Data verification model generation methods, apparatus, computer equipment, and storage media

Abstract

This application relates to a method, apparatus, computer device, and storage medium for generating a data verification model. The method includes: acquiring a target group segmentation model and user historical data; inputting the user historical data into the target group segmentation model; segmenting users into target user groups based on the target group segmentation model; performing unsupervised training on the user historical data corresponding to the target user groups to generate a first target model; inputting the user historical data corresponding to the target user groups into the first target model to obtain first intermediate anomaly data; acquiring second intermediate data; training an initial classification model based on the first intermediate anomaly data and the second intermediate data to obtain a second target model; and sequentially concatenating and combining the target group segmentation model, the first target model, and the second target model to generate a target verification model. This method can improve the efficiency of data quality verification.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a data verification model generation method, apparatus, computer device, and storage medium. Background Technology

[0002] With the development of computer technology, intelligent verification technology based on artificial intelligence and machine learning algorithms is being applied to more fields. In the conventional field of power grid data verification, conventional electricity bill verification mainly involves checking the electricity bill results after the billing is completed. As the volume of power data continues to increase, the amount of electricity bill verification data that business personnel need to process simultaneously is also constantly increasing. Limited by the existing accounting model, the current electricity bill verification process only begins after the electricity billing is completed. As a result, many problems that occur during the work order process are not resolved in a timely manner, leading to a high rate of anomaly triggering.

[0003] It is evident that the increasing volume of electricity data in traditional technologies has led to a growing number of process problems, resulting in a high rate of anomalies caused by data issues. This has a significant impact on system resources and the workload of verification, ultimately leading to low efficiency in data quality verification. Summary of the Invention

[0004] Therefore, it is necessary to provide a data verification model generation method, apparatus, computer equipment, and storage medium that can verify data quality, thereby improving the efficiency of data quality verification, in response to the above-mentioned technical problems.

[0005] A method for generating a data verification model, the method comprising:

[0006] Obtain the target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into groups based on the target group segmentation model to obtain the target user group;

[0007] Unsupervised training is performed on the user history data corresponding to the target user group to generate a first target model. The user history data corresponding to the target user group is then input into the first target model to obtain the first intermediate abnormal data.

[0008] Obtain second intermediate data, and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model;

[0009] The target group segmentation model, the first target model, and the second target model are sequentially connected and combined to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0010] In one embodiment, the user historical data is input into the target group segmentation model, and the users are segmented into groups based on the target group segmentation model to obtain the target user group, which includes:

[0011] Based on the target group segmentation model, the mean of each type of data in the user's historical data is calculated to obtain the target mean of each type of data.

[0012] Calculate the difference between each type of data in the user's historical data and the corresponding target mean to obtain the first target difference;

[0013] Obtain the first target range value, and based on the comparison results between each of the first target differences and the first target range value, obtain the target user group.

[0014] In one embodiment, unsupervised training is performed on the historical user data corresponding to the target user group to generate a first target model, including:

[0015] Based on the historical user data corresponding to the target user group, a corresponding target distribution function is fitted, and the target distribution function is used as the first model.

[0016] Based on the historical user data corresponding to the target user group, a corresponding target density estimation function is fitted, and the target density estimation function is used as the second model.

[0017] A third model is obtained, which is used to obtain the corresponding abnormal data based on the comparison result between the maximum value of the input data and the input data.

[0018] The first target model is generated by combining the first model, the second model, and the third model in parallel.

[0019] In one embodiment, user historical data corresponding to the target user group is input into the first target model to obtain first intermediate abnormal data. The first target model is obtained by combining the first model, the second model, and the third model in parallel, including:

[0020] Based on the first curve corresponding to the first model, data outside the first curve is selected from the user historical data corresponding to the target user group as the first abnormal data;

[0021] Based on the second curve corresponding to the second model, data outside the second curve are selected from the user history data corresponding to the target user group as the second abnormal data;

[0022] Based on the third model in the first target model, calculate the maximum value of each type of data in the user history data corresponding to the target user group, and obtain the target maximum value corresponding to each type of data.

[0023] The differences between various types of historical data of users corresponding to the target user group and the corresponding target maximum values ​​are calculated respectively to obtain the second target difference;

[0024] Obtain the second target range value, and based on the comparison results between each second target difference and the second target range value, obtain the third abnormal data;

[0025] The first abnormal data, the second abnormal data, and the third abnormal data are merged to obtain the first intermediate abnormal data.

[0026] In one embodiment, obtaining second intermediate data, and training an initial classification model based on the first intermediate anomaly data and the second intermediate data to obtain a second target model includes:

[0027] The first intermediate abnormal data is filtered to obtain the first target abnormal data. The first target abnormal data is then fused with the second intermediate data to obtain the target data.

[0028] The initial classification model is trained based on the target data to obtain multiple intermediate classification models;

[0029] Obtain a test set, test the multiple intermediate classification models based on the test set, and obtain the evaluation index corresponding to the multiple intermediate classification models;

[0030] Based on the comparison results of the evaluation indicators corresponding to the multiple intermediate classification models, the target classification model is determined and used as the second target model.

[0031] In one embodiment, after sequentially combining the target group segmentation model, the first target model, and the second target model to generate the target verification model, the method further includes:

[0032] Obtain the target user data to be verified;

[0033] The target user data to be verified is input into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified.

