Power load data forgery detection method, device, storage medium and related equipment

Abstract

The power load data forgery detection method, device, storage medium and related equipment provided by the present application can, when detecting power load data, first convert the power load data into an image to be detected, and determine a target data detection model, which includes a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of multiple forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability; therefore, after the image to be detected is input into the target data detection model, the predicted forgery result output by the target data detection model can be obtained. The target data detection model constructed by the present application can improve the adaptability of the power scenario by using the feature superposition at the time series level during the detection process, and use the predicted forgery results obtained by prediction to perform probability analysis, making it more scientific, thereby improving the practical application value.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a method, apparatus, storage medium, and related equipment for detecting forgery of power load data. Background Art

[0002] With the deepening of digital power grids, power load data has become a core resource for tasks such as power grid situational awareness, status evaluation, and control decision-making. The power grid has an increasingly high demand for data sets that can reflect the overall picture of the power grid. However, since the power grid has been in a stable operation state for a long time, most power data samples reflect stable scenarios and are insufficient to support forward-looking research such as event prediction, fault warning, and low-probability event plans. Therefore, forged data generated by power load data forgery technology is indispensable in the power system.

[0003] However, the generation of massive amounts of forged data is uncontrollable, introducing data with unexpected characteristics. Furthermore, most data labels lack human annotation and traceability, negatively impacting the power system. Therefore, the power system needs to develop forgery detection technology to verify the authenticity of power load data. Currently, forgery detection technology requires the development of a forgery detection model that is customized and trained based on the type of forged data. This results in low adaptability in power scenarios with high randomness, high complexity, and multiple influencing factors, limiting its practical application value. Summary of the Invention

[0004] The purpose of this application is to solve at least one of the above-mentioned technical defects, especially the technical defect that the counterfeit detection technology in the existing technology has low adaptability in power scenarios with high randomness, high complexity and multiple influencing factors, thereby increasing the cost of building a counterfeit detection model.

[0005] The present application provides a method for detecting forged power load data, the method comprising:

[0006] Acquiring power load data to be detected, and encoding the power load data to generate an image to be detected;

[0007] Determining a target data detection model, wherein the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of a plurality of forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability;

[0008] The image to be detected is input into the target data detection model to obtain a predicted forgery result corresponding to the power load data output by the target data detection model.

[0009] Optionally, encoding the power load data to generate an image to be detected includes:

[0010] Determine multiple timestamps in the power load data in chronological order, and obtain a load value corresponding to each timestamp;

[0011] A preset encoding method is used to encode information of each load value to obtain an encoding result, and the encoding result is visualized using a data visualization tool to generate an image to be detected.

[0012] Optionally, determining the target data detection model includes:

[0013] Acquiring sample load data, the sample load data including a plurality of to-be-detected images of different forgery types and a true forgery result corresponding to each to-be-detected image;

[0014] Inputting the sample load data into a preset initial data detection model to obtain a predicted forgery result output by the initial data detection model;

[0015] Training the initial data detection model with the goal of making the predicted forgery result close to the actual forgery result of the sample load data;

[0016] When the initial data detection model meets the preset training conditions, the trained initial data detection model is used as the target data detection model.

[0017] Optionally, obtaining sample load data includes:

[0018] Acquiring pre-collected real load data and converting the real load data into a real load image;

[0019] determining a plurality of forged load data forging models of different types, and inputting the true load image into each of the forged load data forging models respectively to obtain forged load images output by each of the forged load data forging models;

[0020] The real load image and each forged load image are used as images to be detected, and the real and forged results of each image to be detected are marked to form sample load data.

[0021] Optionally, the target data detection model further includes a true-false discrimination network;

[0022] Inputting the image to be detected into the target data detection model to obtain a predicted forgery result corresponding to the power load data output by the target data detection model includes:

[0023] Inputting the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtaining a predicted probability result output by the multi-target prediction sub-model;

[0024] Maximizing the average probability of the predicted probability results using the multi-objective optimization sub-model to obtain a final probability result;

[0025] The authenticity judgment network is used to judge the authenticity of the final probability result, and a prediction forgery result corresponding to the power load data is generated.

[0026] Optionally, the multi-objective prediction sub-model includes a data form conversion network, a feature decomposition network and a feature-to-time series network;

[0027] Inputting the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtaining the predicted probability result output by the multi-target prediction sub-model, includes:

[0028] Using the data form conversion network to perform data forging on the image to be detected, and obtaining forged data corresponding to each forgery type of the image to be detected;

[0029] Decomposing the image to be detected and each forged data using the feature decomposition network to obtain image feature data corresponding to the image to be detected and each forged data, and converting each image feature data into time series feature data using the data form conversion network;

[0030] The feature-to-time series network is used to cross-superimpose the various time series feature data to obtain multiple time series superposition data, and the probability value of the image to be detected belonging to each forgery type is determined based on each time series superposition data to form the prediction probability result output by the multi-target prediction sub-model.

