An artificial intelligence-based account exception data periodic detection method and system

Abstract

The application provides an artificial intelligence-based account exception data periodic detection method, and belongs to the technical field of data processing, and specifically comprises the following steps: when it is determined that the business system is abnormal based on the data volume, the number and types of handled businesses of running data within a set time, data features are extracted based on the running data within the set time, and the system fault time is located by combining the fusion weight of different moments; based on the number and types of handled businesses of the business system within the set time, and in combination with the exception rate of account data, the type and number of supported handled business types of the business system at the historical fault time, an artificial intelligence-based evaluation model is used to determine the extended detection time, the detection time is determined in combination with the system fault time, and the account data generated by the business system is detected based on the detection time, so that the targeted periodic detection of account exception data is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of anomaly detection, and particularly relates to an account anomaly data periodic detection method and system based on artificial intelligence. BACKGROUND

[0002] In order to realize automatic identification of account anomaly data, the prior art often checks all account data of a financial system, which is not only low in processing efficiency, but also reduces the accuracy of evaluation due to a large amount of data. Since account anomaly data is often caused by system anomaly or user operation anomaly, the application patent with the publication number CN115760447A determines target anomaly account data to be processed and a processing strategy of the target anomaly account data by using each anomaly account data corresponding to a target business transaction type generated in a time range in which the system fails, and processes the target anomaly account data according to the processing strategy of the target anomaly account data. However, the following problems exist in general.

[0003] Automatic confirmation and positioning of system failure time cannot be realized. In actual operation process, an operator often cannot accurately position the failure time. Generally, if the system is in a failure state, the amount of uploaded data cannot match the data type of a business handling personnel and the number of business handling personnel. Therefore, if the system failure time cannot be positioned in combination with the above data, account data cannot be accurately and comprehensively detected and checked.

[0004] The determination of detection time in combination with historical identification of historical account data of different systems is not considered. Since when it is determined that the system fails, a certain amount of account data often exists, and the failure rate of financial data of historical account data of different systems is not the same. Therefore, if the above data cannot be combined, account data cannot be accurately and completely detected and checked.

[0005] In view of the above technical problems, the application provides an account anomaly data periodic detection method and system based on artificial intelligence. SUMMARY

[0006] According to one aspect of the application, an account anomaly data periodic detection method based on artificial intelligence is provided.

[0007] An account anomaly data periodic detection method based on artificial intelligence, characterized in that it specifically comprises:

[0008] S11 obtains a data amount of running data of a business system in a set time based on a preset period, and when the data amount is abnormal, proceeds to step S12.

[0009] S12 determines whether the business system is abnormal based on the data amount of the operation data within the set time, the number and type of the handled business, and when the business system is abnormal, proceeds to step S13;

[0010] S13 extracts data features based on the operation data within the set time, determines fusion weights using the number and type of the handled business at different times within the set time, and locates the system failure time based on the data features and the fusion weights;

[0011] S14 determines the expansion detection time of the business system using an artificial intelligence-based evaluation model based on the number and type of the handled business of the business system within the set time, in combination with the abnormality rate of the account data of the business system at the historical failure time and the types and number of the business types supported by the business system, determines the detection time in combination with the system failure time, and detects the account data generated by the business system based on the detection time.

[0012] By judging the abnormality based on the data amount of the operation data of the business system within the set time based on the preset period, the real-time monitoring of the abnormal condition of the business system is realized in a relatively simple manner, and the timely and accurate identification of the abnormal condition of the business system is ensured.

[0013] By combining the data amount of the operation data within the set time, the number and type of the handled business to judge the abnormality of the business system, the problem that the case of less data amount due to less number and type of the handled business of the business system itself is mistakenly considered as abnormal is avoided, and the efficiency of abnormality judgment is further improved.

[0014] By combining the determination of the fusion weights using the number and type of the handled business at different times within the set time, the determination of the weights from the importance and number of the handled business is realized, which lays a foundation for further accurate positioning of the system failure time. And by locating the system failure time based on the data features and the fusion weights, the positioning of the system failure time from multiple angles is realized, the accuracy of the positioning is ensured, and a foundation for further detection of account data is laid.