[0034] In one embodiment, before obtaining the target user data to be verified, the method further includes:

[0035] A billing parameter dictionary is constructed based on the user's historical data;

[0036] Obtain the verification rules, and based on the verification rules, check and update each parameter in the billing parameter dictionary to obtain the target billing parameter dictionary;

[0037] Obtain the user data to be verified, and based on the comparison results between the user data to be verified and the target billing parameter dictionary, obtain the target user data to be verified.

[0038] A data verification model generation device, the device comprising:

[0039] The group segmentation module is used to obtain a target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into a group based on the target group segmentation model to obtain the target user group.

[0040] The first training module is used to perform unsupervised training on the user historical data corresponding to the target user group to generate a first target model, and input the user historical data corresponding to the target user group into the first target model to obtain the first intermediate abnormal data.

[0041] The second training module is used to acquire the second intermediate data, and to train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model;

[0042] The model generation module is used to sequentially combine the target group segmentation model, the first target model, and the second target model to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0043] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0044] Obtain the target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into groups based on the target group segmentation model to obtain the target user group;

[0045] Unsupervised training is performed on the user history data corresponding to the target user group to generate a first target model. The user history data corresponding to the target user group is then input into the first target model to obtain the first intermediate abnormal data.

[0046] Obtain second intermediate data, and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model;

[0047] The target group segmentation model, the first target model, and the second target model are sequentially connected and combined to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0048] A computer-readable storage medium having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0049] Obtain the target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into groups based on the target group segmentation model to obtain the target user group;

[0050] Unsupervised training is performed on the user history data corresponding to the target user group to generate a first target model. The user history data corresponding to the target user group is then input into the first target model to obtain the first intermediate abnormal data.

[0051] Obtain second intermediate data, and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model;

[0052] The target group segmentation model, the first target model, and the second target model are sequentially connected and combined to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0053] The aforementioned data verification model generation method, apparatus, computer equipment, and storage medium, through acquiring a target group segmentation model, acquiring user historical data, inputting the user historical data into the target group segmentation model, and segmenting each user into a target user group based on the target group segmentation model; performing unsupervised training on the user historical data corresponding to the target user group to generate a first target model; inputting the user historical data corresponding to the target user group into the first target model to obtain first intermediate abnormal data; acquiring second intermediate data; training an initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain a second target model; and sequentially combining the target group segmentation model, the first target model, and the second target model to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on user data to be verified. It divides users corresponding to historical user data into target user groups using a target group segmentation model. Unsupervised training is then performed on the historical user data corresponding to the target user groups to generate a first target model. Based on the first target model, anomaly data in the historical user data corresponding to the target user groups is identified as first intermediate anomaly data, and second intermediate data is obtained. Based on the first intermediate anomaly data and the second intermediate data, an initial classification model is trained to obtain a second target model. The target group segmentation model, the first target model, and the second target model are sequentially concatenated and combined to generate the target verification model, thus realizing the data anomaly verification process and improving the efficiency of data quality verification. Attached Figure Description

[0054] Figure 1 This is a diagram illustrating the application environment of a data verification model generation method in one embodiment.

[0055] Figure 2 This is a flowchart illustrating a data verification model generation method in one embodiment;

[0056] Figure 3 This is a schematic diagram of the group division process in one embodiment;

[0057] Figure 4 This is a schematic diagram of the process for generating the first target model in one embodiment;

[0058] Figure 5 This is a schematic diagram of the data acquisition process in one embodiment;

[0059] Figure 6 This is a schematic diagram of the process for generating the second target model in one embodiment;

[0060] Figure 7 This is a flowchart illustrating the process used by the model in one embodiment;

[0061] Figure 8 This is a schematic diagram of the data preprocessing process in one embodiment;

[0062] Figure 9 This is a schematic diagram of an intelligent verification system in one embodiment;

[0063] Figure 10 This is a structural block diagram of a data verification model generation device in one embodiment;

[0064] Figure 11 This is an internal structural diagram of a computer device in one embodiment;

[0065] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0066] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0067] The data verification model generation method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located on the cloud or other network servers. Terminal 102 is used to input historical user data and user data to be verified. Server 104 is used to generate a target verification model, which is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0068] In one embodiment, such as Figure 2 As shown, a data verification model generation method is provided, which can be applied to... Figure 1 Taking the server in the example, the following steps are included:

[0069] Step S200: Obtain the target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into a group based on the target group segmentation model to obtain the target user group.

[0070] The target user segmentation model refers to a model used to segment users into groups. In this case, users are divided into normal user categories and abnormal user categories. Normal user categories refer to users whose historical data contains no abnormal data. Abnormal user categories refer to users whose historical data contains abnormal data. User historical data includes, but is not limited to, various types of user data related to electricity consumption and billing. The target user group refers to the set of users with abnormal historical data.

[0071] Specifically, based on a target group segmentation model, users corresponding to historical user data are categorized into groups. When anomalies exist in a user's historical data, the user is classified into an anomaly category; when all historical data is normal, the user is classified into a normal category. To more accurately learn patterns from the anomaly data, the user clusters corresponding to the anomaly categories are considered the target user group.

[0072] Step S202: Unsupervised training is performed on the user historical data corresponding to the target user group to generate a first target model. The user historical data corresponding to the target user group is then input into the first target model to obtain the first intermediate abnormal data.

