Task scheduling method, device and equipment for power system, medium and product

Abstract

The embodiment of the invention provides a task scheduling method, device and equipment for a power system, a medium and a product, and is applied to the technical field of smart power grids. The method comprises the following steps: in response to a task scheduling request, obtaining multi-dimensional operation data in a power system; for the operation data of each dimension, distributing a corresponding data preprocessing strategy for the operation data based on a preset strategy rule base; performing data preprocessing on the operation data based on the data preprocessing strategy to obtain preprocessed data; performing feature extraction based on the preprocessed data to obtain a data feature vector; inputting the data feature vectors corresponding to the operation data of the multiple dimensions into a pre-trained scheduling task generation model to obtain a task execution list; executing the scheduling task based on the task execution list; wherein the scheduling task generation model is obtained by performing model training based on historical operation and maintenance data of the power system. The method achieves the technical effect of improving the task scheduling efficiency of the power system.

Description

Technical Field

[0001] The present application relates to the field of smart grid technology, and in particular to a task scheduling method, device, equipment, medium and product for a power system. Background Art

[0002] In the task scheduling and management of power systems, the increasing number of data sources in the power system leads to an increase in the complexity and dynamism of operating data. Therefore, data management is required for power system data of various types or dimensions.

[0003] In the existing technology, the data management method for the power system is mainly: obtaining the operating data of the power system, extracting features from the operating data, and inputting the feature-extracted data into the task scheduling model to obtain the scheduling tasks generated for the current power system operating data, and executing the scheduling tasks to realize the task scheduling and management of the power system.

[0004] Since the existing technology cannot accurately process the power system operation data in complex scenarios, the existing technology has a technical problem of low efficiency in power system task scheduling. Summary of the Invention

[0005] The embodiments of the present application provide a task scheduling method, apparatus, equipment, medium and product for a power system, so as to achieve the technical effect of improving the efficiency of task scheduling in the power system.

[0006] In a first aspect, an embodiment of the present application provides a task scheduling method for a power system, comprising:

[0007] Responding to task scheduling requests, obtaining operational data of multiple dimensions in the power system;

[0008] For each dimension of operating data, a corresponding data preprocessing strategy is assigned to the operating data based on the preset strategy rule library;

[0009] Based on the data preprocessing strategy, the operating data is preprocessed to obtain preprocessed data;

[0010] Perform feature extraction based on preprocessed data to obtain data feature vectors;

[0011] Input the data feature vectors corresponding to the running data of multiple dimensions into the pre-trained scheduling task generation model to obtain the task execution list;

[0012] Execute scheduled tasks based on the task execution list;

[0013] Among them, the scheduling task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

[0014] In one possible implementation, for each dimension of operating data, a corresponding data preprocessing strategy is assigned to the operating data based on a preset strategy rule library, including:

[0015] Determining, based on the operating data, at least one attribute information corresponding to the operating data;

[0016] Based on the attribute information, the data preprocessing strategy corresponding to the running data is matched in the preset strategy rule library;

[0017] Among them, the types of attribute information include: the generation time of the operation data, the device status type corresponding to the operation data, and the data source type corresponding to the operation data; the preset strategy rule base is used to indicate the mapping relationship between different types of attribute information and data preprocessing strategies.

[0018] In one possible implementation, based on the data preprocessing strategy, the operation data is preprocessed to obtain preprocessed data, including:

[0019] For the operating data of each dimension, when the data preprocessing strategy indicates a denoising process, performing data reconstruction processing on the operating data to obtain first preprocessed data;

[0020] When the data preprocessing strategy indicates data filling, prediction is performed on the data segments that need to be filled in the running data to obtain second preprocessed data.

[0021] In one possible implementation, feature extraction is performed based on the preprocessed data to obtain a data feature vector, including:

[0022] Perform feature extraction based on preprocessed data to obtain multiple data features;

[0023] Perform feature dimensionality reduction based on multiple data features to obtain multiple low-dimensional data features;

[0024] Combine multiple low-dimensional data features to obtain the data feature vector corresponding to the running data.

[0025] In one possible implementation, data feature vectors corresponding to the operating data of multiple dimensions are input into a pre-trained scheduling task generation model to obtain a task execution list, including:

[0026] Combine the data feature vectors corresponding to the running data of each dimension to obtain the data feature matrix;

[0027] Input the data feature matrix into the scheduling task generation model to obtain the initial task execution list;

[0028] Adjusting the task execution strategy corresponding to the initial task execution list based on the operation data of multiple dimensions to obtain the task execution list;

[0029] Among them, the initial task execution list includes: multiple scheduled tasks, the task type corresponding to each scheduled task, and the task execution strategy corresponding to the initial task execution list; the task execution strategy includes: the execution order of multiple scheduled tasks, the execution time of each scheduled task and the priority of each scheduled task.

[0030] In one possible implementation, executing a scheduling task based on the task execution list includes:

[0031] Based on the task execution strategy corresponding to the task execution list, obtain the execution order and execution time corresponding to each scheduled task;

[0032] Execute scheduled tasks based on execution order and execution time, and obtain the execution results corresponding to each scheduled task;

[0033] Based on the execution results of multiple scheduling tasks and operating data in multiple dimensions, the preset strategy rule library and scheduling task generation model are updated.