[0015] By means of the number and type of the handled businesses of the business system within a set time, in combination with the abnormal rate of the account data of the business system at the time of historical failure and the type and number of the handled businesses supported by the business system, an evaluation model based on artificial intelligence is used to determine the extended detection time of the business system, so as to realize the determination of the extended detection time from the perspective of historical data and the importance and number of business types, further guarantee the comprehensiveness of account data detection, and avoid unnecessary economic losses.

[0016] In another aspect, the present application provides an account abnormal data periodic detection system based on artificial intelligence, which adopts the above-mentioned account abnormal data periodic detection method based on artificial intelligence, and specifically comprises:

[0017] a data amount acquisition module, a system failure time positioning module, a detection time determination module and an account data detection module;

[0018] The data amount acquisition module is responsible for acquiring the data amount of the running data of the business system within a set time based on a preset period.

[0019] The system failure time positioning module is responsible for extracting data features based on the running data within the set time, determining fusion weights by means of the number and type of the handled businesses at different time within the set time, and positioning the system failure time based on the data features and the fusion weights.

[0020] The detection time determination module is responsible for determining the extended detection time of the business system based on the number and type of the handled businesses of the business system within a set time, in combination with the abnormal rate of the account data of the business system at the time of historical failure and the type and number of the handled businesses supported by the business system, and determining the detection time in combination with the system failure time.

[0021] The account data detection module is responsible for detecting the account data generated by the business system based on the detection time.

[0022] In another aspect, the present application provides an account abnormal data periodic detection system based on artificial intelligence, which adopts the above-mentioned account abnormal data periodic detection method based on artificial intelligence, and specifically comprises:

[0023] a running data acquisition module, a running state determination module, a detection time determination module and an account data detection module;

[0024] The running data acquisition module is responsible for acquiring the data amount of the running data of the business system within a set time based on a preset period, and the number and type of the handled businesses.

[0025] The running state determination module is responsible for judging the abnormal state of the business system and the data volume;

[0026] The detection time determination module is responsible for extracting data features based on the running data within the set time, determining fusion weights using the number and types of handled businesses at different moments within the set time, and positioning the system failure moment based on the data features and fusion weights; based on the number and types of handled businesses of the business system within the set time, in combination with the abnormal rate of the account data of the business system at the historical failure moment and the types and number of the business types supported by the business system, an evaluation model based on artificial intelligence is used to determine the extended detection time of the business system, and the detection time is determined in combination with the system failure moment;

[0027] The account data detection module is responsible for detecting the account data generated by the business system based on the detection time.

[0028] On the other hand, the present application provides a computer storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the above-mentioned periodic detection method for abnormal account data based on artificial intelligence.

[0029] Other features and advantages will be set forth in the following description of the application, and in part will become apparent to those skilled in the art upon examination of the following or can be learned by practice of the application. The objects and other advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.

[0030] In order to make the above-mentioned objects, features and advantages of the present application more apparent and easy to understand, the following preferred embodiments are specifically described with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0031] The above and other features and advantages of the present application will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings:

[0032] Figure 1 is a flowchart of a periodic detection method for abnormal account data based on artificial intelligence according to embodiment 1;

[0033] Figure 2 is a flowchart of the specific steps for determining whether the business system is abnormal according to embodiment 1;

[0034] Figure 3 is a flowchart of the specific steps for positioning the system failure moment according to embodiment 1;

[0035] Figure 4is a framework diagram of an artificial intelligence-based account exception data periodicity detection system according to embodiment 2.

[0036] Figure 5 is another framework diagram of an artificial intelligence-based account exception data periodicity detection system according to embodiment 3.

[0037] Figure 6 is a framework diagram of a computer storage medium according to embodiment 4. DETAILED DESCRIPTION

[0038] The embodiments of the present specification provide an artificial intelligence-based account exception data periodicity detection method and system.

[0039] In order to enable those skilled in the art to better understand the technical solutions in the present specification, the technical solutions in the present specification will be described clearly and completely in conjunction with the drawings in the embodiments of the present specification. Obviously, the described embodiments are only part of the embodiments of the present specification, not all embodiments. Based on the embodiments of the present specification, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present specification.