[0031] Optionally, the time series feature data includes power load features, basic model features, superimposed model features and custom model features;

[0032] The feature-to-time series network is used to cross-superimpose various time series feature data to obtain multiple time series superposition data, including:

[0033] Cross-superimposing the basic model features, superimposed model features, and custom model features in each time series feature data based on the power load features of the image to be detected to obtain first time series superimposed data;

[0034] Cross-superimpose the power load features, superposition model features, and custom model features in each time series feature data based on the basic model features of the image to be detected, to obtain second time series superposition data;

[0035] Cross-superimpose the power load features, basic model features, and custom model features in each time series feature data based on the superimposed model features of the image to be detected to obtain third time series superimposed data;

[0036] The power load features, basic model features and superposition model features in each time series feature data are cross-superimposed based on the custom model features of the image to be detected to obtain fourth time series superposition data.

[0037] The present application also provides a device for detecting forged power load data, comprising:

[0038] An information encoding module is used to obtain the power load data to be detected, and to perform information encoding on the power load data to generate an image to be detected;

[0039] a model determination module, configured to determine a target data detection model, wherein the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is configured to predict the forgery probability corresponding to each of a plurality of forgery types, and the multi-objective optimization sub-model is configured to maximize the average probability of each forgery probability;

[0040] The model prediction module is used to input the image to be detected into the target data detection model to obtain a predicted forgery result corresponding to the power load data output by the target data detection model.

[0041] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the power load data forgery detection method as described in any of the above embodiments.

[0042] The present application also provides a computer device, comprising: one or more processors, and a memory;

[0043] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the power load data forgery detection method as described in any one of the above embodiments are performed.

[0044] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0045] The power load data forgery detection method, device, storage medium and related equipment provided by the present application, when the user needs to detect the authenticity of the power load data, can first obtain the power load data to be detected, and encode the information of the power load data to generate an image to be detected, so as to better capture the time series characteristics in the power load data through the image to be detected, thereby improving the detection accuracy; then the target data detection model can be determined, and the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of the multiple forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability; therefore, after the image to be detected is input into the target data detection model, the predicted forgery result corresponding to the power load data output by the target data detection model can be obtained. Through the target data detection model, the present application can improve the adaptability of the power scenario by superimposing features at the time series level during the detection process, and use the predicted forgery results obtained by prediction to perform probability analysis, making it more scientific, thereby improving the practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] 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 only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0047] Figure 1 A flowchart of a method for detecting forged power load data provided in an embodiment of the present application;

[0048] Figure 2 A flowchart of a target data detection model determination process provided in an embodiment of the present application;

[0049] Figure 3 A flowchart of a target data detection model detection process provided in an embodiment of the present application;

[0050] Figure 4 A schematic diagram of a flow chart of a prediction process of a multi-objective prediction sub-model provided in an embodiment of the present application;

[0051] Figure 5 A schematic diagram of the structure of a power load data forgery detection device provided in an embodiment of the present application;

[0052] Figure 6 A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] 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.

[0054] The generation of massive amounts of forged data is uncontrollable, introducing data with unexpected characteristics. Furthermore, data labels are often lacking manual annotation and traceability, negatively impacting the power system. Therefore, the power system needs to develop forgery detection technology to verify the authenticity of power load data. Currently, forgery detection technology requires the development of a forgery detection model that is customized and trained based on the type of forged data. This approach is less suitable for power scenarios characterized by high randomness, high complexity, and multiple influencing factors, limiting its practical application value.

[0055] Based on this, this application proposes the following technical solutions, please refer to the following for details:

[0056] In one embodiment, Figure 1 As shown, Figure 1 This is a flow chart of a method for detecting forged power load data provided in an embodiment of the present application. The present application provides a method for detecting forged power load data, which specifically includes the following steps:

[0057] S110: Obtain power load data to be detected, and perform information encoding on the power load data to generate an image to be detected.

[0058] In this embodiment, when the user needs to verify the authenticity of the power load data, the user can select the corresponding data on the acquisition interface of the computer device so that the computer device can obtain the power load data to be detected, encode the information of the power load data, and generate an image to be detected, so that relevant detection operations can be performed on the image to be detected.

[0059] It is understandable that power load data refers to the power load that power supply equipment at all levels in the power system has to bear. When computer equipment needs to obtain relevant power load data, it can collect data through various power grid companies or open source data platforms. In the process of data collection, computer equipment can collect data according to different classification methods to improve the efficiency of power load data processing. For example, according to the time when the load occurs in the power system, it can be divided into peak load, minimum load, and average load; according to the degree of loss caused by sudden power outage, it can be divided into primary load, secondary load, and tertiary load. In this application, computer equipment mainly collects power load data according to time classification.

[0060] Specifically, after collecting the power load data, the computer equipment can also perform pre-processing operations on the power load data. The vectorization processing here includes but is not limited to denoising, data cleaning, feature extraction and dimensionality reduction, and data normalization. Among them, denoising refers to removing noise and interference in the original data to improve the quality and availability of the data; data cleaning refers to removing outliers, supplementing missing values, and smoothing the original data to improve the accuracy and completeness of the data; feature extraction and dimensionality reduction refer to extracting effective information from the original data and converting the extracted effective information into a low-dimensional vectorized form for subsequent mathematical operations and analysis; and data normalization refers to normalizing the original data to ensure that the weights of data with different features are the same to avoid data bias and errors. Therefore, the authenticity and accuracy of the image to be detected generated by the computer equipment after encoding the power load data are high.