[0073] Unsupervised training refers to unsupervised learning, a machine learning method that automatically classifies or groups input data without providing pre-labeled training data. The first target model refers to the model built after unsupervised learning on the historical user data corresponding to the target user group. The first intermediate outlier data refers to the outlier data within the historical user data corresponding to the target user group, obtained by classifying normal and outlier data based on the first target model.

[0074] Specifically, the historical user data corresponding to the target user group is not yet labeled. Unsupervised training is performed on the historical user data corresponding to the target user group to learn the patterns in the historical user data and generate a corresponding first target model. In order to identify the abnormal data in the historical user data corresponding to the target user group, the historical user data corresponding to the target user group is input into the first target model to obtain the first intermediate abnormal data.

[0075] Step S204: Obtain second intermediate data, and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model.

[0076] Here, the second intermediate data refers to additional user data labeled as normal or abnormal. The initial classification model refers to the model used to classify the labeled data. This model has not yet learned the patterns in the first and second intermediate abnormal data. In this application, the initial classification model is obtained by cascading a random forest model and a multilayer perceptron neural network classifier. The second target model refers to the classification model that has been supervisedly trained on the labeled first and second intermediate abnormal models and has learned the patterns in the first and second intermediate models.

[0077] Specifically, in order to further improve the accuracy of the model in identifying abnormal data, additional user data labeled with normal or abnormal data, namely the second intermediate data, is added as part of the training data for training the initial classification model. The first intermediate abnormal data is also labeled with corresponding tags and added to the initial classification model training data. The initial classification model is used to train the first intermediate abnormal data and the second intermediate data. During the training process, the patterns in the first intermediate abnormal data and the second intermediate data are learned, and the parameters in the initial classification model are adjusted to obtain the corresponding second target model.

[0078] Step S206: The target group segmentation model, the first target model, and the second target model are sequentially connected and combined to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0079] In this context, "serialization" refers to a series of consecutive connections. In this application, a portion of the output of the target group segmentation model serves as the input to the first target model, and the output of the first target model serves as a portion of the input to the second model. The target verification model refers to a model trained on historical user data that learns patterns within that data and is used to classify the user data to be verified. The user data to be verified refers to various types of user-related data collected in the current month. The user anomaly category refers to the user category corresponding to a user whose corresponding user data contains anomalies. The user anomaly list refers to the list of anomalies present in the user data to be verified for each user.

[0080] Specifically, a target group segmentation model, a first target model, and a second target model are connected in series to generate a target verification model. By inputting the user data to be verified into the target group segmentation model, the users corresponding to the user data to be verified are segmented into groups to obtain the group categories corresponding to the users. When the group category corresponding to the user is a user anomaly category, the user data to be verified corresponding to the user is then input into the first target model to obtain the anomaly data to be verified in the user data to be verified. Then, the anomaly data in the user data to be verified is input into the second target model. The second target model further checks whether the anomaly data to be verified found by the first target model is accurate, and outputs the data that the second target model considers to be anomaly data. The data output by the second target model is used as the user anomaly list.

[0081] The above-described data verification model generation method involves: acquiring a target group segmentation model; obtaining user historical data; inputting the user historical data into the target group segmentation model; segmenting users into target user groups based on the target group segmentation model; performing unsupervised training on the user historical data corresponding to the target user groups to generate a first target model; inputting the user historical data corresponding to the target user groups into the first target model to obtain first intermediate anomaly data; acquiring second intermediate data; training an initial classification model based on the first intermediate anomaly data and the second intermediate data to obtain a second target model; and sequentially combining the target group segmentation model, the first target model, and the second target model to generate a target verification model. This method is used to obtain user anomaly categories and user anomaly lists based on user data to be verified. It divides users corresponding to historical user data into target user groups using a target group segmentation model. Unsupervised training is then performed on the historical user data corresponding to the target user groups to generate a first target model. Based on the first target model, anomaly data in the historical user data corresponding to the target user groups is identified as first intermediate anomaly data. Second intermediate data is obtained. Based on the first intermediate anomaly data and the second intermediate data, an initial classification model is trained to obtain a second target model. The target group segmentation model, the first target model, and the second target model are sequentially concatenated and combined to generate a target verification model. This process realizes the verification of data anomalies, thereby improving the efficiency of data quality verification.

[0082] In one embodiment, such as Figure 3 As shown, step S200 includes:

[0083] Step S300: Based on the target group segmentation model, calculate the mean of each type of data in the user's historical data to obtain the target mean corresponding to each type of data.

[0084] The target mean refers to the average value obtained by adding up all the data in each category of the user's historical data. Here, the target mean represents the mean value corresponding to each category of data.

[0085] Specifically, by adding up and averaging the data corresponding to each category in the user's historical data, the average value corresponding to each category of data is obtained, which is the target average value corresponding to each category of data. The target average value is used as the basis for judging whether the data is abnormal data.

[0086] Step S302: Calculate the difference between each type of data in the user's historical data and the corresponding target mean to obtain the first target difference.

[0087] The first target difference refers to the absolute value of the difference between each data point in the user's historical data and the corresponding target mean, which is a value greater than or equal to 0.

[0088] Specifically, the absolute value of the difference between each data point in the user's historical data and the corresponding target mean is calculated, and the absolute value is used as the first target difference to prepare data for subsequent judgment on whether the data is abnormal.