[0040] Embodiment 1

[0041] To solve the above problems, according to one aspect of the present application, as shown in Figure 1 According to one aspect of the present application, an artificial intelligence-based account exception data periodicity detection method is provided, characterized in that it specifically comprises:

[0042] S11, based on a preset period, obtaining the data amount of the running data of the business system within a set time, and when the data amount is abnormal, entering step S12;

[0043] In the present embodiment, the preset period is determined according to the type of the business type supported by the business system and the amount of accepted business, wherein the more the type of the business type supported by the business system and the more the amount of accepted business, the shorter the preset period.

[0044] In the present embodiment, the business system includes but is not limited to a payment system, an online banking system, a credit system, and a business acceptance terminal.

[0045] Specifically, when the data amount of the running data of the business system within a set time is greater than a set amount, it is determined that the data amount is abnormal.

[0046] In the embodiment, the abnormality of the data quantity of the running data of the business system in the set time is determined based on the preset period, so that the real-time monitoring of the abnormal condition of the business system is realized in a relatively simple manner, and the timely and accurate identification of the abnormal condition of the business system is ensured.

[0047] S12 determines whether the business system is abnormal based on the data quantity of the running data in the set time, the number and type of the handled business, and when the business system is abnormal, enters step S13;

[0048] Specifically, as shown in Figure 2 The specific steps of determining whether the business system is abnormal are:

[0049] S21 obtains the data quantity of the uplink data and the data quantity of the downlink data based on the running data in the set time, and determines whether the business system is abnormal based on the data quantity of the uplink data and the data quantity of the downlink data, if yes, determines that the business system is abnormal, and if no, enters step S22;

[0050] It should be noted that when either the data quantity of the uplink data or the data quantity of the downlink data is less than the preset quantity, and the preset quantity is less than the recommended value of the uplink data and the recommended value of the downlink data, the business system basically has no uplink data or downlink data in the set time, and it is determined that the business system is abnormal.

[0051] On the other hand, if the data quantity of the uplink data and the data quantity of the downlink data do not match, for example, the uplink data is 1 GB, and the downlink data is only 2 MB, it indicates that the business system may also be abnormal at this time.

[0052] S22 determines the recommended value of the uplink data and the recommended value of the downlink data of the business system based on the number and type of the handled business in the set time, and determines whether the business system is abnormal based on the data quantity of the uplink data and the recommended value of the uplink data, if yes, determines that the business system is abnormal, and if no, enters step S23;

[0053] S23 determines whether the business system is abnormal based on the data quantity of the downlink data and the recommended value of the downlink data.

[0054] It should be noted that the recommended value of the uplink data and the recommended value of the downlink data are determined according to the average value of the uplink data and the average value of the downlink data of the historical data corresponding to the number and type of the handled business in the set time.

[0055] In the embodiment, by combining the data amount of operation data in the set time, the number and type of handled services to judge the anomaly of the service system, the problem that the case of less data amount due to the number and type of handled services of the service system itself being less is mistakenly considered as abnormal is avoided, and the efficiency of anomaly judgment is further improved.

[0056] S13 extracts data features based on operation data in the set time, determines fusion weights using the number and type of handled services at different times in the set time, and locates the system failure time based on the data features and the fusion weights;

[0057] Specifically, as shown in Figure 3 The specific steps of locating the system failure time are as follows:

[0058] S31 determines the weights of different times in the set time based on the number and type of handled services at different times in the set time using a feature extraction model based on a time neural network;

[0059] In the embodiment, by constructing a data set matrix based on the number and type of handled services at different times in the set time, and based on a feature extraction model of a CNN algorithm, the weights of different times determined based on the number and type of handled services at different times can be obtained. The weights reflect the importance of the account data at different times, thereby laying a foundation for accurately locating the system failure time.

[0060] S32 determines the evaluation weights of different times in the set time based on the number and type of handled services at different times in the set time using a feature extraction model based on a Self-Attention mechanism;

[0061] S33 constructs fusion weights using a softmax function based on the weights of different times and the evaluation weights;

[0062] Specifically, the weights of the weights of different times and the evaluation weights are determined based on a softmax function, and the fusion weights are constructed based on the weights of the weights of different times and the evaluation weights.

