Edge node operation and maintenance method and system based on cloud computing

Abstract

The present invention relates to the field of data anomaly detection technology, and specifically to an edge node operation and maintenance method and system based on cloud computing. The present invention divides the transaction time series within a day into time windows, analyzes the distribution of the transaction time series under each time window, and analyzes the differences of the transaction time series corresponding to different days under the same time window, to obtain the possibility of network failure under each time window; analyzes the outliers of the transaction amount under the time window, to obtain the possibility of operation failure corresponding to all transaction amounts in each time window; according to the possibility of network failure and the possibility of operation failure, different types of alarms and corresponding types of operation and maintenance are performed. The method and system provided by the present invention can timely and accurately identify the fault type, so that timely and targeted operation and maintenance repair of supermarket transaction equipment can be achieved, the operation and maintenance efficiency can be improved, and then the management level and safety can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data anomaly detection, and in particular to an edge node operation and maintenance method and system based on cloud computing. Background Art

[0002] The operation and maintenance process of edge nodes based on cloud computing involves multiple technical fields, including cloud computing, edge computing, and operation and maintenance management. Among them, edge nodes generally refer to computing devices or resource nodes located at the edge of the network in the edge computing architecture. These nodes are usually close to data sources or users to reduce data processing delays and bandwidth consumption. Edge nodes can be IoT devices, routers, gateways, micro data centers, etc. Operation and maintenance refers to the management and maintenance of computing resources to ensure the normal operation, performance optimization, and security of the system. Through the operation and maintenance of edge nodes, the aim is to use cloud computing technology to systematically, efficiently, and securely manage and maintain nodes in edge computing, thereby improving the performance and reliability of the entire system.

[0003] When establishing and optimizing the operation and maintenance methods of edge nodes, for example, when monitoring and maintaining edge node transaction data in supermarkets, it is necessary to accurately monitor the transaction data and check for data anomalies in real time. In existing implementation methods, usually only simple anomaly detection is performed on the transaction amount, and the specific cause of the anomaly cannot be identified, which in turn affects the accuracy and targeting of subsequent operation and maintenance of the transaction terminal, and cannot ensure stable, safe and efficient operation of the system. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide an edge node operation and maintenance method and system based on cloud computing.

[0005] According to a first aspect of an embodiment of the present invention, a cloud computing-based edge node operation and maintenance method is provided, and the technical solution adopted is specifically as follows:

[0006] Collect transaction log records of supermarket transaction equipment to obtain daily transaction time series and transaction amount series;

[0007] Divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window;

[0008] Analyze the differences of the transaction time series corresponding to different days in the same time window to obtain the time differences in each time window;

[0009] Combining the shopping inertia weight value and the time difference, obtaining the possibility of network failure in each time window;

[0010] Analyze the outliers of the transaction amounts in the time window to obtain the probability of operation failures corresponding to all the transaction amounts in each time window;

[0011] According to the possibility of the network failure and the possibility of the operation failure, different types of alarms are issued and corresponding types of operation and maintenance are performed.

[0012] In some embodiments of the present invention, the transaction time series within a day is divided into time windows, and the distribution of the transaction time series in each time window is analyzed to obtain the shopping inertia weight value in each time window, including:

[0013] Divide the transaction time series within a day into time windows, with each time window being half an hour in size;

[0014] Count the amount of transaction data within each time window to obtain the number of transactions within each time window;

[0015] Calculate the time difference between all adjacent transaction times within each time window to obtain the transaction compactness within each time window;

[0016] The shopping inertia weight value in each time window is obtained by combining the transaction times and the transaction compactness.

[0017] In some embodiments of the present invention, analyzing the differences of the transaction time series corresponding to different days in the same time window to obtain the time differences in each time window includes:

[0018] For the same time window corresponding to the current day and other different days, take the time window corresponding to a certain day when the amount of transaction time data is the largest as the reference window, and perform equal-length interpolation on the transaction time series in the same time window corresponding to different days, and the interpolation value is 0, so that the amount of transaction time data in the same time window corresponding to all days is the same as that of the reference window;

[0019] Analyze the time difference between the transaction time at the same sequence position in the transaction time sequence corresponding to the current day and other different days under the same time window to obtain the difference value of the transaction time;

[0020] All transaction times within the time window and all other different days are traversed, and the differences of the transaction times are summed to obtain the time difference under each time window.