[0034] In one possible implementation, before obtaining operation data in the power system in response to the task scheduling request, the method further includes:

[0035] Obtain historical operation and maintenance data generated by the power system within a preset time period;

[0036] Perform feature extraction on historical operation and maintenance data to obtain multiple data features;

[0037] Calculating an importance score for each data feature, and determining a plurality of target data features from the plurality of data features based on the importance score;

[0038] Extract feature data from each sub-data in the historical operation and maintenance data based on multiple target data features to obtain sub-sample data;

[0039] Match the corresponding data label for each sub-sample data based on the preset label matching rules;

[0040] Generate training sample data based on multiple sub-sample data and their corresponding labels;

[0041] The initial scheduling task generation model is trained based on the training sample data to obtain the target scheduling task generation model.

[0042] In a second aspect, an embodiment of the present application provides a task scheduling device for a power system, comprising:

[0043] An acquisition module, configured to acquire operation data of multiple dimensions in the power system in response to a task scheduling request;

[0044] A first processing module is used to assign a corresponding data preprocessing strategy to the operating data of each dimension based on a preset strategy rule library;

[0045] The second processing module is used to perform data preprocessing on the operating data based on the data preprocessing strategy to obtain preprocessed data;

[0046] The third processing module is used to extract features based on the preprocessed data to obtain a data feature vector;

[0047] The fourth processing module is used to input the data feature vectors corresponding to the operation data of multiple dimensions into the pre-trained scheduling task generation model to obtain a task execution list;

[0048] A fifth processing module, configured to execute the scheduled task based on the task execution list;

[0049] Among them, the scheduling task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

[0050] In a possible implementation, the first processing module is further configured to:

[0051] Determining, based on the operating data, at least one attribute information corresponding to the operating data;

[0052] Based on the attribute information, the data preprocessing strategy corresponding to the running data is matched in the preset strategy rule library;

[0053] Among them, the types of attribute information include: the generation time of the operation data, the device status type corresponding to the operation data, and the data source type corresponding to the operation data; the preset strategy rule base is used to indicate the mapping relationship between different types of attribute information and data preprocessing strategies.

[0054] In a possible implementation, the second processing module is further configured to:

[0055] For the operating data of each dimension, when the data preprocessing strategy indicates a denoising process, performing data reconstruction processing on the operating data to obtain first preprocessed data;

[0056] When the data preprocessing strategy indicates data filling, prediction is performed on the data segments that need to be filled in the running data to obtain second preprocessed data.

[0057] In a possible implementation, the third processing module is further configured to:

[0058] Perform feature extraction based on preprocessed data to obtain multiple data features;

[0059] Perform feature dimensionality reduction based on multiple data features to obtain multiple low-dimensional data features;

[0060] Combine multiple low-dimensional data features to obtain the data feature vector corresponding to the running data.

[0061] In a possible implementation manner, the fourth processing module is further configured to:

[0062] Combine the data feature vectors corresponding to the running data of each dimension to obtain the data feature matrix;

[0063] Input the data feature matrix into the scheduling task generation model to obtain the initial task execution list;

[0064] Adjusting the task execution strategy corresponding to the initial task execution list based on the operation data of multiple dimensions to obtain the task execution list;

[0065] Among them, the initial task execution list includes: multiple scheduled tasks, the task type corresponding to each scheduled task, and the task execution strategy corresponding to the initial task execution list; the task execution strategy includes: the execution order of multiple scheduled tasks, the execution time of each scheduled task and the priority of each scheduled task.

[0066] In a possible implementation manner, the fifth processing module is further configured to:

[0067] Based on the task execution strategy corresponding to the task execution list, obtain the execution order and execution time corresponding to each scheduled task;

[0068] Execute scheduled tasks based on execution order and execution time, and obtain the execution results corresponding to each scheduled task;

[0069] Based on the execution results of multiple scheduling tasks and operating data in multiple dimensions, the preset strategy rule library and scheduling task generation model are updated.

[0070] In a possible implementation, the acquisition module is further configured to:

[0071] Obtain historical operation and maintenance data generated by the power system within a preset time period;

[0072] Perform feature extraction on historical operation and maintenance data to obtain multiple data features;

[0073] Calculating an importance score for each data feature, and determining a plurality of target data features from the plurality of data features based on the importance score;

[0074] Extract feature data from each sub-data in the historical operation and maintenance data based on multiple target data features to obtain sub-sample data;

[0075] Match the corresponding data label for each sub-sample data based on the preset label matching rules;

[0076] Generate training sample data based on multiple sub-sample data and their corresponding labels;

[0077] The initial scheduling task generation model is trained based on the training sample data to obtain the target scheduling task generation model.

[0078] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a memory, a processor;

[0079] Memory stores computer-executable instructions;

[0080] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the first aspect as described above, that is, any possible implementation of the first aspect.

[0081] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and any possible implementation method of the first aspect.

[0082] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the first aspect and any possible implementation method of the first aspect.

[0083] The embodiments of the present application provide a task scheduling method, device, equipment, medium and product for a power system. The method obtains operating data of multiple dimensions in the power system by responding to a task scheduling request of the power system; assigns corresponding data preprocessing strategies to operating data of different dimensions according to a preset strategy rule base, performs data preprocessing on the operating data according to the data preprocessing strategy, and obtains preprocessed data suitable for feature extraction; performs feature extraction on the preprocessed data to obtain a data feature vector, uses a pre-trained scheduling task generation model to process the data feature vector to obtain a task execution list, and executes the scheduling task based on the task execution list. Compared with the prior art, the present application uses different data preprocessing strategies to process operating data, improves the processing accuracy of data preprocessing, uses a pre-trained scheduling task generation model in combination with preprocessed data to realize the generation of a task execution list, uses data preprocessing to improve data processing accuracy, and uses a scheduling task generation model to improve the scheduling task generation efficiency, thereby achieving the technical effect of improving the efficiency of power system task scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0085] Figure 1Schematic diagram of the task scheduling solution for the power system provided in this application Figure 1 ;

[0086] Figure 2 Schematic diagram of the process of task scheduling method for power system provided in this application Figure 2 ;

[0087] Figure 3 Schematic diagram of the process of task scheduling method for power system provided in this application Figure 3 ;

[0088] Figure 4 A flowchart of the training method for the scheduling task generation model provided in this application;

[0089] Figure 5 This is a schematic diagram of the structure of the task scheduling device for the power system provided by this application;

[0090] Figure 6 This is a schematic diagram of the structure of the electronic device provided in this application.