[0063] In the embodiment, by constructing the fusion weights, it is further ensured that the number and type of handled services can more accurately reflect the importance of actual account data at different times, thereby more accurately and comprehensively locating the system failure time.

[0064] S34 determines the uplink data and the downlink data in the set time based on the operation data in the set time, and extracts data features based on the uplink data and the downlink data to obtain uplink timing data features, uplink amplitude data features, downlink amplitude data features, and downlink timing data features;

[0065] S35 obtains uplink fusion time domain data features, downlink fusion timing data features, uplink fusion amplitude data features, and downlink fusion amplitude data features based on the uplink timing data features, the downlink timing data features, the uplink amplitude data features, and the downlink amplitude data features in combination with fusion weights, and locates the system fault time based on the uplink fusion time domain data features, the downlink fusion timing data features, the uplink fusion amplitude data features, and the downlink fusion amplitude data features.

[0066] In this embodiment, it should be noted that the system fault time locating specifically includes:

[0067] locating the system fault time of the uplink data based on the uplink fusion time domain data features and the uplink fusion amplitude data features;

[0068] locating the system fault time of the downlink data based on the downlink fusion timing data features and the downlink fusion amplitude data features;

[0069] confirming the system fault time based on the minimum value of the system fault time of the uplink data and the system fault time of the downlink data.

[0070] By combining the fusion weights determined by the number and type of handled businesses at different times in the set time, the determination of the weights from the importance and number of handled businesses is realized, which lays a foundation for further accurate positioning of the system fault time. Moreover, by locating the system fault time based on the data features and the fusion weights, the system fault time is located from multiple angles, ensuring the accuracy of the positioning and laying a foundation for further detection of account data.

[0071] S14 determines the expansion detection time of the business system based on an artificial intelligence-based evaluation model based on the number and type of handled businesses of the business system in the set time, in combination with the abnormal rate of account data of the business system at the historical fault time and the types and number of supported business types of the business system, determines the detection time in combination with the system fault time, and detects the account data generated by the business system based on the detection time.

[0072] It should be noted that the specific steps of determining the expansion detection time are as follows:

[0073] S41 constructs a basic monitoring extension time based on the types and quantities of the business types supported by the business system, and determines whether the basic monitoring extension time needs to be corrected based on the quantity of the handled business of the business system within a set time, if yes, goes to step S44, if no, goes to step S42;

[0074] It should be noted that the basic monitoring extension time is determined according to the importance of the business system, and is also determined according to the types and quantities of the business types supported by the business system. Generally, the value range of the basic extension time is between 15 minutes and 30 minutes, and the basic monitoring extension time can also be determined according to the accumulated fault time of the business system, for example, when the accumulated fault time of the business system is 2 hours or more, the basic monitoring extension time is determined based on 10 to 20 percent of the accumulated fault time.

[0075] S42 determines whether the basic monitoring extension time needs to be corrected based on the types of the handled business of the business system within a set time, if yes, goes to step S44, if no, goes to step S43;

[0076] It should be noted that when the types of the handled business of the business system within the extension time are more, the basic monitoring extension time needs to be corrected, which is also for more accurate and comprehensive analysis of the accounting data.

[0077] S43 determines whether the basic monitoring extension time needs to be corrected based on the abnormal rate of the accounting data of the business system at the historical fault, if yes, goes to step S44, if no, the basic extension time is taken as the extension detection time of the business system;

[0078] S44 determines the correction time amount of the business system based on the abnormal rate of the accounting data of the business system at the historical fault, the types of the handled business of the business system within a set time, and the quantity of the handled business of the business system within a set time, using an evaluation model based on an artificial intelligence algorithm, and determines the extension detection time of the business system based on the correction time amount and the basic monitoring extension time.

[0079] It should be noted that the correction time amount of the business system is determined using an evaluation model based on the FOA-GRU algorithm, which optimizes the hyperparameters, the number of neurons, and the learning rate of the GRU using the improved FOA algorithm, and constructs the evaluation model of FOA-GRU to obtain the correction time amount of the business system. The specific steps of constructing the evaluation model are:

[0080] (1) Data processing.