[0021] In some embodiments of the present invention, the network failure probability in each time window is obtained, and then the following steps are further included:

[0022] Set a likelihood threshold;

[0023] According to the network failure possibility, comparing the possibility threshold, obtaining a network failure time window and a non-network failure time window;

[0024] For the network failure time window, a network failure type alarm is performed.

[0025] In some embodiments of the present invention, analyzing the outliers of the transaction amounts in a time window to obtain the possibility of operation failures corresponding to all the transaction amounts in each time window includes:

[0026] The outliers of the transaction amounts in the non-network failure time window are analyzed, and combined with the difference in the transaction time, the possibility of operation failure corresponding to all the transaction amounts in each time window is obtained.

[0027] In some embodiments of the present invention, the operation failure probability corresponding to all the transaction amounts in each time window is obtained, and then the following is further included:

[0028] Sort the probability of operation failures corresponding to all the transaction amounts in the time window from large to small, and determine the previous The transaction amount corresponding to the possibility of an operation failure is the operation failure, and the previous The sum of the probability of operation failures is greater than half of the sum of the probability of operation failures corresponding to all transaction amounts in the time window;

[0029] For transaction amounts with operational failures, an alarm will be issued based on the type of operational failure.

[0030] In some embodiments of the present invention, corresponding types of operation and maintenance are performed, including:

[0031] In response to network fault type alarms, local cache mechanism settings and asynchronous processing of transaction data are performed, including:

[0032] The local cache mechanism setting includes: setting up a local storage cache on the supermarket transaction device to temporarily save transaction log records, setting a cache size limit, and regularly clearing processed data;

[0033] Asynchronous processing of transaction data includes uploading transaction log records from the local cache to the server, setting up an automatic retry mechanism when the upload fails, and adjusting the upload frequency according to the network status.

[0034] In some embodiments of the present invention, the transaction log records of supermarket transaction devices are collected, and then the following steps are further included: data cleaning, data normalization and data storage of the transaction log records.

[0035] According to a second aspect of an embodiment of the present invention, a cloud computing-based edge node operation and maintenance system is provided, including: a memory and a processor, wherein:

[0036] The memory is used to store program code;

[0037] The processor is used to read the program code stored in the memory and execute the method described in the first aspect of the embodiment of the present invention.

[0038] In some embodiments of the present invention, the processor comprises:

[0039] A transaction data collection module, used to collect transaction log records of supermarket transaction devices, wherein the transaction log records include transaction time series and transaction amount series;

[0040] The network failure possibility analysis module is first used to divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window; then used to analyze the difference of the transaction time series corresponding to different days in the same time window, and obtain the time difference in each time window; finally used to obtain the network failure possibility in each time window according to the shopping inertia weight value and the time difference;

[0041] An operation failure possibility analysis module is used to analyze the outliers of the transaction amounts in a time window and obtain the operation failure possibilities corresponding to all the transaction amounts in each time window;

[0042] The alarm and operation and maintenance module is used to issue different types of alarms and perform corresponding types of operation and maintenance according to the possibility of the network failure and the possibility of the operation failure.

[0043] Compared with the prior art, the edge node operation and maintenance method and system based on cloud computing provided by the present invention obtains the possibility of network failure in each time window through the analysis of transaction time history matching and the shopping inertia weight value of different time windows, and judges the time window where network fluctuation failure may occur; for more comprehensive monitoring, for the time window where no network failure occurs, further analyzes the outliers of the transaction amount in the time window, obtains the possibility of operation failure corresponding to all transaction amounts in each time window, and judges the time window where operation failure may occur; finally, according to the fault type, the corresponding type of alarm and operation and maintenance processing are performed. The method and system can timely and accurately identify the fault type, can timely and targetedly perform operation and maintenance repairs on supermarket transaction equipment, improve the operation and maintenance efficiency, and thus improve the management level and safety of supermarket transaction equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A basic flow chart of an edge node operation and maintenance method based on cloud computing provided by an embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the basic composition of an edge node operation and maintenance system based on cloud computing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the edge node operation and maintenance method and system based on cloud computing proposed by the present invention, its specific implementation method, structure, features and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0048] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. Terms such as "comprises", "comprising" or any other variants thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of further restrictions, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the article or device including the element.

[0049] The following is a detailed description of a cloud computing-based edge node operation and maintenance method provided by the present invention with reference to the accompanying drawings.