[0061] Furthermore, after generating the image to be detected, the computer device can further preprocess the image to improve its clarity. However, unlike data preprocessing, the preprocessing operations here include sharpening, denoising, and other operations. Specifically, sharpening refers to compensating for image contours, enhancing image edges and grayscale transitions, and making the image clearer. It can be categorized into spatial domain processing and frequency domain processing. By highlighting the edges and contours of elements in the image, or the features of certain linear target elements, the contrast between the element edges and surrounding pixels is increased. Denoising refers to the process of reducing noise in digital images. Generally, the digitization and transmission of images are often affected by interference from the imaging device and the external environment. That is, the received image information generally contains noise, which can be a significant source of image interference. Denoising can remove the noise in the image, further improving the authenticity and accuracy of the resulting image.

[0062] S120: Determine a target data detection model.

[0063] In this embodiment, after the image to be detected is generated in step S110, the computer device can determine a target data detection model for detecting the image to be detected, so that the target data detection model can be used to detect the authenticity of the image to be detected.

[0064] Specifically, the target data detection model of this application includes a multi-objective prediction sub-model and a multi-objective optimization sub-model. There are many types of forged data, and each type has significantly different characteristics. Therefore, the computer device can use the multi-objective prediction sub-model to predict the probability that the image under inspection belongs to a forgery type based on the forgery type. After obtaining the forgery probabilities corresponding to each forgery type, the computer device uses the multi-objective optimization sub-model to maximize the average probability of each forgery probabilities. This allows the predicted forgery results generated by the model to more quickly and accurately converge with the actual forgery results.

[0065] Maximizing the probability average refers to determining the probability distribution of multiple events by adjusting these probability values to maximize their weighted average given a set of probability values, thereby achieving a specific optimal goal. Therefore, this application can use maximizing the probability average to improve the accuracy of predicting forged results.

[0066] It is understandable that the multi-target prediction sub-model in the target data detection model is reproducible and specific. Therefore, through the multi-target prediction sub-model, forgery features that are unrelated to the power load features can be extracted from the image to be detected, and then the predicted forgery results can be calculated. However, considering the mixed superposition between customized forgery detection models and the incomplete decoupling between model features and power load features caused by factors such as mixed superposition, the predicted forgery results output by the target data detection model constructed in this application not only give the prediction result of whether the power load data is forged data, but also give the probability results of the power load data belonging to each forgery type and the probability results of the combination of forgery types, making it more scientific and more practical.

[0067] Furthermore, the target data detection model in the present application can be pre-stored in a computer device, so that when the authenticity of the power load data needs to be detected later, the computer device can directly call the pre-stored target data detection model to perform the detection operation on the power load data. In addition, the target data detection model in the present application can select a neural network model as the initial model for improvement and training to obtain a target data detection model that can detect the authenticity of the image to be detected.

[0068] Model.

[0069] S130: Inputting the image to be detected into the target data detection model to obtain the predicted forgery result corresponding to the power load data output by the target data detection model.

[0070] In this embodiment, after determining the target data detection model through step S120, the computer device can input the image to be detected into the target data detection model, and detect the image to be detected through the target data detection model, thereby obtaining the predicted forgery result corresponding to the power load data output by the target data detection model after detection.

[0071] Specifically, after the present application inputs the image to be detected into the target data detection model, the target data detection model can use the multi-objective prediction sub-model to extract the forgery features of the image to be detected, and cross-superimpose the extracted forgery features, and then determine the probability value of the image to be detected belonging to each forgery type through the cross-superposition result. Then, the target data detection model can use the multi-objective optimization sub-model to maximize the probability average of each probability value, and then output the predicted forgery result corresponding to the power load data based on the maximization result.

[0072] In the above embodiment, when the user needs to detect the authenticity of the power load data, the power load data to be detected can be obtained first, and the power load data can be encoded to generate an image to be detected, so as to better capture the time series features in the power load data through the image to be detected and improve the detection accuracy; then the target data detection model can be determined, and the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of the multiple forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability; therefore, after the image to be detected is input into the target data detection model, the predicted forgery result corresponding to the power load data output by the target data detection model can be obtained. Through the target data detection model, the present application can improve the adaptability of the power scene by superimposing features at the time series level during the detection process, and use the predicted forgery results obtained by prediction to perform probability analysis, making it more scientific, thereby improving the practical application value.

[0073] In one embodiment, encoding the power load data to generate an image to be detected in step S110 may include:

[0074] S111: Determine multiple timestamps in the power load data in chronological order, and obtain a load value corresponding to each timestamp.

[0075] S112: Encode information of each load value using a preset encoding method to obtain an encoding result, and visualize the encoding result using a data visualization tool to generate an image to be detected.

[0076] In this embodiment, when the computer device converts the power load data into the image to be detected, it can first determine multiple timestamps in the power load data in chronological order, and obtain the load value corresponding to each timestamp, so that the preset encoding method can be used to encode the information of each load value to obtain the encoding result, and the encoding result can be visualized through a data visualization tool to generate the image to be detected.

[0077] It is understood that since power load data is collected in a time series, the computer device can first determine multiple timestamps based on a preset time window before encoding the power load data, and then obtain the load value corresponding to each timestamp in the power load data. The computer device can then select a corresponding preset encoding method based on the detection requirements of the model and use the preset encoding method to encode the information of each load value to obtain an encoded result. The preset encoding method here can be a line chart, heat map, bar chart, etc., without limitation.