[0081] Firstly, the fault characteristic parameter data and weather data collected from the photovoltaic array experimental platform are constructed into vectors, and after normalization preprocessing, they are divided into training data and test data. The historical fault characteristic parameters and corresponding irradiance and environmental temperature are taken as the independent variables of the model, and the predicted future fault characteristic parameters at a certain time are taken as the dependent variable of the model.

[0082] (2) Set the initial parameters of FOA algorithm.

[0083] The basic parameters of FOA algorithm include: population size N, maximum iteration number Maxt, convergence adjustment factor μ, convergence factor initial and final value ω1 and ω2, and dimension Dim. Cubic chaotic mapping, logistic mapping and random function are used to improve the initial position distribution of fruit fly population, improve its randomness, and improve the convergence performance.

[0084] (3) Select the concentration determination function.

[0085] The root mean squared error between the predicted value and the actual value of the fault characteristic parameter is used as the concentration determination function, which is represented by RMSE (root mean squared error, RMSE), that is:

[0086]

[0087] In the formula, n is the number of data; y i is the actual value of the fault characteristic parameter of the photovoltaic array; is the predicted value of the fault characteristic parameter of the photovoltaic array, and the smaller the RMSE, the better the prediction effect, that is, the smaller the concentration determination value, indicating that the prediction effect is better.

[0088] (4) FOA algorithm optimizes the parameters of GRU.

[0089] After initializing the parameters of the algorithm, the fitness value of fruit fly is calculated according to the concentration determination function, that is, the root mean squared error. If the current error value is less than the error value of the last cycle, the current error value is taken as the best taste concentration for updating in the next cycle. In the iteration, the optimization and updating are not stopped until the maximum iteration number is reached, and then the best parameter combination is output. The evaluation model of FOA-GRU is established by using the parameter combination.

[0090] It should be noted that, for the algorithm in the iteration process, the random step mechanism is easy to cause the instability of the algorithm performance and the blindness of the optimization process. The solution adopted in the present research is to introduce the most common Cubic chaotic mapping and logistic mapping to improve the influence of random initial population on the performance of FOA, dynamically adjust the search step of fruit fly population, so that the optimized fruit fly position has the characteristics of chaotic randomness, ergodicity and regularity, and further introduce a random quantity, so as to improve the randomness of the fruit fly position and improve the overall convergence efficiency, and the iteration formula is:

[0091]

[0092] Where rand(0,1) is a random number conforming to normal distribution, n is the number of iterations, f n , f n+1 are the fruit fly positions at the n-th iteration and the n+1-th iteration respectively, K1, K2, a, b are control parameters, the present application makes the fruit fly positions uniformly distributed in the solution space, and then brings them into the objective function calculation, and selects the optimal value as the initial population position.

[0093] It should be noted that the detection of the account data generated by the business system includes but is not limited to transaction type, account data change, and transaction volume.

[0094] In the present embodiment, the determination of the extended detection time of the business system is carried out by using an evaluation model based on artificial intelligence based on the number and type of business handled by the business system within a set time, in combination with the abnormal rate of the account data of the business system at the time of historical failure and the types and quantities of the business types supported by the business system, thereby realizing the determination of the extended detection time from the perspective of historical data and the importance and quantity of business types, further ensuring the comprehensiveness of account data detection, and avoiding unnecessary economic losses.

[0095] A best embodiment is given below:

[0096] The data amount of the running data of the business system within a set time is acquired based on a preset period, and when the data amount of the running data within the set time is greater than a set amount, it is determined that the data amount is abnormal, and the next step is entered;

[0097] As shown in Figure 2 , based on the data amount of the running data within the set time, the number and type of handled business, when it is determined that the business system is abnormal, the next step is entered;

[0098] Figure 3 ​extract data features based on the running data within the set time, determine fusion weights using the number and types of handled businesses at different time points within the set time, and locate the system failure time point based on the data features and the fusion weights;

[0099] determine the detection time point of the business system in combination with the system failure time point, and detect the account data generated by the business system based on the detection time point.