[0091] The above drawings illustrate specific embodiments of the present application, which will be described in more detail below. These drawings and the textual description are not intended to limit the scope of the present application in any way, but rather to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0092] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0093] In the existing technology, when performing data management on the power system, the main management method is: obtaining the operating data of the power system, obtaining a scheduling task execution strategy that matches the operating data through rule matching or model calculation for the operating data, and performing data management and task scheduling based on the scheduling task execution strategy, thereby realizing data management of the power system.

[0094] However, prior art requires analysis of operational data when generating dispatch tasks for power systems. In complex power system scenarios, operational data generated comes from different devices, has different types, and has different structures. When processing data using a unified processing strategy, this results in low data processing accuracy and inaccurate data features extracted from the operational data. Consequently, the resulting dispatch tasks do not conform to the current power system's operating conditions, leading to low efficiency in dispatching power system tasks in prior art.

[0095] In response to the above technical problems, the present application proposes the following technical concept: targeted data processing for complex operating data. Specifically: according to the preset strategy rule base, the corresponding data preprocessing strategy is assigned to the operating data, and the operating data is preprocessed according to the data preprocessing strategy to obtain data suitable for feature extraction, and the pre-trained scheduling task generation model is used to realize the generation of the task execution list, thereby executing the scheduling task. Compared with the existing technology, the present application uses different data preprocessing strategies to process operating data, improves the processing accuracy of data preprocessing, and uses the preprocessed data to extract features and generate scheduling tasks, thereby reducing the probability of the scheduling task being unsuitable and requiring frequent adjustments or rework, thereby achieving the technical effect of improving the efficiency of power system task scheduling.

[0096] The following specific embodiments describe in detail the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0097] Figure 1 Schematic diagram of the task scheduling solution for the power system provided in this application Figure 1 ,like Figure 1 As shown, the method includes:

[0098] S101 : In response to a task scheduling request, obtain operation data of multiple dimensions in a power system.

[0099] In this step, the multi-dimensional operational data includes at least one of the following: electrical measurement data of the power system, equipment status data, energy and market data, grid topology data, environmental and meteorological data, and stability and control data. The operational data can come from monitoring and data acquisition systems, energy management systems, real-time monitoring sensors, and historical databases.

[0100] Optionally, a possible implementation method for obtaining operation data of multiple dimensions in the power system is:

[0101] S1011. Determine requirements corresponding to the operating data based on the task scheduling request.

[0102] In this step, the requirements corresponding to the operational data may be: the time period in which the data is generated, the type of data, the dimension of the data, and the source of the data.

[0103] S1012. According to the requirements corresponding to the operating data, the required operating data of multiple dimensions is extracted from different data sources and databases.

[0104] For example, according to the requirements corresponding to the operating data, the equipment status data within one hour is extracted from the monitoring and data acquisition system; wherein the equipment status data may be: the status of the transformer and switchgear and the corresponding fault records and maintenance logs of the equipment.

[0105] In this step, the multi-dimensional operating data obtained from the power system refers to time series data, which is used to characterize the operating changes of the power system at multiple consecutive time nodes.

[0106] S102: For the operation data of each dimension, assign a corresponding data preprocessing strategy to the operation data based on a preset strategy rule library.

[0107] Optionally, a possible implementation of assigning a corresponding data preprocessing strategy to the running data is:

[0108] S1021. Based on the operation data, determine at least one attribute information corresponding to the operation data.

[0109] In this step, the attribute information types include: the time the operating data was generated, the device status type corresponding to the operating data, and the data source type corresponding to the operating data. The time the operating data was generated is used to determine whether the operating data is real-time or historical data; the device status types corresponding to the operating data include normal, abnormal, alarm, and fault states; and the data source types corresponding to the operating data include real-time monitoring data, device status data, and historical data.

[0110] S1022: Match the preset strategy rule library based on the attribute information to obtain a data preprocessing strategy corresponding to the running data.

[0111] In this step, the preset strategy rule base is used to indicate the mapping relationship between different types of attribute information and data preprocessing strategies.

[0112] Exemplarily, the attribute information of the running data includes the data source type and the generation time. The data preprocessing strategy is determined as follows:

[0113] When the operational data is real-time monitoring data and the time interval between the generation times of two data nodes exceeds the maximum time interval corresponding to real-time monitoring data, the data preprocessing strategy is determined to be to supplement the missing data. Data supplementation can be performed using real-time interpolation or recent historical data, linear interpolation, or model-based interpolation. The purpose of data supplementation is to ensure the continuity of the operational data.

[0114] When the operation data is historical data and the generation time of the operation data exceeds the validity period, the data preprocessing strategy is determined as: performing data cleaning on the operation data to ensure the quality of the operation data.