[0078] Specifically, when the preset encoding method is a line graph, the computer device can convert the power load data into a line graph with time as the horizontal axis and the load value as the vertical axis. When the preset encoding method is a heat map, the computer device can convert the power load data into a two-dimensional heat map with time as the horizontal axis, the load level as the vertical axis, and the load value as the matrix value. When the preset encoding method is a bar graph, the computer device can divide the power load data into time periods, calculate the average load data value for each time period, and then express the average value as the height of the bar graph to obtain a bar graph. Finally, the computer device can visualize the graph data using a data visualization tool to generate an image to be detected.

[0079] Furthermore, after the computer equipment generates the image to be detected, it can also adjust and optimize the image to be detected, such as adjusting the color scheme, line thickness, image size and resolution, etc., to ensure that the image is clear and easy to read and can accurately convey the characteristics and trends of the power load data.

[0080] In one embodiment, Figure 2 As shown, Figure 2 A flowchart of a target data detection model determination process provided in an embodiment of the present application; Figure 2 In step S120, determining the target data detection model may include:

[0081] S121: Acquire sample load data, where the sample load data includes a plurality of to-be-detected images of different forgery types and a true forgery result corresponding to each to-be-detected image.

[0082] S122: Input the sample load data into a preset initial data detection model to obtain a predicted forgery result output by the initial data detection model.

[0083] S123: Training the initial data detection model with the goal of predicting forged results close to the actual forged results of the sample load data.

[0084] S124: When the initial data detection model meets the preset training conditions, the trained initial data detection model is used as the target data detection model.

[0085] In this embodiment, when determining a target data detection model, the computer device may first obtain sample load data, where the sample load data includes multiple images to be detected of different forgery types and the actual forgery results corresponding to each image to be detected. Therefore, after inputting the sample load data into a preset initial data detection model and obtaining the predicted forgery results output by the initial data detection model, the computer device may train the initial data detection model with the goal of ensuring that the predicted forgery results approximate the actual forgery results of the sample load data. When the initial data detection model meets preset training conditions, the trained initial data detection model is used as the target data detection model.

[0086] It should be noted that the images to be detected in the sample load data may include not only images to be detected converted from forged data of different forgery types, but also images to be detected converted from power load data collected from the power system, and there is no limitation here.

[0087] It is understood that the target data detection model of the present application refers to a model that detects the input image to be detected and obtains the forgery result. When the target data detection model is trained, the images to be detected of different forgery types can be used as training samples, and each training sample is labeled with a sample label, that is, the corresponding true forgery result. After all training samples are labeled, the training samples with sample labels can be input into the preset initial data detection model for forward propagation to train the model. In the reverse propagation process of the model, the preset target loss function is used to tune the model parameters. When the model meets certain training conditions or parameter convergence conditions, such as the number of iterations reaches the set value, the training is considered complete. At this time, the trained model can be used as the final target data detection model.

[0088] Of course, before inputting the sample load data into the preset initial cutout model, the present application can also pre-process the image to be detected, such as normalization, sharpening, denoising and other operations, so as to effectively improve the clarity of the image to be detected itself, thereby improving the efficiency of model training.

[0089] In one embodiment, obtaining sample load data in step S121 may include:

[0090] S1211: Acquire pre-collected real load data and convert the real load data into a real load image.

[0091] S1212: Determine multiple load data forgery models of forgery types, and input the real load image into each load data forgery model respectively to obtain forged load images output by each load data forgery model.

[0092] S1213: The real load image and each forged load image are used as images to be detected, and the real or forged result of each image to be detected is marked to form sample load data.

[0093] In this embodiment, when obtaining sample load data, the computer device can first obtain the real load data collected in advance and convert the real load data into a real load image; then, it can determine multiple forged types of load data forgery models, and input the real load image into each load data forgery model respectively to obtain the forged load image output by each load data forgery model; finally, the computer device can use the real load image and each forged load image as the images to be detected, and mark the real forgery result of each image to be detected, thereby forming sample load data.

[0094] It is understandable that when a computer device uses real load data to generate forged load data, it can first determine multiple forged types of load data forgery models based on actual needs or detection purposes, so that these load data forgery models can simulate different forgery means or attack methods, such as random noise addition, trend modification, periodic interference, etc. to generate forged load data.

[0095] Specifically, the load data forgery model in this application can select a neural network model as a preset model for improvement and training. The neural network model can be a basic model such as a generative adversarial network, a deep convolutional network, an autoencoder, or a hybrid superposition model of different basic models, or a customized model that is customized and improved from the above models, without limitation. Specifically, for each forgery type, the computer device can pre-acquire real load data and forged load data of the corresponding forgery type obtained by expanding the real load image, and then use this data to train the preset initial model, thereby obtaining and storing a load data forgery model of the forgery type.

[0096] Therefore, when the computer equipment needs to falsify the real load data, it can first convert the real load image of the real load data, and directly call the pre-stored load data falsification models of multiple falsification types, and then input the real load image into each load data falsification model respectively to obtain the falsified load image output by each load data falsification model.