[0100] Embodiment 2

[0101] In another aspect, as Figure 4 shown, the present application provides an account abnormal data periodic detection system based on artificial intelligence, which adopts the above-mentioned account abnormal data periodic detection method based on artificial intelligence, and specifically comprises:

[0102] a data amount acquisition module, a system failure time point locating module, a detection time point determining module, and an account data detecting module;

[0103] The data amount acquisition module is responsible for acquiring the data amount of the running data of the business system within a set time based on a preset period.

[0104] The system failure time point locating module is responsible for extracting data features based on the running data within the set time, determining fusion weights using the number and types of handled businesses at different time points within the set time, and locating the system failure time point based on the data features and the fusion weights.

[0105] The detection time point determining module is responsible for determining the detection time point of the business system based on the number and types of handled businesses of the business system within a set time in combination with the abnormal rate of the account data of the business system at a historical failure time and the types and number of the business types supported by the business system, and determining the detection time point of the business system using an evaluation model based on artificial intelligence.

[0106] The account data detecting module is responsible for detecting the account data generated by the business system based on the detection time point.

[0107] Embodiment 3

[0108] In another aspect, as Figure 5 shown, the present application provides an account abnormal data periodic detection system based on artificial intelligence, which adopts the above-mentioned account abnormal data periodic detection method based on artificial intelligence, and specifically comprises:

[0109] The running data acquisition module, the running state determination module, the detection time determination module, and the account data detection module;

[0110] The running data acquisition module is responsible for acquiring the data volume of the running data of the business system within the set time, the number and type of handled businesses based on a preset period;

[0111] The running state determination module is responsible for judging the abnormal state of the business system and the data volume;

[0112] The detection time determination module is responsible for extracting data features based on the running data within the set time, determining fusion weights using the number and type of handled businesses at different times within the set time, and positioning the system failure time based on the data features and the fusion weights; determining the extended detection time of the business system using an artificial intelligence-based evaluation model based on the number and type of handled businesses of the business system within the set time, in combination with the abnormal rate of the account data of the business system at the historical failure time and the types and number of the business types supported by the business system, and determining the detection time in combination with the system failure time;

[0113] The account data detection module is responsible for detecting the account data generated by the business system based on the detection time.

[0114] Embodiment 4

[0115] On the other hand, as Figure 6 indicated, the present application provides a computer storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the above-mentioned artificial intelligence-based account abnormal data periodic detection method.

[0116] In this embodiment, the artificial intelligence-based financial abnormal data periodic detection method specifically includes:

[0117] The data volume of the running data of the business system within the set time is acquired based on a preset period, and when the data volume of the running data within the set time is greater than a set amount, it is determined that the data volume is abnormal, and the next step is entered;

[0118] As Figure 2 indicated, when it is determined that the business system is abnormal based on the data volume of the running data within the set time, the number and type of handled businesses, the next step is entered;

[0119] An Figure 3In a manner, based on the operation data within the set time, the data features are extracted, the number and type of handled businesses at different time within the set time are used to determine the fusion weight, and based on the data features and the fusion weight, the system failure time is located;

[0120] Based on the types and quantities of the supported business types of the business system, the basic monitoring extension time is constructed, and when it is determined that the basic monitoring extension time needs to be corrected according to the number of handled businesses within the set time, the next step is entered;

[0121] It should be noted that the basic monitoring extension time is determined according to the importance of the business system, and it is also determined according to the types and quantities of the supported business types. Generally, the value range of the basic extension time is between 15 minutes and 30 minutes, and the basic monitoring extension time can also be determined according to the accumulated failure time of the business system, for example, when the accumulated failure time of the business system is more than 2 hours, the basic monitoring extension time is determined based on 10 to 20 percent of the accumulated failure time.

[0122] Based on the abnormal rate of the business system at the historical failure time, the types of handled businesses of the business system within the set time, and the number of handled businesses of the business system within the set time, an evaluation model based on artificial intelligence algorithm is used to determine the correction time of the business system, and based on the correction time and the basic monitoring extension time, the extension detection time of the business system is determined.

