Fault diagnosis method, device, terminal equipment and network platform

Abstract

The application provides a fault diagnosis method and device, a terminal device and a network platform. The method comprises the following steps: collecting running interaction data of the terminal device under a network connection fault condition; performing fault analysis according to the running interaction data to obtain fault analysis data; and reporting the fault analysis data to the network platform after the network is restored to normal. By using the method, the terminal device and the network platform perform data collection and fault detection respectively, the network fault reason is obtained through end-side collaborative analysis, the network fault can be effectively solved, and the labor cost and operation and maintenance cost are reduced.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a fault diagnosis method, device, terminal equipment, and network platform. Background Technology

[0002] Current automated and digital operation and maintenance methods are mainly achieved through remote network operation and maintenance platforms. Terminal devices report data such as cellular network operation and performance to the remote network operation and maintenance platform in real time, and the remote network operation and maintenance platform performs unified analysis, fault diagnosis, and location and delimitation.

[0003] In industry networks, device outages are a significant, urgent, and frequent failure, closely related to production safety. According to user feedback work orders compiled by multiple manufacturers, connection anomalies account for a large proportion of the issues. Causes of outages can include SIM card problems (e.g., unpaid bills), environmental factors (e.g., power outages), device malfunctions (e.g., system crashes), or network link failures leading to communication interruptions. Current methods primarily rely on real-time data analysis. However, regardless of the cause, the network platform cannot obtain real-time data from the terminal devices to analyze the cause. Consequently, most outages require maintenance personnel to physically inspect logs on-site, then contact the equipment manufacturer or operator to resolve the issue, consuming substantial manpower and time. Summary of the Invention

[0004] The purpose of this invention is to provide a fault diagnosis method, device, terminal equipment, and network platform to solve the problem that when existing equipment fails and goes offline, manual troubleshooting is required, which makes it impossible to solve the problem of equipment disconnection in a timely and effective manner, and also incurs a lot of manpower and time costs.

[0005] This invention provides a fault diagnosis method, which is executed by a terminal device, and the method includes:

[0006] In the event of a network connection failure, the operational interaction data of the terminal device shall be collected;

[0007] Fault analysis is performed based on the operational interaction data to obtain fault analysis data;

[0008] After the network is restored to normal, the fault analysis data is reported to the network platform.

[0009] Optionally, the fault diagnosis method further includes:

[0010] Under normal network connectivity conditions, predictive analysis is performed based on the operational interaction data of the terminal devices to obtain a prediction result on whether a network failure will occur within a first preset time period.

[0011] When the prediction indicates that a network failure will occur, an alarm message is reported to the network platform.

[0012] Optionally, in the fault diagnosis method, the operational interaction data includes AT commands and / or module interaction logs.

[0013] Optionally, the fault diagnosis method, wherein fault analysis is performed based on the operational interaction data to obtain fault analysis data, includes:

[0014] Based on the operational interaction data, perform one or more of the following fault analyses to obtain the fault analysis data:

[0015] SIM card problem analysis;

[0016] Module problem analysis;

[0017] Network coverage problem analysis;

[0018] Network connectivity problem analysis;

[0019] Authentication issue analysis;

[0020] Analysis of invalid message issues.

[0021] Optionally, in the fault diagnosis method, after performing fault analysis based on the operational interaction data to obtain fault analysis data, the method further includes:

[0022] The fault type of the network connection failure is obtained based on the fault analysis data;

[0023] The gateway indicator light controlling the terminal device flashes in a manner corresponding to the fault type to indicate the problem.

[0024] This invention also provides a fault diagnosis method, which is executed by a network platform, the method comprising:

[0025] Acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs;

[0026] Based on the fault analysis data, an abnormal sequence template for the terminal device is established;

[0027] When a network connection failure of the terminal device is detected, abnormal status detection is performed on the real-time network data to obtain abnormal data.

[0028] The abnormal data is matched with the abnormal sequence template to determine the fault type of the network connection failure.

[0029] Optionally, in the fault diagnosis method, the real-time network data includes SIM card data of the terminal device obtained by the SIM card management platform, and / or network status data collected by the network platform within a second preset time period before the network connection failure occurs.

[0030] Optionally, the fault diagnosis method, wherein abnormal state detection of real-time network data includes one or more of the following:

[0031] Perform abnormal status detection on the SIM card data;

[0032] Perform abnormal state detection on the network quality index data in the network status data;

[0033] Perform abnormal state detection on the traffic indicator data in the network status data;

[0034] Anomaly detection is performed on the device indicator data in the network status data.