[0089] Step S304: Obtain the first target range value, and obtain the target user group based on the comparison results of each first target difference with the first target range value.

[0090] The first target range value refers to the boundary value used to distinguish whether various types of data in user historical data belong to normal data or abnormal data.

[0091] Specifically, when the first target difference is greater than the first target range value, users whose absolute value of the difference between their historical data and the corresponding target mean is equal to the first target difference are classified as abnormal users. When the first target difference is less than or equal to the first target range value, the corresponding users are classified as normal users. The user clusters corresponding to the abnormal user categories are taken as the target user group.

[0092] In this embodiment, users corresponding to historical user data are divided into groups based on the target group segmentation model to obtain the target user group. This achieves the identification and segmentation of user groups, improves the speed of obtaining corresponding data in subsequent processes, and helps to improve the efficiency of related process execution.

[0093] In one embodiment, such as Figure 4 As shown, step S202 includes:

[0094] Step S400: Fit the corresponding target distribution function based on the user historical data corresponding to the target user group, and use the target distribution function as the first model.

[0095] The target distribution function refers to the function used to reflect the distribution of historical user data corresponding to the target user group. The first model refers to the model used to check whether the data input to the model is on the curve represented by the model. If it is not on the curve, the input data is classified as outlier data.

[0096] Specifically, by fitting the target distribution function corresponding to the historical user data of the target user group, and using the target distribution function as the first model, abnormal data in the input data of the first model are identified based on the first model.

[0097] Step S402: Fit the corresponding target density estimation function based on the user historical data corresponding to the target user group, and use the target density estimation function as the second model.

[0098] Here, the density estimation function refers to the probability density function, which is an impulse function that converges to the true value of the parameter when the data samples are infinite. The second model refers to another classification model used to determine whether the input data is outlier.

[0099] Specifically, by fitting a target density estimation function corresponding to the historical user data of the target user group, and using the target density estimation function as a second model, abnormal data in the input data of the second model are identified based on the second model.

[0100] Step S404: Obtain the third model, which is used to obtain the corresponding abnormal data based on the comparison result between the maximum value of the input data and the input data.

[0101] Here, the third model refers to the model used to determine whether the input data is outlier. The input data refers to the data in the third model that will be used to determine if it is outlier. A maximum value refers to a value in a finite dataset that is not less than any other value.

[0102] Specifically, the third model primarily relies on obtaining a target maximum value and using the difference between the target maximum value and the input data to determine whether the input data is normal or abnormal. Since there may be multiple target maximum values, the nearest target maximum value to the input data is selected for comparison.

[0103] Step S406: Generate the first target model by combining the first model, the second model, and the third model in parallel.

[0104] Parallel connection refers to connecting the first model, the second model, and the third model in parallel between the starting point and the ending point. Here, the starting point refers to the data input point, and the ending point refers to the data output point. That is, the data input to the first model, the second model, and the third model are the same data, and the set of output data from the first model, the second model, and the third model is used as the output data of the first target model.

[0105] Specifically, by connecting the first model, the second model, and the third model in parallel, with all three models receiving the same input data, and using the three models to determine whether the input data is normal or abnormal, the accuracy of finding abnormal data can be improved, ensuring that more abnormal data can be found.

[0106] In this embodiment, a first model and a second model are fitted based on the user historical data corresponding to the target user group, and a third model is obtained. The first target model is generated by combining the first model, the second model and the third model in parallel, which improves the accuracy of finding abnormal data and the efficiency of verifying abnormal data.

[0107] In one embodiment, such as Figure 5 As shown, step S202 further includes:

[0108] Step S500: Based on the first curve corresponding to the first model, filter out data outside the first curve from the user historical data corresponding to the target user group as the first abnormal data.

[0109] Here, the first abnormal data refers to the abnormal data filtered out from the historical user data corresponding to the target user group based on the first model. The first curve refers to the curve that can represent the first model.

[0110] Specifically, when the data input into the first model is on the curve represented by the first model, it means that the input data is within the range of normal data. When the data input into the first model is outside the curve represented by the first model, it means that the input data is not within the range of normal data. In this case, the input data is regarded as abnormal data, and the set of all abnormal data is regarded as the first abnormal data.

[0111] Step S502: Based on the second curve corresponding to the second model, filter out data outside the second curve from the user historical data corresponding to the target user group as the second abnormal data.

[0112] Here, the second abnormal data refers to the abnormal data filtered out from the historical user data corresponding to the target user group based on the second model. The second curve refers to the curve that can represent the second model.

[0113] Specifically, when the data input into the second model is on the curve represented by the second model, it means that the input data is within the range of normal data. When the data input into the second model is outside the curve represented by the second model, it means that the input data is not within the range of normal data. In this case, the input data is regarded as abnormal data, and the collection of all abnormal data is regarded as the second abnormal data.

[0114] Step S504: Based on the third model in the first target model, calculate the maximum value of each type of data in the user historical data corresponding to the target user group, and obtain the target maximum value corresponding to each type of data.

[0115] The target maximum value refers to a certain number in the historical data of users corresponding to the target user group, where all data in the vicinity of this data are smaller than this data.

[0116] Specifically, each type of data contains a large amount of data. By finding the maximum value corresponding to each type of data, the target maximum value corresponding to each type of data is obtained.