[0123] Specifically, the correction time of the business system is determined by using an evaluation model based on FOA-GRU algorithm, the improved FOA algorithm is used to optimize the hyperparameters, the number of neurons, and the learning rate of GRU, and the FOA-GRU evaluation model is constructed to obtain the correction time of the business system

[0124] In the present instance, it will be appreciated by those skilled in the art that all or part of the processes of the above-described embodiments can be implemented by way of a computer program, which can be stored in a non-volatile storage medium, and which, when executed, can include the processes of the above-described embodiments. Any reference in the present application to a storage medium should be interpreted to include a non-volatile storage medium and / or a volatile storage medium. Non-volatile storage media include, for example, optical and / or magnetic disks, and / or the like. Volatile storage media include, for example, RAM, and / or the like. As an illustration and not a limitation, RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), Rambus Direct RAM (RDRAM), and / or the like.

[0125] Each of the embodiments in the present specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each of the embodiments focuses on the difference from other embodiments. In particular, the device, apparatus, and non-transitory computer storage medium embodiments are described simply because they are substantially similar to the method embodiments, and the relevant parts can be referred to the part of the method embodiments.

[0126] The above describes specific embodiments of the present specification. Other embodiments are within the scope of the appended claims. In some cases, the acts or steps recited in the claims can be performed in a different order than the order in which they are recited and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

[0127] The above only describes one or more embodiments of the present specification and is not intended to limit the present specification. One or more embodiments of the present specification can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of one or more embodiments of the present specification should be included in the scope of the claims of the present specification.

Claims

1. A method for periodic detection of accounting anomalies based on artificial intelligence, characterized in that, Specifically, it includes: The system acquires the amount of operational data from the business system within a set time period based on a preset cycle, and proceeds to the next step if the data volume is abnormal. Based on the amount of data, the number and type of transactions processed within the set time period, determine whether there is an anomaly in the business system, and if there is an anomaly in the business system, proceed to the next step; Data features are extracted based on the operational data within the set time period. The number and type of business processed at different times within the set time period are used to determine the fusion weight. The system fault time is located based on the data features and the fusion weight. Based on the number and type of business transactions processed by the business system within a set time period, and combined with the anomaly rate of the accounting data during historical failures of the business system, as well as the types and number of business transactions supported by the business system, an artificial intelligence-based evaluation model is used to determine the extended detection time of the business system. The detection time is determined in conjunction with the system failure time, and the accounting data generated by the business system is detected based on the detection time. The location of the system failure moment specifically includes: The system fault time of the uplink data is obtained based on the uplink fused time domain data characteristics and uplink fused amplitude data characteristics; The system fault time of the downlink data is obtained based on the downlink fused time series data characteristics and downlink fused amplitude data characteristics; The system failure time is determined by the minimum of the system failure time of the uplink data and the system failure time of the downlink data.

2. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The preset period is determined based on the types of business that the business system supports and the volume of business handled. The more types of business that the business system supports and the more business handled, the shorter the preset period will be.

3. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The business systems include, but are not limited to, payment systems, online banking systems, credit systems, and business acceptance terminals.

4. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The specific steps to determine whether the business system has any anomalies are as follows: Based on the running data within the set time period, the data volume of uplink data and downlink data is obtained, and based on the data volume of uplink data and downlink data, it is determined whether there is an anomaly in the business system. If so, it is determined that there is an anomaly in the business system; otherwise, proceed to the next step. Based on the number and type of business processed within the set time period, the recommended values ​​for uplink and downlink data of the business system are determined. Based on the amount of uplink data and the recommended values ​​for uplink data, it is determined whether there is an anomaly in the business system. If so, it is determined that there is an anomaly in the business system. If not, proceed to the next step. The system determines whether there are any anomalies based on the amount of downlink data and the recommended values ​​of the downlink data.

5. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The recommended values ​​for uplink and downlink data are determined based on the average of the uplink and downlink data of historical data corresponding to the number and type of transactions processed within the set time period.

6. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The specific steps for locating a system failure are as follows: Based on the number and type of business processed at different times within the set time period, a feature extraction model based on a time neural network is used to determine the weights of different times within the set time period. Based on the number and type of business processed at different times within a set time period, a feature extraction model based on the Self-Attention mechanism is used to determine the evaluation weights at different times within the set time period. Based on the weights at different times and the evaluation weights, the softmax function is used to construct the fusion weights; Based on the running data within the set time period, the uplink data and downlink data within the set time period are determined, and data features are extracted based on the uplink data and downlink data to obtain uplink time series data features, uplink amplitude data features, downlink amplitude data features, and downlink time series data features. Based on the aforementioned uplink time-domain data features, downlink time-series data features, uplink amplitude data features, and downlink amplitude data features, combined with the fusion weights, the following uplink fused time-domain data features, downlink fused time-series data features, uplink fused amplitude data features, and downlink fused amplitude data features are obtained. Based on these uplink fused time-domain data features, downlink fused time-series data features, uplink fused amplitude data features, and downlink fused amplitude data features, the system fault time is located.

7. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The specific steps for determining the extended detection time are as follows: S41 constructs a basic monitoring extension time based on the types and quantities of business types supported by the business system, and determines whether the basic monitoring extension time needs to be corrected by the number of business processed by the business system within the set time. If yes, proceed to step S44; otherwise, proceed to step S42. S42 determines whether the basic monitoring extension time needs to be corrected based on the types of business processed by the business system within a set time. If yes, proceed to step S44; otherwise, proceed to step S43. S43 determines whether the basic monitoring extension time needs to be corrected by the abnormality rate of the accounting data of the business system during historical failures. If yes, proceed to step S44; otherwise, use the basic extension time as the extension detection time of the business system. S44 determines the correction time of the business system based on the anomaly rate of the accounting data during historical failures, the types of business processed by the business system within a set time, and the number of business processed by the business system within a set time using an evaluation model based on artificial intelligence algorithms, and determines the extended detection time of the business system based on the correction time and the basic monitoring extension time.

8. The method for periodically detecting accounting anomalies as described in claim 1, characterized in that, The detection of accounting data generated by the business system includes, but is not limited to, transaction type, changes in accounting data, and transaction volume.

9. An artificial intelligence-based system for periodic detection of accounting anomaly data, employing the artificial intelligence-based method for periodic detection of accounting anomaly data as described in any one of claims 1-8, specifically comprising: Data volume acquisition module; System fault location module; Detection timing determination module; Accounting data inspection module; The data acquisition module is responsible for acquiring the amount of data from the business system's operation within a set time period based on a preset cycle. The system failure time location module is responsible for extracting data features based on the operating data within the set time period, determining the fusion weight by using the number and type of business processed at different times within the set time period, and locating the system failure time based on the data features and the fusion weight. The detection time determination module is responsible for determining the extended detection time of the business system based on the number and type of business processed by the business system within a set time, combined with the abnormality rate of the accounting data of the business system during historical failures and the types and number of business types supported by the business system, using an artificial intelligence-based evaluation model, and determining the detection time in conjunction with the system failure time. The accounting data detection module is responsible for detecting the accounting data generated by the business system based on the detection time.

10. An artificial intelligence-based system for periodic detection of accounting anomaly data, employing the artificial intelligence-based method for periodic detection of accounting anomaly data as described in any one of claims 1-8, specifically comprising: Run the data acquisition module; Module for determining running status; Detection timing determination module; Account data detection module; The operational data acquisition module is responsible for acquiring the amount of operational data, the number of transactions processed, and the type of transactions of the business system within a set time period based on a preset cycle. The operation status determination module is responsible for judging the abnormal status of the business system and the amount of data; The detection time determination module is responsible for extracting data features based on the operating data within the set time period, determining the fusion weight using the number and type of business processed at different times within the set time period, and locating the system failure time based on the data features and fusion weight; based on the number and type of business processed by the business system within the set time period, and combined with the abnormality rate of the accounting data of the business system during historical failures and the types and number of business types supported by the business system, the extended detection time of the business system is determined using an artificial intelligence-based evaluation model, and the detection time is determined in conjunction with the system failure time; The account data detection module is responsible for detecting the accounting data generated by the business system based on the detection time.

11. A computer storage medium storing a computer program thereon, wherein when the computer program is executed in a computer, the computer executes the artificial intelligence-based periodic detection method for accounting anomaly data as described in any one of claims 1-8.

Images