[0097] In one embodiment, Figure 3 As shown, Figure 3 A flowchart of a target data detection model detection process provided in an embodiment of the present application; Figure 3 In step S130, the target data detection model may further include a true / false discrimination network; wherein, inputting the image to be detected into the target data detection model and obtaining the predicted forgery result corresponding to the power load data output by the target data detection model may include:

[0098] S131: Input the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtain the predicted probability result output by the multi-target prediction sub-model.

[0099] S132: Utilize the multi-objective optimization sub-model to maximize the probability average of the predicted probability results to obtain the final probability result.

[0100] S133: The authenticity of the final probability result is judged by the authenticity judgment network to generate a predicted forged result corresponding to the power load data.

[0101] In this embodiment, the target data detection model also includes a true-false discrimination network. Therefore, the computer device can first input the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtain the predicted probability result output by the multi-target prediction sub-model. Then, the multi-target optimization sub-model can be used to maximize the probability average value of the predicted probability result to obtain the final probability result. Finally, the computer device can perform true-false discrimination on the final probability result through the true-false discrimination network to generate a predicted forgery result corresponding to the power load data.

[0102] Specifically, the multi-target prediction sub-model can predict the probability value of the image to be tested belonging to each preset forgery type. It includes multiple output nodes, one for each forgery type, and the output value of the output node represents the probability of the image to be tested belonging to the corresponding forgery type. Therefore, the output of the multi-target prediction sub-model is a series of probability values, which reflect the model's judgment on whether the image to be tested belongs to a certain forgery type. The multi-target optimization sub-model can further process the predicted probability results of the multi-target prediction sub-model, with the goal of maximizing the average value of the predicted probability results. In this way, the probability values of multiple forgery types can be integrated to obtain a more comprehensive and accurate judgment.

[0103] Furthermore, the authenticity discrimination network, as a binary classifier, can determine whether the image being tested is real or forged based on the final probability result. If the image being tested is real, the authenticity discrimination network directly outputs a predicted forgery result. If the image being tested is forged, the authenticity discrimination network outputs a predicted forgery result that includes the final probability result, allowing users to effectively identify forgeries in power load data, thereby enhancing its practical application value.

[0104] In one embodiment, Figure 4 As shown, Figure 4 A schematic diagram of a flow chart of a prediction process of a multi-objective prediction sub-model provided in an embodiment of the present application; Figure 4 In step S131, the multi-target prediction sub-model may include a data form conversion network, a feature decomposition network, and a feature-to-time series network; wherein, the image to be detected is input into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and the prediction probability result output by the multi-target prediction sub-model may include:

[0105] S1311: Using a data form conversion network to perform data forging on the image to be detected, and obtain forged data corresponding to each forgery type of the image to be detected.

[0106] S1312: Decompose the image to be detected and each forged data using a feature decomposition network to obtain image feature data corresponding to the image to be detected and each forged data, and convert each image feature data into time series feature data through a data form conversion network.

[0107] S1313: Use the feature-to-time series network to cross-superimpose each time series feature data to obtain multiple time series superposition data, and determine the probability value of the image to be detected belonging to each forgery type based on each time series superposition data, forming a prediction probability result output by the multi-target prediction sub-model.

[0108] In this embodiment, the multi-target prediction sub-model may include a data form conversion network, a feature decomposition network and a feature-to-time series network. Therefore, the computer device can first use the data form conversion network to perform data forgery on the image to be detected to obtain the forged data corresponding to each forgery type of the image to be detected. Then, the feature decomposition network can be used to perform feature decomposition on the image to be detected and each forged data to obtain image feature data corresponding to the image to be detected and each forged data, and each image feature data can be converted into time series feature data through the data form conversion network. Finally, the computer device can use the feature-to-time series network to cross-superimpose each time series feature data to obtain multiple time series superposition data, and determine the probability value of the image to be detected belonging to each forgery type based on each time series superposition data to form the prediction probability result output by the multi-target prediction sub-model.

[0109] It should be noted that the data format conversion network can convert data between time series and image formats. Therefore, in one application scenario, if a computer directly inputs time series power load data into the target data detection model, the data format conversion network can first convert the power load data from time series format to image format, i.e., the image to be detected, before further detection operations are performed.

[0110] It is understandable that most data detection models in the prior art can only detect data in the form of images or videos, and cannot directly detect data in the form of time series. Therefore, the target data detection model built by this application can use a data form conversion network to convert power load data from time series form to image form for feature extraction, and restore the extracted feature image to time series form before information fusion. In this way, the target data detection model of this application can improve the adaptability of power scenarios.

[0111] Specifically, the feature decomposition network can perform feature decomposition on the image to be detected and each piece of forged data to obtain image feature data. This image feature data can also be subdivided into multiple model features to express it, thereby avoiding large differences in model features for the same forgery type, which in turn affects the subsequent feature superposition effect. The computer device can then convert the decomposed image feature data into time series feature data through a data format conversion network, so that the feature-to-time series network can cross-superimpose each piece of time series feature data at the time series level to obtain multiple time series superposition data. In this way, the probability value corresponding to each forgery type calculated based on each time series superposition data can be closer to the truth.