[0117] Step S506: Calculate the differences between various types of historical data of the target user group and the corresponding target maximum values ​​to obtain the second target difference.

[0118] The second target difference refers to the absolute value of the difference between each data point in the historical data of the target user group and the target maximum value, which is a value greater than or equal to 0.

[0119] Specifically, there may be multiple target maxima for each type of data. When there are multiple target maxima for a type of data, the target maxima corresponding to the interval closest to the input data of the third model is taken as the target maxima to be compared with the input data. The absolute value of the difference between the input data and the target maxima is calculated and the absolute value is taken as the second target difference.

[0120] Step S508: Obtain the second target range value, and obtain the third abnormal data based on the comparison results of each second target difference with the second target range value.

[0121] The second target range value refers to the boundary value used to distinguish between normal and abnormal data in user historical data. The third abnormal data refers to the user historical data when the difference between the second target and the second target range value is less than the second target range value.

[0122] Specifically, when the difference between the second target and the second target is less than or equal to the second target range value, it indicates that the corresponding user's historical data is normal data. When the difference between the second target and the second target is greater than the second target range value, it indicates that the corresponding user's historical data exceeds the normal range and is considered abnormal data.

[0123] Step S510: The first abnormal data, the second abnormal data, and the third abnormal data are fused together to obtain the first intermediate abnormal data.

[0124] Among them, fusion refers to combining the first abnormal data, the second abnormal data, and the third abnormal data set into a single dataset.

[0125] Specifically, the first intermediate abnormal data is a collection of the first abnormal data, the second abnormal data, and the third abnormal data, which contains all the data in the first abnormal data, the second abnormal data, and the third abnormal data, and contains some duplicate data.

[0126] In this embodiment, abnormal data in the historical data of users corresponding to the target user group is filtered out based on the first model, the second model and the third model corresponding to the first target model. This improves the accuracy of finding abnormal data and provides a large data foundation for subsequent supervised training, thereby improving the efficiency of data quality verification.

[0127] In one embodiment, such as Figure 6 As shown, step S204 includes:

[0128] Step S600: Filter the first intermediate abnormal data to obtain the first target abnormal data, and fuse the first target abnormal data with the second intermediate data to obtain the target data.

[0129] Here, the filtering operation refers to the deletion of duplicate data from the first intermediate outlier data. The first target outlier data refers to the data remaining after removing duplicate data from the first intermediate outlier data. The target data refers to the training data required for supervised training of the initial classification model, which includes labeled normal data and outlier data.

[0130] Specifically, there may be a lot of duplicate data in the first intermediate abnormal data. In order to ensure the accuracy of the initial classification model training process, it is necessary to remove the duplicate data in the first intermediate abnormal data to obtain the first target abnormal data. The set of the first target abnormal data and the second intermediate data is used as the training data for the subsequent training of the initial classification model, that is, the target data.

[0131] Step S602: Train the initial classification model based on the target data to obtain multiple intermediate classification models.

[0132] Among them, the intermediate classification model refers to a classification model whose parameters differ from those of the initial classification model obtained by supervised training based on the target data. This intermediate classification model has learned the patterns existing in the target data.

[0133] Specifically, by training the initial classification model multiple times to learn the patterns in the target data and continuously adjusting the parameters of the initial classification model, multiple intermediate classification models are obtained. Each intermediate classification model has a different ability to classify the data in the corresponding intermediate classification model. It is also necessary to select the model with the best classification performance from the multiple intermediate classification models according to the corresponding evaluation index.

[0134] Step S604: Obtain a test set, test the multiple intermediate classification models based on the test set, and obtain the evaluation index corresponding to the multiple intermediate classification models.

[0135] The test set is used to test the classification performance of the intermediate classification model. The evaluation metric is used to measure the classification performance of the intermediate classification model.

[0136] Specifically, the test set is used to test the generalization performance of the multiple intermediate classification models. By inputting the test set into each intermediate classification model and calculating the evaluation index corresponding to each intermediate classification model, a basis for judgment is provided for selecting the model with the best classification ability among the multiple intermediate classification models in the subsequent process.

[0137] Step S606: Based on the comparison results of the evaluation indicators corresponding to the multiple intermediate classification models, determine the target classification model and use the target classification model as the second target model.

[0138] The target classification model refers to the model with the best classification performance among the multiple intermediate classification models.

[0139] Specifically, the model with the highest evaluation index among the multiple intermediate classification models is selected as the target classification model, and this target classification model is used as the second target model.

[0140] In this embodiment, by filtering out duplicate data in the first intermediate abnormal data, the first target abnormal data is obtained. The first target abnormal data and the second intermediate data are used as target data. The initial classification model is trained multiple times based on the target data to obtain multiple intermediate classification models. A test set is obtained, and the multiple intermediate classification models are tested based on the test set. The evaluation index corresponding to the intermediate classification model is calculated. Based on the comparison results of the evaluation index, the target classification model is determined and used as the second target model. This further ensures the accuracy of checking the user data to be verified, thereby improving the efficiency of data quality verification.

[0141] In one embodiment, such as Figure 7 As shown, after step S206, the following steps are also included:

[0142] Step S700: Obtain the target user data to be verified.

[0143] Among them, the target user data to be verified refers to the data obtained after checking whether there are any input errors in the entered data to be verified.