[0115] Exemplarily, the attribute information of the operating data includes the device status type. The data preprocessing strategy is determined as follows:

[0116] The device status type is normal, and the data preprocessing strategy is determined to be: data denoising and data supplementation.

[0117] When the equipment status type is overload or abnormal, the data preprocessing strategy is determined as: using dynamic interpolation to combine historical data for data supplementation.

[0118] If the device status type is alarm or fault, determine the data preprocessing strategy as: data anomaly detection, denoising, and data supplementation.

[0119] S103: Based on the data preprocessing strategy, perform data preprocessing on the operating data to obtain preprocessed data.

[0120] Optionally, a possible implementation of performing data preprocessing on the operating data based on the data preprocessing strategy to obtain preprocessed data is:

[0121] S1031 . For the operating data of each dimension, when the data preprocessing strategy indicates denoising, perform data reconstruction on the operating data to obtain first preprocessed data.

[0122] In this step, data reconstruction processing refers to data denoising using an autoencoder network.

[0123] For example, the method of using an autoencoder network to perform data denoising is as follows: the encoder in the autoencoder network maps the input running data to a low-dimensional space to obtain a low-dimensional representation; the decoder reconstructs the input running data from the low-dimensional representation to obtain the running data after denoising. As shown in Formula 1, the optimization goal of the autoencoder network is to minimize the difference between the input and the reconstructed output:

[0124]

[0125] Among them, x i is the input operation data, is the output running data reconstructed by the decoder, is the reconstruction loss function, and the goal is to minimize this loss. By training the autoencoder network, noise and outliers in the power system operation data can be detected and removed.

[0126] S1032: When the data preprocessing strategy indicates data filling, predict the data segments that need to be filled in the running data to obtain second preprocessed data.

[0127] In this step, predicting the data segment to be filled refers to predicting the specific data corresponding to the data segment to be filled by using rule matching or model prediction.

[0128] For example, the random forest algorithm of ensemble learning can be used to fill in the missing data in the running data. Each time, a decision tree is trained and the corresponding missing data is predicted. The final data segment is selected through ensemble voting. As shown in Formula 2:

[0129]

[0130] Among them, f i is the prediction function of the i-th tree. N is the number of decision trees in the ensemble, is the data segment obtained for prediction.

[0131] S104: Perform feature extraction based on the preprocessed data to obtain a data feature vector.

[0132] In this step, the preprocessed data is time series data after data cleaning, data denoising and missing value filling. The method of feature extraction for the preprocessed data can be: extracting the time domain features of the preprocessed data to reflect the direct law of data change over time; frequency domain features, which refers to extracting periodic signal features; statistical features, which are used to describe the central trend, degree of dispersion and distribution form of the data; time series pattern features, which are used to characterize the dynamic evolution law of data; event features, which are used to extract correlation features from discrete events; context features, which are used to combine additional information of the external environment or system status; and high-order features obtained through feature engineering or deep learning.

[0133] Exemplarily, a method of performing feature extraction on the pre-processed data may be: performing feature extraction on the processed data corresponding to the running data of each dimension to obtain a corresponding data feature vector.

[0134] It should be noted that the data feature extraction method in this step is as follows Figure 2 Further explanation is given in the embodiment shown and no redundant description is given here.

[0135] S105: Input the data feature vectors corresponding to the operation data of multiple dimensions into a pre-trained scheduling task generation model to obtain a task execution list.

[0136] In this step, the dispatching task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

[0137] Optionally, a possible implementation method for obtaining the task execution list is:

[0138] S1051. Combine the data feature vectors corresponding to the operating data of each dimension to obtain a data feature matrix.

[0139] In this step, the data feature matrix is ​​obtained by combining data feature vectors corresponding to multiple running data. The data feature vectors can be combined in a manner of: vector splicing or vector weighted fusion.

[0140] S1052: Input the data feature matrix into the scheduling task generation model to obtain an initial task execution list.

[0141] In this step, the initial task execution list includes: multiple scheduled tasks, the task type corresponding to each scheduled task, and the task execution strategy corresponding to the initial task execution list.

[0142] Optionally, the task execution list may also include: an identifier of each scheduling task, which is used to identify different scheduling tasks during the task scheduling process of the power system; a description of each scheduling task, which is used to locate the fault or conflict according to the description of the scheduling task when there is a conflict or failure in the execution of the scheduling task.

[0143] S1053: Adjust the task execution strategy corresponding to the initial task execution list based on the operation data of multiple dimensions to obtain a task execution list.

[0144] In this step, the task execution strategy includes: the execution order of multiple scheduled tasks, the execution time of each scheduled task, and the priority of each scheduled task.

[0145] Optionally, the task execution strategy also includes a resource allocation strategy for each scheduled task, which is used to allocate resources for scheduled tasks of different priorities to ensure that key tasks can be executed smoothly.

[0146] For example, the method of adjusting the task execution strategy corresponding to the initial task execution list based on the operation data can be: using the Q learning mechanism to calculate the reward value of each scheduled task under the system state corresponding to the operation data, and adjusting the priority and execution order of the scheduled tasks according to the reward value of each scheduled task. The Q learning mechanism can refer to Formula 3:

[0147]

[0148] Where Q(s,a) is the reward value for selecting scheduling task a in system state s, a is the learning rate, r is the reward value, γ is the discount factor, s′ is the next system state, and a′ is the next scheduled task.

[0149] S106: Execute the scheduled task based on the task execution list.