[0050] See also Figure 1 , which shows the basic process of an edge node operation and maintenance method based on cloud computing provided by an embodiment of the present invention.

[0051] like Figure 1 As shown, an edge node operation and maintenance method based on cloud computing provided by an embodiment of the present invention specifically includes:

[0052] S100: Collect transaction log records of supermarket transaction devices to obtain daily transaction time series and transaction amount series.

[0053] Obtain transaction log records from the supermarket's transaction equipment (cash register terminal) in real time. Transaction log records include transaction time, transaction amount, terminal ID, operator ID, etc. Among them, transaction time (Transaction Time) is the exact time when the transaction occurred, for example: xxxx-xx-xx (year-month-day) 14:25:32; transaction amount (Transaction Amount) is the amount of each transaction, for example: 99.99 yuan; terminal ID (Terminal ID) is the unique ID that identifies the transaction device, for example: POS001; operator ID (Operator ID) is the operator number responsible for the transaction, for example: OP123. Transaction log record data can be transmitted in real time through API interfaces, message queues, or database scheduled capture.

[0054] In some embodiments of the present invention, in order to facilitate subsequent data analysis, the transaction log records of supermarket transaction devices are collected, and then the transaction log records are cleaned, normalized and stored.

[0055] Data cleaning includes:

[0056] Deduplication: Remove duplicate transaction log records. For example, if the same transaction is recorded multiple times, only one of them will be kept.

[0057] Format standardization: All fields have the same format, transaction time is converted to a unified time zone format, and amount fields are uniformly reserved to two decimal places;

[0058] Missing value processing: Check whether there is any missing key data (such as transaction amount, transaction time), supplement the missing data (for example, fill it with the last valid data) or delete invalid records.

[0059] Data normalization includes: converting transaction amounts into a unified currency unit (such as yuan); converting transaction times into a unified transaction timestamp format for subsequent analysis.

[0060] Data storage includes: storing processed data in a database or distributed storage system to facilitate subsequent query, analysis and report generation.

[0061] In addition, the transaction log records for each day exist in the form of a matrix, and the transaction time series and transaction sequence of each day are extracted. In addition to reflecting the transaction time and transaction amount, the transaction log record matrix also corresponds to parameters such as the terminal ID and operator ID of each transaction. Such vectors are not listed one by one here.

[0062] In order to ensure efficient troubleshooting, anomaly repair, and improve the safety level during the operation and maintenance of supermarket transaction equipment, the transaction data parameters and their types in the transaction log records are analyzed. The unique characteristics of different transaction data parameter types are analyzed according to their attributes, and the unique characteristics are quantified as the fault reference index value corresponding to each transaction data parameter. For example, for transaction time data, its main characteristics are historical similarity and interval activity. Historical similarity is manifested in having a certain shopping inertia weight value at different times. For example, the purchase of breakfast by users in the morning and the purchase of other needs by users in the afternoon are all concentrated at the commuting time nodes. There are more people, so the transaction time is more concentrated; and the transaction time and transaction quantity will not change significantly at the same time on the second day. The time interval is manifested in that if there is no transaction for a long time or the transaction interval time increases abnormally during the daily active time period, then the corresponding abnormal characteristics in the time period are higher, and the delay characteristics caused by the abnormality due to network fluctuations are higher. Therefore, while obtaining the possibility of abnormality, it also has a certain ability to identify the cause of abnormality.

[0063] Regardless of transaction time or transaction amount, its purpose is to detect the real-time possibility of abnormality in the edge node of supermarket transaction data, and to operate and maintain it to maintain the security of the transaction process. Therefore, this embodiment performs real-time abnormality possibility analysis on the edge node based on the transaction time sequence and the transaction amount sequence, and the specific steps include S200 to S500.

[0064] S200: Divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window.

[0065] The information corresponding to the transaction time series indicates the time when each transaction occurred. Among general transaction equipment anomalies, the majority of transaction data transmission and sending anomalies are caused by network problems (unstable or interrupted network connection, resulting in the inability to transmit transaction data normally or loss). The manifestation is: the network fluctuates within a period of time, and the transaction data cannot be transmitted in time. The corresponding transaction time series will have relatively discrete time gaps, which means that there is no transaction data at a time node.

[0066] The above performance characteristics can be quantified through specific time differences in the transaction time series. For example, in the historical data of the previous few days, there is a certain amount of transaction data in the same time node or time period, but at the current moment, the data volume and transaction time are unstable or abnormal, which needs to be analyzed.