[0144] Specifically, there may be errors in the data of users to be verified. Therefore, before inputting the data to be verified into the corresponding target verification model, it is necessary to conduct a preliminary verification of the data of users to be verified to ensure that the data of users to be verified entered is consistent with the data input into the target verification model. The data obtained from the preliminary verification of the data of users to be verified is the target data of users to be verified.

[0145] Step S702: Input the target user data to be verified into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified.

[0146] Among them, the target user anomaly category refers to the user category corresponding to the target user data to be verified. The target user anomaly list refers to each anomaly data in the target user data to be verified.

[0147] Specifically, after the target user data to be verified is input into the target verification model, the target group segmentation model corresponding to the target verification model will segment the users corresponding to the target user data to be verified into groups. When the user is in the user abnormal category, the target user data to be verified corresponding to the user is input into the first model corresponding to the target verification model. After the first model verifies the target user data to be verified, it obtains the first output data. The second model further verifies the first output data to obtain abnormal data with higher accuracy. The abnormal data is the abnormal data found from the target user data to be verified, which is the target user abnormal list.

[0148] In this embodiment, by acquiring target user data to be verified and inputting the target user data to be verified into the target verification model, the target user anomaly category and target user anomaly list corresponding to the target user data to be verified are obtained, thereby improving the efficiency of verifying the quality of target user data to be verified.

[0149] In one embodiment, such as Figure 8 As shown, before step S700, the following steps are also included:

[0150] Step S800: Construct a billing parameter dictionary based on the user's historical data.

[0151] The billing parameter dictionary includes, but is not limited to, a data dictionary used to compare whether there are any errors in the entered data, and parameters used to study the results of electricity billing.

[0152] Specifically, the billing parameter dictionary inputs a portion of the user's historical data. This is mainly used for pre-verification to ensure that the user information data is entered correctly. For example, the billing parameter dictionary contains certain rules: if the user's metering point is marked as time-of-use, then the user's meter must be a time-of-use meter, and the meter's reading type must have peak and valley reading types.

[0153] Step S802: Obtain the verification rules, and based on the verification rules, check and update each parameter in the billing parameter dictionary to obtain the target billing parameter dictionary.

[0154] Among them, the verification rules refer to the rules formulated by business personnel based on their experience in collecting user data, used to verify the correctness of each parameter in the billing parameter dictionary. The target billing parameter dictionary refers to the data dictionary obtained after correcting incorrect parameters in the billing parameter dictionary based on the verification rules.

[0155] Specifically, the verification rules are rules formulated by personnel based on business experience. When the verification rules find that the corresponding parameters in the billing parameter dictionary do not conform to the corresponding verification rules, the relevant business personnel are prompted to correct the data to obtain the corresponding target billing parameter dictionary.

[0156] Step S804: Obtain the user data to be verified, and obtain the target user data to be verified based on the comparison result between the user data to be verified and the target billing parameter dictionary.

[0157] Among them, the user data to be verified refers to the user information data collected in the current month, and the data for which no operation has been performed yet.

[0158] Specifically, the accuracy of the user data to be verified is checked by using the target billing parameters. When there is an entry error in the user data to be verified, the relevant business personnel are prompted to resolve the corresponding entry error.

[0159] In this embodiment, a billing parameter dictionary is constructed based on user historical data, and a target billing parameter dictionary is obtained by correcting the billing parameter dictionary according to the verification rules. The corresponding target user data to be verified is obtained based on the comparison results between the target parameter dictionary and the user data to be verified, which ensures the accuracy of the entered data and improves the efficiency of data verification to a certain extent.

[0160] In one embodiment, the archives collected by the intelligent verification system in one embodiment are used as user historical data, and the archive data collected by the intelligent verification system in the current month are used as user data to be verified. Based on the user historical data, a corresponding target verification model and target billing parameter dictionary are constructed. The target billing parameter dictionary is used to perform preliminary verification on the user data to be verified to obtain target user data to be verified. The target user data to be verified is input into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified. The target verification model involves acquiring a target group segmentation model, obtaining user historical data, inputting the user historical data into the target group segmentation model, and segmenting each user into a target user group based on the target group segmentation model. The model then performs unsupervised training on the user historical data corresponding to the target user group to generate a first target model. The user historical data corresponding to the target user group is then input into the first target model to obtain first intermediate anomaly data. Second intermediate data is acquired, and an initial classification model is trained based on the first intermediate anomaly data and the second intermediate data to obtain a second target model. Finally, the target group segmentation model, the first target model, and the second target model are sequentially concatenated and combined to generate the target model. Figure 9 This is an example of an intelligent verification system based on the quality of existing power grid data. The system includes a billing parameter dictionary tagging rule module, which uses artificial intelligence algorithms to set parameter dictionaries for billing data and tag verification rules, corresponding to the billing parameter dictionary and verification rule operations in this case. The archive quality comprehensive evaluation module provides anomaly alerts for billing parameter configuration during work order processing and performs a comprehensive evaluation of work order archiving, corresponding to the billing parameter dictionary correction operations in this case. The archive quality anomaly quantity model processes abnormal information in the work order processing. The user electricity consumption and electricity bill anomaly identification model identifies user electricity consumption and billing data. These models correspond to the construction and use of the target verification model in this case. Each execution process performs its specific function, improving both the accuracy of abnormal data verification and the execution efficiency of each process, thereby improving the efficiency of data quality verification.