[0150] In this step, in the process of executing the scheduling task based on the task execution list, the preset strategy rule library and the scheduling task generation model may also be updated according to the execution result of each scheduling task.

[0151] The embodiment of the present application provides a task scheduling method for an electric power system. The method obtains operating data of multiple dimensions in the electric power system by responding to a task scheduling request of the electric power system; assigns corresponding data preprocessing strategies to operating data of different dimensions according to a preset strategy rule base, performs data preprocessing on the operating data according to the data preprocessing strategy, and obtains preprocessed data suitable for feature extraction; performs feature extraction on the preprocessed data to obtain a data feature vector, uses a pre-trained scheduling task generation model to process the data feature vector to obtain a task execution list, and executes the scheduling task based on the task execution list. Compared with the prior art, the present application uses different data preprocessing strategies to process operating data, improves the processing accuracy of data preprocessing, uses a pre-trained scheduling task generation model in combination with preprocessed data to realize the generation of a task execution list, uses data preprocessing to improve data processing accuracy, and uses a scheduling task generation model to improve scheduling task generation efficiency, thereby achieving the technical effect of improving the efficiency of task scheduling in the electric power system.

[0152] Figure 2 Schematic diagram of the process of task scheduling method for power system provided in this application Figure 2 ,exist Figure 1 Based on the embodiment shown, this embodiment further explains the extraction of the data feature vector in step 104, as shown in FIG. Figure 2 As shown, the method includes:

[0153] S201: Perform feature extraction based on preprocessed data to obtain multiple data features.

[0154] In this step, the method of extracting multiple data features based on the preprocessed data can be: extracting statistical features of the preprocessed data by mathematical calculation; extracting high-order features of the preprocessed data by using a deep learning model or a machine learning model.

[0155] Exemplarily, a method for extracting multiple data features may be to extract multiple data features of the preprocessed data using a wavelet transform method, as shown in Formula 4:

[0156]

[0157] Among them, W ψ (a,b) are the wavelet transform coefficients, a is the scale factor, b is the time translation factor, ψ(t) is the wavelet basis function, and x(t) is the input preprocessed data.

[0158] For example, the wavelet transform method can be used to extract the load fluctuation and equipment status change characteristics corresponding to the operation data of the power system in different dimensions.

[0159] S202: Perform feature dimensionality reduction based on multiple data features to obtain multiple low-dimensional data features.

[0160] In this step, the goal of feature dimensionality reduction is to optimize the feature space and reduce feature redundancy and dimensionality. Methods for feature dimensionality reduction include, but are not limited to, principal component analysis for linear dimensionality reduction and unified manifold approximation and projection for nonlinear dimensionality reduction.

[0161] For example, principal component analysis is used to perform feature dimensionality reduction, projecting high-dimensional features to low-dimensional controls, reducing the complexity of calculations while maintaining the main information of the data. The feature dimensionality reduction formula is shown in Formula 5:

[0162]

[0163] Among them, X centered is the centralized input data, X new is the data after dimensionality reduction; V is the eigenvector matrix. Feature dimensionality reduction can reduce redundant information, improve computational efficiency, and preserve key features, further enhancing the accuracy of scheduling task generation.

[0164] S203: Combine multiple low-dimensional data features to obtain a data feature vector corresponding to the operating data.

[0165] In this step, the low-dimensional data features are combined in series, and each low-dimensional data feature is arranged together according to the feature sorting method to form a long vector.

[0166] For example, there are three low-dimensional data features: electrical characteristics, environmental characteristics, and temperature characteristics. Each feature is normalized. For example, the voltage and temperature characteristics are subtracted from their mean and divided by their standard deviation to ensure they are on the same scale. All normalized features are sequentially combined into a data feature vector, which contains information from all low-dimensional data features.

[0167] In this embodiment, data feature vectors are generated by feature dimensionality reduction, thereby achieving the purpose of reducing data calculation complexity.

[0168] Figure 3 Schematic diagram of the process of task scheduling method for power system provided in this application Figure 3 ,exist Figure 1 Based on the embodiment shown, this embodiment further explains the execution of the scheduling task in step 106, as shown in FIG. Figure 2 As shown, the method includes:

[0169] S301 : Based on the task execution strategy corresponding to the task execution list, obtain the execution order and execution time corresponding to each scheduled task.

[0170] In this step, the execution order is used to determine the triggering method of the scheduling task, and the execution time is used to determine the triggering time of the scheduling task.

[0171] For example, there is a task execution list of a power system scheduling task, including four scheduling tasks and detailed information corresponding to each scheduling task, specifically:

[0172] Task ID: Task 1; Task type: Real-time data acquisition; Task description: Obtain the grid bus voltage from the monitoring and data acquisition system; Task priority: High; Execution time: Executed every five minutes.

[0173] Task ID: Task 2; Task type: Load forecast; Task description: Predict the load for one hour based on historical data; Task priority: Medium; Execution time: Every hour.

[0174] Task ID: Task 3; Task type: Control instruction issuance; Task description: Adjust the unit output balance frequency; Task priority: High; Execution time: Execute every ten seconds.

[0175] Task ID: Task 4; Task type: Equipment status assessment; Task description: Transformer oil temperature over-limit analysis; Task priority: Low; Execution time: Executed every thirty minutes.

[0176] The corresponding task execution strategies include:

[0177] Execution order: priority-driven execution, high-priority tasks preempt low-priority tasks; time window triggered, scheduled tasks are executed according to the preset time.

[0178] Resource allocation: High-priority tasks exclusively use computing resources.