[0067] Based on the above analysis, this implementation divides the transaction time series within a day into time windows, analyzes the distribution of the transaction time series in each time window, and obtains the shopping inertia weight value in each time window.

[0068] Specifically, first, the transaction time series within a day is divided into time windows, and the size of each time window is half an hour. There are several transaction time data in each time window. The more the total amount of transaction time data in a time window, the higher the corresponding transaction volume; and the smaller the transaction time interval in a time window, that is, the more compact the transaction, the higher the corresponding transaction volume; therefore, this embodiment obtains the number of transactions within each time window by counting the amount of transaction time data within each time window; and calculates the time difference between all adjacent transaction times within each time window to obtain the transaction compactness within each time window; and then, combining the number of transactions and the transaction compactness, obtains the shopping inertia weight value under each time window.

[0069] The calculation formula for constructing the shopping inertia weight value in each time window is:

[0070]

[0071] in, Indicates the current day The corresponding shopping inertia weight value in the time window; Indicates the number of The number of transactions within a time window, that is, the specific amount of transaction time. The record of one transaction time is one transaction volume. Indicates the current day The transaction time series in the time window The transaction timestamp value corresponding to the transaction time; Indicates the current day The transaction time series in the time window The transaction timestamp value corresponding to the transaction time; Represents the maximum value function.

[0072] The model determines the shopping inertia weight value in the corresponding time window by analyzing the number of transactions in the time window and the compactness of the corresponding transaction time. The weight value is the weight of the historical matching of the transaction time in the analysis time window. The total amount of transaction time data in a time window The higher it is, the more active the corresponding shopping is, which means the corresponding shopping inertia weight value in this time window is higher. It represents the maximum value of the time difference between the transaction timestamp value corresponding to all subsequent transaction times and the transaction timestamp value corresponding to the previous transaction time in the transaction time series within the time window. The larger the time difference, the wider the corresponding transaction interval within the time window, and the less compact the transaction. That is, during this time period, there are generally fewer customers and the shopping volume is lower. In this case, the corresponding shopping inertia weight value is lower, because there is no corresponding shopping inertia weight value. On the contrary, when the time difference is smaller, the corresponding shopping inertia weight value is higher, and the weight value is higher.

[0073] S300: Analyze the differences of transaction time series corresponding to different days in the same time window to obtain the time differences in each time window.

[0074] Within a day, different time window ranges correspond to different shopping inertia weight values. Combined with the differences in shopping inertia and transaction time series in different parts of the world, the historical matching of transaction time is obtained.

[0075] Therefore, this embodiment obtains the time difference under each time window by analyzing the difference of the transaction time series corresponding to different days under the same time window. Specifically, first, for the same time window corresponding to the current day and other different days, take the time window corresponding to a certain day when the transaction time data volume is the largest as the reference window, and perform equal-length interpolation on the transaction time series in the same time window corresponding to different days, and the interpolation value is 0, so that the transaction time data volume in the same time window corresponding to all days is the same as the reference window; then, analyze the time difference of the transaction time at the same sequence position in the transaction time series corresponding to the current day and other different days under the same time window, and obtain the difference of transaction time; finally, traverse all transaction times in the time window, and traverse all other different days, sum the difference of transaction time, and obtain the time difference under each time window.

[0076] The time difference calculation formula for each time window is constructed as follows:

[0077]

[0078] in, Indicates the number of days corresponding to the current day and other different days. Time differences under a time window; Indicates Day (Current Day) In the time window The transaction timestamp value corresponding to the transaction time; Indicates Days ago Day) In the time window The transaction timestamp value corresponding to the transaction time; Indicates The total amount of transaction time data within a time window; Indicates the number of previous days involved in the calculation.

[0079] By analyzing the difference in transaction timestamps corresponding to different days in the same time window , determine the corresponding time difference. Among them, when the number of transaction times in a certain day's time window does not match (there is no object to be subtracted), the time window with less transaction time data is filled in by the equal-length interpolation method, and the transaction timestamp value filled in is 0. In this way, the corresponding time difference will increase, because the number of transaction times does not match, and the corresponding transaction time is more difficult to align.

[0080] S400: Combining the shopping inertia weight value and the time difference, the possibility of network failure in each time window is obtained.

[0081] After obtaining the shopping inertia weight value and the time difference, the shopping inertia weight value is used as the weight of the time difference to obtain the possibility of network failure in each time window.