[0161] Based on the same inventive concept, this application also provides a data verification model generation apparatus for implementing the data verification model generation method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more data verification model generation apparatus embodiments provided below can be found in the limitations of the data verification model generation method described above, and will not be repeated here.

[0162] In one embodiment, such as Figure 10 As shown, a data verification model generation device is provided, including: a group partitioning module 1000, a first training module 1002, a second training module 1004, and a model generation module 1006, wherein:

[0163] The group segmentation module 1000 is used to obtain a target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into a group based on the target group segmentation model to obtain the target user group.

[0164] The first training module 1002 is used to perform unsupervised training on the user historical data corresponding to the target user group to generate a first target model, and input the user historical data corresponding to the target user group into the first target model to obtain the first intermediate abnormal data.

[0165] The second training module 1004 is used to acquire second intermediate data and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model.

[0166] The model generation module 1006 is used to sequentially connect and combine the target group segmentation model, the first target model, and the second target model to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified.

[0167] In one embodiment, the group segmentation module 1000 is further configured to calculate the mean of each type of data in the user's historical data based on the target group segmentation model, and obtain the target mean corresponding to each type of data; calculate the difference between each type of data in the user's historical data and the corresponding target mean, and obtain a first target difference; obtain a first target range value, and obtain the target user group based on the comparison results of each first target difference and the first target range value.

[0168] In one embodiment, the first training module 1002 is further configured to fit a corresponding target distribution function based on the user historical data corresponding to the target user group, and use the target distribution function as a first model; fit a corresponding target density estimation function based on the user historical data corresponding to the target user group, and use the target density estimation function as a second model; obtain a third model, wherein the third model is used to obtain corresponding abnormal data based on the comparison result between the maximum value of the input data and the input data; and generate the first target model by combining the first model, the second model and the third model in parallel.

[0169] In one embodiment, the first training module 1002 is further configured to: filter data outside the first curve from the user historical data corresponding to the target user group as first abnormal data based on the first curve corresponding to the first model; filter data outside the second curve from the user historical data corresponding to the target user group as second abnormal data based on the second curve corresponding to the second model; calculate the maximum value of each type of data in the user historical data corresponding to the target user group based on the third model in the first target model, and obtain the target maximum value corresponding to each type of data; calculate the difference between each type of data in the user historical data corresponding to the target user group and the corresponding target maximum value, and obtain the second target difference value; obtain the second target range value; obtain the third abnormal data based on the comparison results of each second target difference value and the second target range value; and fuse the first abnormal data, the second abnormal data and the third abnormal data to obtain the first intermediate abnormal data.

[0170] In one embodiment, the second training module 1004 is further configured to filter the first intermediate abnormal data to obtain first target abnormal data, fuse the first target abnormal data with the second intermediate data to obtain target data; train the initial classification model based on the target data to obtain multiple intermediate classification models; obtain a test set, test the multiple intermediate classification models based on the test set to obtain evaluation metrics corresponding to the multiple intermediate classification models; and determine the target classification model based on the comparison results of the evaluation metrics corresponding to the multiple intermediate classification models, and use the target classification model as the second target model.

[0171] In one embodiment, the data verification model generation device further includes a model usage module 1008, which is used to acquire target user data to be verified; input the target user data to be verified into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified.

[0172] In one embodiment, the data verification model generation device further includes a data preprocessing module 1010, which is used to construct a billing parameter dictionary based on the user's historical data; obtain verification rules, and based on the verification rules, check and update each parameter in the billing parameter dictionary to obtain a target billing parameter dictionary; obtain user data to be verified, and based on the comparison result between the user data to be verified and the target billing parameter dictionary, obtain the target user data to be verified.

[0173] Each module in the aforementioned data verification model generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0174] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 11 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores historical user data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When executed by the processor, the computer program implements a data verification model generation method.

[0175] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 12As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a data verification model generation method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0176] Those skilled in the art will understand that Figure 11 and Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0177] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method examples.

[0178] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0179] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0180] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0181] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0182] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0183] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for generating a data verification model, characterized in that, The method includes: Obtain a target group segmentation model, obtain user historical data, input the user historical data into the target group segmentation model, and segment each user into groups based on the target group segmentation model to obtain the target user group; wherein, the target user group is a user cluster corresponding to the user anomaly category, and the user historical data includes various data related to user electricity consumption and electricity bills collected; Unsupervised training is performed on the user history data corresponding to the target user group to generate a first target model. The user history data corresponding to the target user group is then input into the first target model to obtain the first intermediate abnormal data. Obtain second intermediate data, and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model; wherein, the second intermediate data refers to additional user data that has been labeled with normal or abnormal data labels, and the second target model refers to a classification model that has been supervisedly trained on the first intermediate abnormal data and the second intermediate data with labels and has learned the patterns in the first intermediate abnormal data and the second intermediate data. The target group segmentation model, the first target model, and the second target model are sequentially connected and combined to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified. Obtain the target user data to be verified; The target user data to be verified is input into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified; The target verification model includes: a target group segmentation model, a first target model, and a second target model. The step of inputting the target user data to be verified into the target verification model to obtain the target user anomaly category and the target user anomaly list corresponding to the target user data includes: Input the target user data to be verified into the target group segmentation model, and segment the users corresponding to the target user data to be verified into groups; When the user is an abnormal user category, the target user data to be verified corresponding to the user is input into the first target model corresponding to the target verification model. After the target user data to be verified is verified by the first target model, the first output data is obtained. The first output data is re-verified based on the second objective model to obtain user anomaly categories and user anomaly lists.