[0179] S302: Execute the scheduled tasks based on the execution order and execution time, and obtain the execution result corresponding to each scheduled task.

[0180] Exemplarily, based on the example of S301, the execution of the scheduling task is performed as follows:

[0181] When the start time of the time period coincides with the execution time of Task 2, the process of executing the scheduled task based on the task execution list is as follows:

[0182] a1. When the hour is detected, Task 2 is started to read the load data within 24 hours from the historical database and call the load forecasting model to generate the forecast results.

[0183] a2. According to the execution time and priority, start task 1, collect voltage data from the monitoring and data acquisition system, and store it in the database.

[0184] a3. According to the execution time and priority, start Task 3, obtain the system frequency, calculate the power adjustment amount, and output the control instructions for adjusting the unit output balance frequency.

[0185] a4. Start Task 4 according to the execution time and priority to query the transformer oil temperature time series. If the oil temperature exceeds the limit and lasts for more than 10 minutes, an alarm is triggered.

[0186] In this step, the execution result includes: the data generated during the execution of the scheduling task and the feedback information obtained after the scheduling task is executed.

[0187] S303: Based on the execution results of multiple scheduling tasks and the operation data of multiple dimensions, the preset strategy rule library and the scheduling task generation model are updated.

[0188] In this step, when task execution results deviate from expectations or new devices are connected, the preset policy rule base and the scheduling task generation model need to be updated. The scheduling task generation model can be updated through incremental updates, model retraining, and ensemble learning. The goal is to optimize and update the model based on feedback data after scheduling task execution to ensure its timeliness and accuracy.

[0189] In this embodiment, the scheduling task generation model and the preset policy rule base are updated using the execution result of the scheduling task, thereby achieving the technical effect of improving the accuracy of scheduling task generation.

[0190] Figure 4 A flow chart of the training method for the scheduling task generation model provided in this application, such as Figure 4 As shown, the method includes:

[0191] S401. Obtain historical operation and maintenance data of the power system generated within a preset time period.

[0192] In this step, the historical operation and maintenance data may be obtained by obtaining the historical operation and maintenance data generated within a preset time period from the database corresponding to the monitoring system and the operation and maintenance system of the power system.

[0193] For example, equipment status data, event log data, and environmental log data from the past year are collected from the power system's corresponding operation and maintenance management system database. Event log data includes alarm logs and maintenance work orders; environmental log data includes temperature and humidity data; and equipment status data includes electrical appliance data and equipment operation and maintenance records.

[0194] S402: Extract features from historical operation and maintenance data to obtain multiple data features.

[0195] In this step, the data features extracted from the historical operation and maintenance data can be divided into four categories: statistical features, time domain features, frequency domain features, and semantic features.

[0196] S403: Calculate the importance score of each data feature, and determine multiple target data features from multiple data features based on the importance score.

[0197] Optionally, the target feature can be selected in the following ways:

[0198] S4031. Calculate the importance score for each data feature.

[0199] In this step, the importance score can be calculated using a statistical method or a model-based method.

[0200] For example, the statistical method may be: evaluating the importance of each feature by analyzing its variance, or calculating the correlation coefficient between each feature and the target variable.

[0201] Model-based approaches can be: using a decision tree model, feature importance can be assessed by calculating the contribution of each feature to the tree node split, or in linear regression or logistic regression, feature importance can be determined by looking at the absolute value of the model coefficients.

[0202] S4032. Analyze the importance score, set a threshold, or select the top-N features, where N is a positive integer greater than 0.

[0203] S4033. Retain important features, eliminate redundant features, and obtain multiple target data features.

[0204] S404: Extract feature data from each sub-data in the historical operation and maintenance data based on multiple target data features to obtain sub-sample data.

[0205] In this step, the feature data can be extracted by performing word segmentation on each sub-data of the historical operation and maintenance data, and matching the data after word segmentation and sorting using the target data feature set to obtain sub-sample data.

[0206] S405: Match a corresponding data label for each sub-sample data based on a preset label matching rule.

[0207] In this step, the data label marks the type of scheduling task and is used to provide guidance for model training. The data label of each sub-sample data is generated using the preset label matching rules as shown in Formula 6:

[0208]

[0209] Among them, y is the generated data label, x features is the extracted time series feature, and θ is the parameter of the label generation rule.

[0210] It should be noted that the data labels here are used to guide model training, but the ultimate effect of model training is to generate a corresponding task execution list based on the input data feature matrix; the data labels serve as verification conditions for the accuracy of model training.

[0211] S406: Generate training sample data based on the multiple sub-sample data and their corresponding labels.

[0212] In this step, each piece of training sample data includes a sub-sample data and its corresponding label.

[0213] Optionally, after obtaining the training sample data, the training sample data can also be saved in a readable format to facilitate the call of model training.

[0214] S407: Train the initial scheduling task generation model based on the training sample data to obtain the target scheduling task generation model.

[0215] In this step, when training the model based on the training sample data, the training sample data can be divided into a training set, a validation set, and a test set. The initial task scheduling model can select multiple prediction models with different parameters or different frameworks.

[0216] Exemplarily, there are three initial scheduling task generation models: Model 1, Model 2, and Model 3.

[0217] The training set is used to train each model, and the test set is used to calculate the loss value of each model. When the loss value is lower than the preset loss threshold, the model training of each model is stopped; the performance of each model is verified using the validation set, and the model with the best performance is selected as the final target scheduling task generation model.

[0218] In this embodiment, a target scheduling task generation model for scheduling task generation is obtained by model training, thereby achieving the purpose of improving the scheduling task generation efficiency.