[0082] The calculation formula for the probability of network failure in each time window is:

[0083]

[0084] In the formula, Indicates The probability of network failure in a time window; Indicates the number of days corresponding to the current day and other different days. Time differences under a time window; Indicates the current day The corresponding shopping inertia weight value in the time window; represents the linear normalization function.

[0085] Possibility of network failure Value size and time difference It is proportional to the difference between transaction timestamps. The higher the difference, the higher the possibility of failure caused by network fluctuation. This is because the network fluctuates within a period of time, and transaction data cannot be transmitted in time. The corresponding transaction time series will have relatively discrete time gaps, which means that there is no transaction data in this time segment. In addition, the possibility of network failure The value is inversely proportional to the shopping inertia weight value, because the higher the weight value, the more normal the shopping inertia and shopping logic reflected, and the less likely it is that a network failure will occur.

[0086] In some embodiments of the present invention, the possibility of network failure in each time window is obtained, and then the following steps are further included: setting a possibility threshold, which may be 0.8; comparing the possibility threshold according to the network failure possibility, and when the network failure possibility is greater than the possibility threshold, that is, When the time window is a network failure time window, it is determined that the time window is a network failure time window; otherwise, the time window is determined to be a non-network failure time window; for the network failure time window, a network failure type alarm is issued.

[0087] S500: Analyze the outliers of the transaction amounts in the time window to obtain the possibility of operation failures corresponding to all transaction amounts in each time window.

[0088] Through the analysis of the possibility of network failure, we can get the network failure time window and the non-network failure time window. For the non-network failure time window, it means that the constants such as the number of transactions and transaction time in this part of the time window are relatively normal, and there is no delay caused by network fluctuations. However, in order to achieve the purpose of comprehensive monitoring, it is necessary to capture and analyze other parameter values ​​corresponding to each transaction time, such as transaction amount and terminal ID.

[0089] Based on the above analysis, this embodiment analyzes the outliers of the transaction amounts in the time window to obtain the operational failure probability corresponding to all transaction amounts in each time window. Further, it includes: analyzing the outliers of the transaction amounts in the non-network failure time window by the Z-score algorithm (with related calculation weights), and combining the difference in transaction time to obtain the operational failure probability corresponding to all transaction amounts in each time window.

[0090] The calculation formula for constructing the probability of operation failure corresponding to all transaction amounts in each time window is:

[0091]

[0092] in, Indicates In the window The probability of operational failure corresponding to each transaction amount; , indicating the The first day corresponding to the current day and all other different days in the window The sum of the differences of transaction timestamps; The normalized sum of the differences in transaction timestamps is used as the weight; Indicates In the window The size of the outlier corresponding to the transaction amount.

[0093] The size of the outlier corresponding to the transaction amount The larger the value is, the greater the possibility of operational failure of the transaction amount data, because the greater the difference in its value distribution, the more likely it is that the failure is caused by staff misoperation. In addition, when the transaction time data corresponding to the transaction amount is less different from the historical data, the corresponding operational failure possibility is smaller, because the more normal the transaction time is, the lower the possibility of misoperation.

[0094] Obtain the operational failure probability corresponding to all transaction amounts in each time window, and then also include: sorting the operational failure probability corresponding to all transaction amounts in the time window from large to small, and judging the previous The transaction amount corresponding to the possibility of an operation failure is the operation failure, and the previous The sum of the possibilities of operation failures is greater than half of the sum of the possibilities of operation failures corresponding to all transaction amounts within the time window; for the transaction amounts with operation failures, an operation failure type alarm is issued.

[0095] S600: Different types of alarms are issued and corresponding types of operation and maintenance are performed according to the possibility of network failure and the possibility of operation failure.

[0096] The network failure time window is obtained based on the network failure possibility; the operation failure time window is obtained based on the operation failure possibility. For the network failure time window, a network failure type alarm is performed, which can be a network fluctuation, and the alarm form can be network fluctuation + failure possibility; for the operation failure time window, an operation failure type alarm is performed, which can be an operation error or an amount error, and the alarm form can be an operation error or an amount error + failure possibility.

[0097] In addition, since the terminal ID is unique, any faults can be detected by matching analysis, so the faults of such edge nodes are easy to be discovered and eliminated.