2. The method according to claim 1, characterized in that, The step involves inputting the user's historical data into the target group segmentation model, and then segmenting each user into groups based on the target group segmentation model to obtain the target user groups, including: Based on the target group segmentation model, the mean of each type of data in the user's historical data is calculated to obtain the target mean of each type of data. Calculate the difference between each type of data in the user's historical data and the corresponding target mean to obtain the first target difference; Obtain the first target range value, and based on the comparison results between each of the first target differences and the first target range value, obtain the target user group.

3. The method according to claim 1, characterized in that, The step of performing unsupervised training on the user historical data corresponding to the target user group to generate the first target model includes: Based on the historical user data corresponding to the target user group, a corresponding target distribution function is fitted, and the target distribution function is used as the first model. Based on the historical user data corresponding to the target user group, a corresponding target density estimation function is fitted, and the target density estimation function is used as the second model. A third model is obtained, which is used to obtain the corresponding abnormal data based on the comparison result between the maximum value of the input data and the input data. The first target model is generated by combining the first model, the second model, and the third model in parallel.

4. The method according to claim 3, characterized in that, The step involves inputting historical user data corresponding to the target user group into the first target model to obtain first intermediate abnormal data. The first target model is obtained by combining the first model, the second model, and the third model in parallel. Based on the first curve corresponding to the first model, data outside the first curve is selected from the user historical data corresponding to the target user group as the first abnormal data; Based on the second curve corresponding to the second model, data outside the second curve are selected from the user historical data corresponding to the target user group as the second abnormal data; Based on the third model in the first target model, calculate the maximum value of each type of data in the user history data corresponding to the target user group, and obtain the target maximum value corresponding to each type of data. The differences between various types of historical data of users corresponding to the target user group and the corresponding target maximum values ​​are calculated respectively to obtain the second target difference; Obtain the second target range value, and based on the comparison results between each second target difference and the second target range value, obtain the third abnormal data; The first abnormal data, the second abnormal data, and the third abnormal data are merged to obtain the first intermediate abnormal data.

5. The method according to claim 1, characterized in that, The step of obtaining the second intermediate data and training the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model includes: The first intermediate abnormal data is filtered to obtain the first target abnormal data. The first target abnormal data is then fused with the second intermediate data to obtain the target data. The initial classification model is trained based on the target data to obtain multiple intermediate classification models; Obtain a test set, test the multiple intermediate classification models based on the test set, and obtain the evaluation index corresponding to the multiple intermediate classification models; Based on the comparison results of the evaluation indicators corresponding to the multiple intermediate classification models, the target classification model is determined and used as the second target model.

6. The method according to claim 1, characterized in that, Before obtaining the target user data to be verified, the process also includes: A billing parameter dictionary is constructed based on the user's historical data; Obtain the verification rules, and based on the verification rules, check and update each parameter in the billing parameter dictionary to obtain the target billing parameter dictionary; Obtain the user data to be verified, and based on the comparison results between the user data to be verified and the target billing parameter dictionary, obtain the target user data to be verified.

7. A data verification model generation device, characterized in that, The device includes: The group segmentation module is used to obtain a target group segmentation model, acquire user historical data, input the user historical data into the target group segmentation model, and segment each user into groups based on the target group segmentation model to obtain the target user group; wherein, the target user group is a user cluster corresponding to the user anomaly category, and the user historical data includes various data related to user electricity consumption and electricity bills collected; The first training module is used to perform unsupervised training on the user historical data corresponding to the target user group to generate a first target model, and input the user historical data corresponding to the target user group into the first target model to obtain the first intermediate abnormal data. The second training module is used to acquire the second intermediate data and train the initial classification model based on the first intermediate abnormal data and the second intermediate data to obtain the second target model. The second intermediate data refers to the additional user data that has been labeled with normal or abnormal data labels. The second target model refers to the classification model that has been supervisedly trained on the first intermediate abnormal data and the second intermediate data with labels and has learned the patterns in the first intermediate abnormal data and the second intermediate data. The model generation module is used to sequentially combine the target group segmentation model, the first target model, and the second target model to generate a target verification model. The target verification model is used to obtain user anomaly categories and user anomaly lists based on the user data to be verified. Obtain the target user data to be verified; The target user data to be verified is input into the target verification model to obtain the target user anomaly category and target user anomaly list corresponding to the target user data to be verified; The target verification model includes: a target group segmentation model, a first target model, and a second target model. The step of inputting the target user data to be verified into the target verification model to obtain the target user anomaly category and the target user anomaly list corresponding to the target user data includes: Input the target user data to be verified into the target group segmentation model, and segment the users corresponding to the target user data to be verified into groups; When the user is an abnormal user category, the target user data to be verified corresponding to the user is input into the first target model corresponding to the target verification model. After the target user data to be verified is verified by the first target model, the first output data is obtained. The first output data is re-verified based on the second objective model to obtain user anomaly categories and user anomaly lists.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images