[0219] Figure 5 This is a schematic diagram of the structure of the task scheduling device for the power system provided in this application, such as Figure 5 As shown, the task scheduling device for the power system provided in this embodiment includes:

[0220] The acquisition module 501 is used to acquire operation data of multiple dimensions in the power system in response to a task scheduling request.

[0221] The first processing module 502 is configured to allocate a corresponding data preprocessing strategy to the operating data of each dimension based on a preset strategy rule library.

[0222] The second processing module 503 is configured to perform data preprocessing on the operating data based on a data preprocessing strategy to obtain preprocessed data.

[0223] The third processing module 504 is configured to perform feature extraction based on the preprocessed data to obtain a data feature vector.

[0224] The fourth processing module 505 is used to input the data feature vectors corresponding to the operation data of multiple dimensions into the pre-trained scheduling task generation model to obtain a task execution list.

[0225] The fifth processing module 506 is configured to execute the scheduled task based on the task execution list.

[0226] Among them, the scheduling task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

[0227] In a possible implementation, the first processing module 502 is further configured to:

[0228] Based on the operating data, at least one type of attribute information corresponding to the operating data is determined.

[0229] Based on the attribute information, a match is performed in the preset strategy rule library to obtain the data preprocessing strategy corresponding to the running data.

[0230] The types of attribute information include: the generation time of the operation data, the device status type corresponding to the operation data, and the data source type corresponding to the operation data. The preset strategy rule base is used to indicate the mapping relationship between different types of attribute information and data preprocessing strategies.

[0231] In a possible implementation, the second processing module 503 is further configured to:

[0232] For the operating data of each dimension, when the data preprocessing strategy indicates denoising, data reconstruction is performed on the operating data to obtain first preprocessed data.

[0233] When the data preprocessing strategy indicates data filling, prediction is performed on the data segments that need to be filled in the running data to obtain second preprocessed data.

[0234] In a possible implementation, the third processing module 504 is further configured to:

[0235] Feature extraction is performed based on the preprocessed data to obtain multiple data features.

[0236] Feature dimensionality reduction is performed based on multiple data features to obtain multiple low-dimensional data features.

[0237] Combine multiple low-dimensional data features to obtain the data feature vector corresponding to the running data.

[0238] In a possible implementation, the fourth processing module 505 is further configured to:

[0239] The data feature vectors corresponding to the running data of each dimension are combined to obtain the data feature matrix.

[0240] The data feature matrix is ​​input into the scheduling task generation model to obtain the initial task execution list.

[0241] The task execution strategy corresponding to the initial task execution list is adjusted based on the operation data of multiple dimensions to obtain the task execution list.

[0242] The initial task execution list includes: multiple scheduled tasks, the task type corresponding to each scheduled task, and the task execution strategy corresponding to the initial task execution list. The task execution strategy includes: the execution order of multiple scheduled tasks, the execution time of each scheduled task, and the priority of each scheduled task.

[0243] In a possible implementation, the fifth processing module 506 is further configured to:

[0244] Based on the task execution strategy corresponding to the task execution list, obtain the execution order and execution time corresponding to each scheduled task.

[0245] Execute the scheduled tasks based on the execution order and execution time, and obtain the execution results corresponding to each scheduled task.

[0246] Based on the execution results of multiple scheduling tasks and operating data in multiple dimensions, the preset strategy rule library and scheduling task generation model are updated.

[0247] In a possible implementation, the acquisition module 501 is further configured to:

[0248] Obtain historical operation and maintenance data generated by the power system within a preset time period.

[0249] Feature extraction is performed on historical operation and maintenance data to obtain multiple data features.

[0250] An importance score of each data feature is calculated, and a plurality of target data features are determined from the plurality of data features based on the importance score.

[0251] Based on multiple target data features, feature data is extracted from each sub-data in the historical operation and maintenance data to obtain sub-sample data.

[0252] Match the corresponding data label for each sub-sample data based on the preset label matching rules.

[0253] Generate training sample data based on multiple sub-sample data and their corresponding labels.

[0254] The initial scheduling task generation model is trained based on the training sample data to obtain the target scheduling task generation model.

[0255] The task scheduling device for the power system provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effects are similar and will not be described in detail in this embodiment.

[0256] Figure 6 This is a schematic diagram of the structure of the electronic device provided in this application. Figure 6 As shown, the electronic device provided in this embodiment includes: at least one processor 601 and a memory 602. Optionally, the device further includes a communication component 603. The processor 601, the memory 602, and the communication component 603 are connected via a bus 604.

[0257] In a specific implementation process, at least one processor 601 executes the computer-executable instructions stored in the memory 602 , so that the at least one processor 601 executes the above-mentioned task scheduling method for the power system.

[0258] The specific implementation process of the processor 601 can be found in the above method embodiment. Its implementation principle and technical effects are similar and will not be repeated here in this embodiment.

[0259] In the above embodiments, it should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASICs), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly executed by a hardware processor or by a combination of hardware and software modules within the processor.

[0260] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage.

[0261] A bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be categorized as address buses, data buses, and control buses. For ease of illustration, the buses in the drawings of this application are not limited to just one bus or just one type of bus.

[0262] The present application also provides a computer program product, including a computer program, which implements the above-mentioned task scheduling method for the power system when executed by a processor.

[0263] The present application also provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the above-mentioned task scheduling method for the power system is implemented.