[0112] In one embodiment, the time series feature data in step S1313 may include power load features, basic model features, superposition model features, and custom model features; wherein, the time series feature data are cross-superimposed using a feature-to-time series network to obtain multiple time series superposition data, which may include:

[0113] S3131: Cross-superimpose the basic model features, superposition model features, and user-defined model features in each time series feature data based on the power load features of the image to be detected to obtain first time series superposition data.

[0114] S3132: Based on the basic model features of the image to be detected, the power load features, the superposition model features and the custom model features in each time series feature data are cross-superimposed to obtain second time series superposition data.

[0115] S3133: Cross-superimpose the power load features, basic model features, and custom model features in each time series feature data based on the superimposed model features of the image to be detected to obtain third time series superimposed data.

[0116] S3134: Cross-superimpose the power load features, basic model features, and superposition model features in each time series feature data based on the custom model features of the image to be detected to obtain fourth time series superposition data.

[0117] In this embodiment, the time series feature data obtained by decomposing the target data detection model may include power load features, basic model features, superimposed model features, and custom model features. Therefore, when cross-superimposing the various time series feature data, the basic model features, superimposed model features, and custom model features in the various time series feature data may be cross-superimposed based on the power load features of the image to be detected to obtain first time series superimposed data, and the power load features, superimposed model features, and custom model features in the various time series feature data may be cross-superimposed based on the basic model features of the image to be detected to obtain second time series superimposed data, the power load features, basic model features, and custom model features in the various time series feature data may be cross-superimposed based on the superimposed model features of the image to be detected to obtain third time series superimposed data, and the power load features, basic model features, and superimposed model features in the various time series feature data may be cross-superimposed based on the custom model features of the image to be detected to obtain fourth time series superimposed data.

[0118] In a specific implementation, a computer device can decompose a target data detection model to obtain n sets of time series feature data, specifically including power load features de_n, basic model features db_n, superposition model features da_n, and custom model features dc_n. The computer can first use the n sets of time series feature data for cross-superposition to obtain an n*n*n*n superposition time series form: de_w+db_x+da_y+dc_z, where w=1, 2, ..., n; x=1, 2, ..., n; y=1, 2, ..., n; z=1, 2, ..., n. The computer device can then constrain the superposition time series form by minimizing the connection time series classification loss, thereby forming four different sets of time series superposition data, specifically including the following:

[0119] (1) Based on the power load characteristics of the image to be detected, the basic model features, superposition model features and custom model features in each time series feature data are cross-superimposed to obtain the first time series superposition data: d1 = de_w + db_x + da_y + dc_z, where x = 1, 2, ..., n; y = 1, 2, ..., n; z = 1, 2, ..., n; so that the de_w obtained by processing the target data detection model is as close as possible to the original de_w, and then the probability η1 of complete closeness is obtained.

[0120] (2) Based on the basic model features of the image to be detected, the power load features, superposition model features and custom model features in each time series feature data are cross-superimposed to obtain the second time series superposition data: d2 = de_w + db_x + da_y + dc_z, where w = 1, 2, ..., n; y = 1, 2, ..., n; z = 1, 2, ..., n; and then the probability η2 that d2 is classified as the basic model corresponding to db_x is obtained.

[0121] (3) Based on the superposition model features of the image to be detected, the power load features, basic model features, and custom model features in each time series feature data are cross-superimposed to obtain the third time series superposition data: d3 = de_w + db_x + da_y + dc_z, where w = 1, 2, ..., n; x = 1, 2, ..., n; z = 1, 2, ..., n; and the probability η3 that d3 is classified as the superposition model corresponding to db_y is obtained.

[0122] (4) Based on the custom model features of the image to be detected, the power load features, basic model features and superposition model features in each time series feature data are cross-superimposed to obtain the fourth time series superposition data: d4 = de_w + db_x + da_y + dc_z, where w = 1, 2, ..., n; x = 1, 2, ..., n; y = 1, 2, ..., n; and then the probability η4 that d4 is classified as the custom model corresponding to db_z is obtained.

[0123] The following describes an apparatus for detecting forged power load data provided in an embodiment of the present application. The apparatus for detecting forged power load data described below and the method for detecting forged power load data described above may refer to each other.

[0124] In one embodiment, Figure 5 As shown, Figure 5 This is a schematic diagram of the structure of a power load data forgery detection device provided in an embodiment of the present application. The present application also provides a power load data forgery detection device, including an information encoding module 210, a model determination module 220, and a model prediction module 230, specifically including the following:

[0125] The information encoding module 210 is used to obtain the power load data to be detected, and perform information encoding on the power load data to generate an image to be detected.

[0126] The model determination module 220 is used to determine the target data detection model. The target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model. The multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of multiple forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability.

[0127] The model prediction module 230 is used to input the image to be detected into the target data detection model to obtain the predicted forged result corresponding to the power load data output by the target data detection model.

[0128] In the above embodiment, when the user needs to detect the authenticity of the power load data, the power load data to be detected can be obtained first, and the power load data can be encoded to generate an image to be detected, so as to better capture the time series features in the power load data through the image to be detected and improve the detection accuracy; then the target data detection model can be determined, and the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of the multiple forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability; therefore, after the image to be detected is input into the target data detection model, the predicted forgery result corresponding to the power load data output by the target data detection model can be obtained. Through the target data detection model, the present application can improve the adaptability of the power scene by superimposing features at the time series level during the detection process, and use the predicted forgery results obtained by prediction to perform probability analysis, making it more scientific, thereby improving the practical application value.

