiple training samples and a sample label of each training sample, and each training sample is obtained by preprocessing structured measurement data and unstructured knowledge data in historical power project data; A selection subunit is used to select a target training sample currently used for training from the training sample set; A first execution subunit is used to input the target training sample into the initial model to obtain the recognition result of the initial model; A second execution subunit is configured to input the recognition result and the sample label of the target training sample into a preset loss function to obtain a loss function value; the loss function is generated based on the safety specification constraints extracted by the power cognitive model, including the safety threshold of the equipment operating parameters and the line loss compliance boundary; An updating subunit, configured to update the model parameters of the initial model according to the loss function value; A third execution subunit is configured to, if the initial model after updating the model parameters does not meet the training completion condition, return to the step of selecting a target training sample currently used for training from the training sample set; The fourth execution subunit is used to determine the initial model after the model parameters are updated as the trained spatiotemporal knowledge fusion model when the initial model after the model parameters are updated meets the training completion condition.

9. The system according to claim 6, wherein: The third execution unit includes: The fifth execution sub-unit is used to input the dynamic line loss rate of the distribution network, the equipment health vector and the environmental monitoring index into the prediction model to obtain a dynamic threshold. The prediction model is trained based on a historical operating condition data set. The historical operating condition data set includes multiple historical operating condition data. The historical operating condition data includes historical line loss rate change time series data, historical environmental monitoring indicators and historical equipment health vectors.

10. The system according to claim 6, wherein: Also includes: A monitoring unit, configured to monitor in real time the dynamic line loss rate of the distribution network, equipment operating parameters, and environmental monitoring indicators after the execution of the control plan, and generate an execution feedback data stream; a calculation unit, configured to perform deviation analysis on the execution feedback data stream and the simulation result of the digital twin verification module, and calculate a dynamic error coefficient between the actual line loss rate change and the simulation result; A training unit is configured to input the execution feedback data stream into the spatiotemporal knowledge fusion model for incremental training when the dynamic error coefficient exceeds a preset error tolerance threshold, so as to update the weight parameters of the device risk scoring matrix.

Images