[0264] The readable storage medium may be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium may be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0265] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium may also be an integral part of the processor. The processor and the readable storage medium may reside in an application-specific integrated circuit (ASIC). The processor and the readable storage medium may also reside in a device as discrete components.

[0266] The division of units is merely a logical functional division; actual implementations may employ alternative divisions, such as combining or integrating multiple units or components into another system, or omitting or disabling certain features. Furthermore, any coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between devices or units, whether electrical, mechanical, or otherwise, through some interface.

[0267] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0268] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0269] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present invention. The aforementioned storage medium includes various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0270] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0271] Finally, it should be noted that those skilled in the art will readily identify other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. The present invention is not limited to the precise structure described above and illustrated in the accompanying drawings, and various modifications and variations may be made without departing from the scope thereof. The scope of the present invention is limited solely by the appended claims.

Claims

1. A task scheduling method for a power system, characterized in that: include: Responding to task scheduling requests, obtaining operational data of multiple dimensions in the power system; For each dimension of operating data, a corresponding data preprocessing strategy is assigned to the operating data based on a preset strategy rule library; Based on the data preprocessing strategy, the operation data is preprocessed to obtain preprocessed data; Perform feature extraction based on the preprocessed data to obtain a data feature vector; Input the data feature vectors corresponding to the running data of multiple dimensions into the pre-trained scheduling task generation model to obtain the task execution list; Execute the scheduled task based on the task execution list; The scheduling task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

2. The method according to claim 1, characterized in that For the operation data of each dimension, a corresponding data preprocessing strategy is assigned to the operation data based on a preset strategy rule library, including: determining, based on the operating data, at least one attribute information corresponding to the operating data; Matching the attribute information in the preset strategy rule library to obtain the data preprocessing strategy corresponding to the operation data; Among them, the types of attribute information include: the generation time of the operation data, the device status type corresponding to the operation data, and the data source type corresponding to the operation data; the preset strategy rule base is used to indicate the mapping relationship between different types of attribute information and data preprocessing strategies.

3. The method according to claim 1, characterized in that The step of performing data preprocessing on the operating data based on the data preprocessing strategy to obtain preprocessed data includes: For the operating data of each dimension, when the data preprocessing strategy indicates denoising, performing data reconstruction processing on the operating data to obtain first preprocessed data; When the data preprocessing strategy indicates data filling, prediction is performed on the data segments that need to be filled in the running data to obtain second preprocessed data.

4. The method according to claim 3, characterized in that The performing feature extraction based on the preprocessed data to obtain a data feature vector includes: Performing feature extraction based on the preprocessed data to obtain multiple data features; Performing feature dimensionality reduction based on the multiple data features to obtain multiple low-dimensional data features; Multiple low-dimensional data features are combined to obtain a data feature vector corresponding to the operating data.

5. The method according to claim 4, characterized in that The data feature vectors corresponding to the operation data of multiple dimensions are input into the pre-trained scheduling task generation model to obtain the task execution list, including: Combine the data feature vectors corresponding to the running data of each dimension to obtain the data feature matrix; Input the data feature matrix into the scheduling task generation model to obtain the initial task execution list; Adjusting the task execution strategy corresponding to the initial task execution list based on the operating data of the multiple dimensions to obtain the task execution list; Among them, the initial task execution list includes: multiple scheduled tasks, the task type corresponding to each scheduled task, and the task execution strategy corresponding to the initial task execution list; wherein the task execution strategy includes: the execution order of the multiple scheduled tasks, the execution time of each scheduled task and the priority of each scheduled task.

6. The method according to claim 5, characterized in that The executing the scheduling task based on the task execution list includes: Based on the task execution strategy corresponding to the task execution list, obtaining the execution order and execution time corresponding to each of the scheduled tasks; Execute the scheduled tasks based on the execution order and the execution time, and obtain the execution result corresponding to each scheduled task; Based on the execution results of the plurality of scheduling tasks and the operation data of the plurality of dimensions, the preset strategy rule library and the scheduling task generation model are updated.

7. The method according to claim 1, characterized in that Before obtaining the operation data in the power system in response to the task scheduling request, the method further includes: Obtain historical operation and maintenance data generated by the power system within a preset time period; Perform feature extraction on historical operation and maintenance data to obtain multiple data features; Calculating an importance score for each data feature, and determining a plurality of target data features from the plurality of data features based on the importance score; Extracting feature data from each sub-data in the historical operation and maintenance data based on the multiple target data features to obtain sub-sample data; Match the corresponding data label for each sub-sample data based on the preset label matching rules; Generate training sample data based on multiple sub-sample data and their corresponding labels; An initial scheduling task generation model is trained based on the training sample data to obtain a target scheduling task generation model.

8. A task scheduling device for a power system, characterized in that: include: An acquisition module, configured to acquire operation data of multiple dimensions in the power system in response to a task scheduling request; A first processing module is configured to allocate a corresponding data preprocessing strategy to the operating data of each dimension based on a preset strategy rule library; A second processing module is configured to perform data preprocessing on the operating data based on the data preprocessing strategy to obtain preprocessed data; A third processing module is used to perform feature extraction based on the preprocessed data to obtain a data feature vector; The fourth processing module is used to input the data feature vectors corresponding to the operation data of multiple dimensions into the pre-trained scheduling task generation model to obtain a task execution list; A fifth processing module, configured to execute the scheduled task based on the task execution list; The scheduling task generation model is obtained by model training based on the historical operation and maintenance data of the power system.

9. An electronic device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor performs the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

11. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 7 when executed by a processor.

Images