[0129] In one embodiment, the information encoding module 210 may include:

[0130] The value determination submodule is used to determine multiple time stamps in the power load data in chronological order and obtain the load value corresponding to each time stamp.

[0131] The numerical coding submodule is used to encode the information of each load value using a preset coding method to obtain the coding result, and to visualize the coding result through a data visualization tool to generate an image to be detected.

[0132] In one embodiment, the model determination module 220 may include:

[0133] The data acquisition submodule is used to acquire sample load data, where the sample load data includes a plurality of images to be detected of different forgery types and a true forgery result corresponding to each image to be detected.

[0134] The model prediction submodule is used to input the sample load data into the preset initial data detection model to obtain the predicted forgery result output by the initial data detection model.

[0135] The model training submodule is used to train the initial data detection model with the goal of predicting forged results close to the real forged results of the sample load data.

[0136] The model generation submodule is used to use the trained initial data detection model as the target data detection model when the initial data detection model meets the preset training conditions.

[0137] In one embodiment, the data acquisition submodule may include:

[0138] The data conversion unit is used to obtain the real load data collected in advance and convert the real load data into a real load image.

[0139] The image forging unit is used to determine multiple forging types of load data forging models, and input the real load image into each load data forging model respectively to obtain the forged load image output by each load data forging model.

[0140] The data forming unit is used to take the real load image and each forged load image as the images to be detected, and mark the real and forged results of each image to be detected to form sample load data.

[0141] In one embodiment, the model prediction module 230 may include:

[0142] The probability prediction submodule is used to input the image to be detected into the multi-target prediction submodel to predict the probability value of the image to be detected belonging to each forgery type, and obtain the predicted probability result output by the multi-target prediction submodel.

[0143] The probability maximization submodule is used to maximize the probability average of the predicted probability results using the multi-objective optimization submodel to obtain the final probability result.

[0144] The authenticity discrimination submodule is used to perform authenticity discrimination on the final probability result through the authenticity discrimination network and generate a predicted forged result corresponding to the power load data.

[0145] In one embodiment, the probability prediction submodule may include:

[0146] The data forging unit is used to forge data of the image to be detected by using a data form conversion network to obtain forged data corresponding to each forgery type of the image to be detected.

[0147] The image conversion unit is used to perform feature decomposition on the image to be detected and each forged data using a feature decomposition network to obtain image feature data corresponding to the image to be detected and each forged data, and convert each image feature data into time series feature data through a data form conversion network.

[0148] The result output unit is used to cross-superimpose various time series feature data using the feature-to-time series network to obtain multiple time series superposition data, and determine the probability value of the image to be detected belonging to each forgery type based on each time series superposition data, forming the prediction probability result output by the multi-target prediction sub-model.

[0149] In one embodiment, the result output unit may include:

[0150] The first cross-overlap subunit is configured to cross-overlap the basic model features, the superposition model features, and the user-defined model features in each time series feature data based on the power load features of the image to be detected, to obtain first time series superposition data.

[0151] The second cross-overlap subunit is configured to cross-overlap the power load features, the overlay model features and the custom model features in each time series feature data based on the basic model features of the image to be detected to obtain second time series overlay data.

[0152] The third cross-overlap subunit is configured to cross-overlap the power load features, basic model features, and user-defined model features in each time series feature data based on the overlay model features of the image to be detected to obtain third time series overlay data.

[0153] The fourth cross-overlap subunit is configured to cross-overlap the power load features, basic model features, and superposition model features in each time series feature data based on the custom model features of the image to be detected, to obtain fourth time series superposition data.

[0154] In one embodiment, the present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the power load data forgery detection method as described in any of the above embodiments.

[0155] In one embodiment, the present application also provides a computer device having computer-readable instructions stored therein. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the power load data forgery detection method as described in any of the above embodiments.

[0156] Schematically, as Figure 6 As shown, Figure 6 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 can be provided as a server. Figure 6 Computer device 300 includes a processing component 302, which further includes one or more processors, and a memory resource represented by memory 301 for storing instructions executable by processing component 302, such as an application. The application stored in memory 301 may include one or more modules, each corresponding to a set of instructions. Furthermore, processing component 302 is configured to execute the instructions to perform the power load data forgery detection method according to any of the above-described embodiments.

[0157] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server™, Mac OS X™, Unix™, Linux™, Free BSD™, or the like.

[0158] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0159] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only 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. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0160] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referenced to each other.