[0098] Then, the corresponding type of operation and maintenance is carried out according to the alarm type. For the network failure type alarm, the local cache mechanism is set up and the transaction data is processed asynchronously, where:

[0099] The local cache mechanism setting includes: setting up a local storage cache on the supermarket transaction device to temporarily save transaction log records, setting a cache size limit, and regularly clearing processed data;

[0100] Asynchronous processing of transaction data includes: uploading transaction log records from the local cache to the server without blocking terminal operations; setting an automatic retry mechanism when the upload fails to ensure that the data can eventually be uploaded successfully; and adjusting the upload frequency according to the network status, that is, increasing the upload frequency when the network recovers and reducing the upload frequency when the network is unstable.

[0101] In addition, if there are problems such as erroneous operation and terminal ID, you can simply repeat the operation again.

[0102] Based on the same inventive concept as the above method, this embodiment also provides an edge node operation and maintenance system based on cloud computing.

[0103] See also Figure 2 , which shows the basic composition of an edge node operation and maintenance system based on cloud computing provided by an embodiment of the present invention.

[0104] like Figure 2 As shown, an edge node operation and maintenance system based on cloud computing includes: a memory 10 and a processor 20, wherein:

[0105] A memory 10, used for storing program codes;

[0106] The processor 20 is used to read the program code stored in the memory 10, and execute the transaction log records of the supermarket transaction equipment to obtain the transaction time series and transaction amount series of each day; divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window; analyze the difference of the transaction time series corresponding to different days in the same time window to obtain the time difference in each time window; combine the shopping inertia weight value and the time difference to obtain the possibility of network failure in each time window; analyze the outliers of the transaction amount in the time window to obtain the possibility of operation failure corresponding to all transaction amounts in each time window; according to the possibility of network failure and the possibility of operation failure, different types of alarms and corresponding types of operation and maintenance are performed.

[0107] Furthermore, the processor 20 includes a transaction data collection module 21, a network fault possibility analysis module 22, an operation fault possibility analysis module 23 and an alarm and operation and maintenance module 24. Among them:

[0108] The transaction data collection module 21 is used to collect transaction log records of supermarket transaction devices, and the transaction log records include transaction time series and transaction amount series;

[0109] The network failure possibility analysis module 22 is first used to divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window; then used to analyze the difference of the transaction time series corresponding to different days in the same time window, and obtain the time difference in each time window; finally, used to obtain the network failure possibility in each time window according to the shopping inertia weight value and time difference;

[0110] Operational failure possibility analysis module 23, used to analyze the outliers of transaction amounts in a time window, and obtain the operational failure possibilities corresponding to all transaction amounts in each time window;

[0111] The alarm and operation and maintenance module 24 is used to generate different types of alarms and perform corresponding types of operation and maintenance according to the possibility of network failure and the possibility of operation failure.

[0112] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0113] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A cloud computing-based edge node operation and maintenance method, characterized in that: The method comprises: Collect transaction log records of supermarket transaction equipment to obtain daily transaction time series and transaction amount series; Divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window; Analyze the differences of the transaction time series corresponding to different days in the same time window to obtain the time differences in each time window; Combining the shopping inertia weight value and the time difference, obtaining the possibility of network failure in each time window; Analyze the outliers of the transaction amounts in the time window to obtain the probability of operation failures corresponding to all the transaction amounts in each time window; According to the possibility of the network failure and the possibility of the operation failure, different types of alarms are issued and corresponding types of operation and maintenance are performed; Analyze the differences of the transaction time series corresponding to different days in the same time window to obtain the time differences in each time window, including: For the same time window corresponding to the current day and other different days, take the time window corresponding to a certain day when the amount of transaction time data is the largest as the reference window, and perform equal-length interpolation on the transaction time series in the same time window corresponding to different days, and the interpolation value is 0, so that the amount of transaction time data in the same time window corresponding to all days is the same as that of the reference window; Analyze the time difference between the transaction time at the same sequence position in the transaction time sequence corresponding to the current day and other different days under the same time window to obtain the difference value of the transaction time; Traversing all transaction times within the time window and traversing all other different days, summing the differences of the transaction times, and obtaining the time difference under each time window; The method for obtaining the shopping inertia weight value includes: dividing the transaction time series within a day into time windows, each time window being half an hour in size; counting the amount of transaction time data within each time window to obtain the number of transactions within each time window; calculating the time difference between all adjacent transaction times within each time window to obtain the transaction compactness within each time window; combining the transaction number and the transaction compactness to obtain the shopping inertia weight value under each time window; The method for obtaining the possibility of an operation failure includes: performing an outlier analysis on the transaction amount in a non-network failure time window by using a Z-score algorithm, and combining the difference in transaction time to obtain the possibility of an operation failure corresponding to all transaction amounts in each time window.

2. The edge node operation and maintenance method based on cloud computing according to claim 1, characterized in that: Get the probability of network failure in each time window, and then include: Set a likelihood threshold; According to the network failure possibility, comparing the possibility threshold, obtaining a network failure time window and a non-network failure time window; For the network failure time window, a network failure type alarm is performed.

3. The edge node operation and maintenance method based on cloud computing according to claim 2, characterized in that: Obtaining the operational failure probability corresponding to all the transaction amounts in each time window, and then further including: Sort the probability of operation failures corresponding to all the transaction amounts in the time window from large to small, and determine the previous The transaction amount corresponding to the possibility of an operation failure is the operation failure, and the previous The sum of the probability of operation failures is greater than half of the sum of the probability of operation failures corresponding to all transaction amounts in the time window; For transaction amounts with operational failures, an alarm will be issued based on the type of operational failure.

4. The edge node operation and maintenance method based on cloud computing according to claim 1, characterized in that: Perform corresponding types of operations and maintenance, including: In response to network fault type alarms, local cache mechanism settings and asynchronous processing of transaction data are performed, including: The local cache mechanism setting includes: setting up a local storage cache on the supermarket transaction device to temporarily save transaction log records, setting a cache size limit, and regularly clearing processed data; Asynchronous processing of transaction data includes uploading transaction log records from the local cache to the server, setting up an automatic retry mechanism when the upload fails, and adjusting the upload frequency according to the network status.

5. The edge node operation and maintenance method based on cloud computing according to claim 1, characterized in that: The transaction log records of supermarket transaction devices are collected, and then the transaction log records are cleaned, normalized and stored.

6. An edge node operation and maintenance system based on cloud computing, characterized in that: The system comprises: a memory and a processor, wherein: The memory is used to store program codes; The processor is configured to read the program code stored in the memory and execute the method according to any one of claims 1 to 5.

7. The edge node operation and maintenance system based on cloud computing according to claim 6, characterized in that: The processor comprises: A transaction data collection module, used to collect transaction log records of supermarket transaction devices, wherein the transaction log records include transaction time series and transaction amount series; The network failure possibility analysis module is first used to divide the transaction time series within a day into time windows, analyze the distribution of the transaction time series in each time window, and obtain the shopping inertia weight value in each time window; then used to analyze the difference of the transaction time series corresponding to different days in the same time window, and obtain the time difference in each time window; finally used to obtain the network failure possibility in each time window according to the shopping inertia weight value and the time difference; An operation failure possibility analysis module is used to analyze the outliers of the transaction amounts in a time window and obtain the operation failure possibilities corresponding to all the transaction amounts in each time window; An alarm and operation and maintenance module, used to generate different types of alarms and perform corresponding types of operation and maintenance according to the possibility of the network failure and the possibility of the operation failure; Analyze the differences of the transaction time series corresponding to different days in the same time window to obtain the time differences in each time window, including: For the same time window corresponding to the current day and other different days, take the time window corresponding to a certain day when the amount of transaction time data is the largest as the reference window, and perform equal-length interpolation on the transaction time series in the same time window corresponding to different days, and the interpolation value is 0, so that the amount of transaction time data in the same time window corresponding to all days is the same as that of the reference window; Analyze the time difference between the transaction time at the same sequence position in the transaction time sequence corresponding to the current day and other different days under the same time window to obtain the difference value of the transaction time; Traversing all transaction times within the time window and traversing all other different days, summing the differences of the transaction times, and obtaining the time difference under each time window; The method for obtaining the shopping inertia weight value includes: dividing the transaction time series within a day into time windows, each time window being half an hour in size; counting the amount of transaction time data within each time window to obtain the number of transactions within each time window; calculating the time difference between all adjacent transaction times within each time window to obtain the transaction compactness within each time window; combining the transaction number and the transaction compactness to obtain the shopping inertia weight value under each time window; The method for obtaining the possibility of an operation failure includes: performing an outlier analysis on the transaction amount in a non-network failure time window by using a Z-score algorithm, and combining the difference in transaction time to obtain the possibility of an operation failure corresponding to all transaction amounts in each time window.

Images