[0161] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting forged power load data, characterized in that: The method comprises: Acquiring power load data to be detected, and encoding the power load data to generate an image to be detected; Determining a target data detection model, wherein the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is used to predict the forgery probability corresponding to each of a plurality of forgery types, and the multi-objective optimization sub-model is used to maximize the probability average of each forgery probability; Inputting the image to be detected into the target data detection model to obtain a predicted forgery result corresponding to the power load data output by the target data detection model; Wherein, the target data detection model also includes a true-false discrimination network; Inputting the image to be detected into the target data detection model to obtain a predicted forgery result corresponding to the power load data output by the target data detection model includes: Inputting the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtaining a predicted probability result output by the multi-target prediction sub-model; Maximizing the average probability of the predicted probability results using the multi-objective optimization sub-model to obtain a final probability result; Performing authenticity discrimination on the final probability result through the authenticity discrimination network to generate a prediction forgery result corresponding to the power load data; The multi-objective prediction sub-model includes a data form conversion network, a feature decomposition network and a feature conversion time series network; Inputting the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtaining the predicted probability result output by the multi-target prediction sub-model, includes: Using the data form conversion network to perform data forging on the image to be detected, and obtaining forged data corresponding to each forgery type of the image to be detected; Decomposing the image to be detected and each forged data using the feature decomposition network to obtain image feature data corresponding to the image to be detected and each forged data, and converting each image feature data into time series feature data using the data form conversion network; The feature-to-time series network is used to cross-superimpose the various time series feature data to obtain multiple time series superposition data, and the probability value of the image to be detected belonging to each forgery type is determined based on each time series superposition data to form the prediction probability result output by the multi-target prediction sub-model.

2. The method for detecting forged power load data according to claim 1, characterized in that: The step of encoding the power load data to generate an image to be detected includes: Determine multiple timestamps in the power load data in chronological order, and obtain a load value corresponding to each timestamp; A preset encoding method is used to encode information of each load value to obtain an encoding result, and the encoding result is visualized using a data visualization tool to generate an image to be detected.

3. The method for detecting forged power load data according to claim 1, wherein: Determining the target data detection model includes: Acquiring sample load data, the sample load data including a plurality of to-be-detected images of different forgery types and a true forgery result corresponding to each to-be-detected image; Inputting the sample load data into a preset initial data detection model to obtain a predicted forgery result output by the initial data detection model; Training the initial data detection model with the goal of making the predicted forgery result close to the actual forgery result of the sample load data; When the initial data detection model meets the preset training conditions, the trained initial data detection model is used as the target data detection model.

4. The method for detecting forged power load data according to claim 3, wherein: The obtaining of sample load data includes: Acquiring pre-collected real load data and converting the real load data into a real load image; determining a plurality of forged load data forging models of different types, and inputting the true load image into each of the forged load data forging models respectively to obtain forged load images output by each of the forged load data forging models; The real load image and each forged load image are used as images to be detected, and the real and forged results of each image to be detected are marked to form sample load data.

5. The method for detecting forged power load data according to claim 1, wherein: The time series feature data includes power load features, basic model features, superimposed model features and custom model features; The feature-to-time series network is used to cross-superimpose various time series feature data to obtain multiple time series superposition data, including: Cross-superimposing the basic model features, superimposed model features, and custom model features in each time series feature data based on the power load features of the image to be detected to obtain first time series superimposed data; Cross-superimpose the power load features, superposition model features, and custom model features in each time series feature data based on the basic model features of the image to be detected, to obtain second time series superposition data; Cross-superimpose the power load features, basic model features, and custom model features in each time series feature data based on the superimposed model features of the image to be detected to obtain third time series superimposed data; The power load features, basic model features and superposition model features in each time series feature data are cross-superimposed based on the custom model features of the image to be detected to obtain fourth time series superposition data.

6. A device for detecting forgery of power load data, characterized in that: include: An information encoding module is used to obtain the power load data to be detected, and to perform information encoding on the power load data to generate an image to be detected; a model determination module, configured to determine a target data detection model, wherein the target data detection model is composed of a multi-objective prediction sub-model and a multi-objective optimization sub-model, wherein the multi-objective prediction sub-model is configured to predict the forgery probability corresponding to each of a plurality of forgery types, and the multi-objective optimization sub-model is configured to maximize the average probability of each forgery probability; a model prediction module, configured to input the image to be detected into the target data detection model, and obtain a predicted forgery result output by the target data detection model corresponding to the power load data; The target data detection model further includes a true / false discrimination network; and the model prediction module includes: Inputting the image to be detected into the multi-target prediction sub-model to predict the probability value of the image to be detected belonging to each forgery type, and obtaining a predicted probability result output by the multi-target prediction sub-model; Maximizing the average probability of the predicted probability results using the multi-objective optimization sub-model to obtain a final probability result; Performing authenticity discrimination on the final probability result through the authenticity discrimination network to generate a prediction forgery result corresponding to the power load data; The multi-objective prediction sub-model includes a data form conversion network, a feature decomposition network and a feature conversion time series network; The model prediction module also includes: Using the data form conversion network to perform data forging on the image to be detected, and obtaining forged data corresponding to each forgery type of the image to be detected; Decomposing the image to be detected and each forged data using the feature decomposition network to obtain image feature data corresponding to the image to be detected and each forged data, and converting each image feature data into time series feature data using the data form conversion network; The feature-to-time series network is used to cross-superimpose the various time series feature data to obtain multiple time series superposition data, and the probability value of the image to be detected belonging to each forgery type is determined based on each time series superposition data to form the prediction probability result output by the multi-target prediction sub-model.

7. A storage medium, characterized in that: The storage medium stores computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the method for detecting forgery of power load data as claimed in any one of claims 1 to 5.

8. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, execute the steps of the method for detecting forgery of power load data according to any one of claims 1 to 5.

Images