[0035] Optionally, in the fault diagnosis method, the step of detecting abnormal states in real-time network data to obtain abnormal data includes:

[0036] The real-time network data is preprocessed to obtain processed data of the real-time network data;

[0037] Feature extraction is performed on the processed data to obtain the feature matrix of the processed data;

[0038] Multiple detection sample data of the feature matrix are obtained by constructing a binary tree;

[0039] For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal data.

[0040] Optionally, in the fault diagnosis method, the anomaly judgment for each test sample data, determining whether the corresponding test sample data is abnormal data, includes:

[0041] Calculate the path length of the detected sample data in the constructed binary tree;

[0042] Calculate the anomaly score of the detected sample data based on the path length;

[0043] When the abnormal score is greater than a preset judgment value, the corresponding detection sample data is determined to be abnormal data.

[0044] Optionally, in the fault diagnosis method, feature extraction of the processed data to obtain a feature matrix of the processed data includes:

[0045] The processed data are calculated using different mean calculation methods to obtain the predicted value corresponding to each mean calculation method;

[0046] A feature matrix of the processed data is generated based on the difference between each processed data and a different predicted value.

[0047] Optionally, in the fault diagnosis method, matching the abnormal data with the abnormal sequence template to determine the fault type of the network fault includes:

[0048] Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template;

[0049] Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix;

[0050] Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters;

[0051] Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template;

[0052] If the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network fault corresponding to the abnormal data is determined according to the matching abnormal sequence template.

[0053] Optionally, in the fault diagnosis method, determining whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost includes:

[0054] Determine whether the cumulative cost is less than a preset value;

[0055] If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

[0056] Optionally, the fault diagnosis method further includes:

[0057] Based on the fault analysis data and / or the abnormal data obtained by detecting abnormal states in real-time network data, model update information is generated.

[0058] The model update information is sent to the terminal device; the model update information includes first model update information and / or second model update information; the first model update information is used to instruct the terminal device to update the first fault analysis model used for fault analysis, and the second model update information is used to instruct the first fault prediction model used for fault prediction to be updated.

[0059] This invention also provides a terminal device, including a processor and a transceiver, wherein:

[0060] The processor is used to collect operational interaction data of the terminal device in the event of a network connection failure; and

[0061] Fault analysis is performed based on the operational interaction data to obtain fault analysis data;

[0062] The transceiver is used to report the fault analysis data to the network platform after the network returns to normal.

[0063] This invention also provides a network platform, including a transceiver and a processor, wherein:

[0064] The transceiver is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs;

[0065] The processor is configured to: establish an abnormal sequence template for the terminal device based on the fault analysis data; perform abnormal state detection on real-time network data to obtain abnormal data when a network connection fault of the terminal device is detected; and match the abnormal data with the abnormal sequence template to determine the fault type of the network connection fault.

[0066] This invention also provides a fault diagnosis device, which is applied to a terminal device, and the device includes:

[0067] The data acquisition module is used to collect the operational interaction data of the terminal device in the event of a network connection failure.

[0068] The analysis module is used to perform fault analysis based on the operational interaction data and obtain fault analysis data.

[0069] The reporting module is used to report the fault analysis data to the network platform after the network returns to normal.

[0070] This invention also provides a fault diagnosis device, which is applied to a network platform, and the device includes:

[0071] The information acquisition module is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs;

[0072] The template creation module is used to create an anomaly sequence template for the terminal device based on the fault analysis data.

[0073] The detection module is used to detect abnormal states in real-time network data and obtain abnormal data when a network connection failure of the terminal device is detected.

[0074] The matching module is used to match the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure.

[0075] This invention also provides a terminal device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the fault diagnosis method as described in any of the preceding embodiments.

[0076] This invention also provides a network platform, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the fault diagnosis method as described in any of the preceding embodiments.

[0077] This invention also provides a readable storage medium, wherein a program is stored on the readable storage medium, and when the program is executed by a processor, it implements the steps in the fault diagnosis method described in any of the preceding claims.

[0078] At least one of the above technical solutions of the present invention has the following beneficial effects:

[0079] Using the fault diagnosis method described in this embodiment of the invention, when a terminal device experiences a network connection failure, the terminal device collects operational interaction data and performs fault analysis to obtain fault analysis data. After the network returns to normal, the terminal device reports the fault analysis data to the network platform, enabling the network platform to use the fault analysis data to establish an anomaly sequence template. After detecting network faults in the terminal, the detected anomaly data can be matched with the anomaly sequence template to determine the cause of the network fault. This achieves the effect of collaborative analysis and determination of network faults between the terminal device and the network platform, thereby reducing the cost of manual analysis and maintenance. Attached Figure Description

[0080] Figure 1 This is a flowchart illustrating the fault diagnosis method according to one embodiment of the present invention;

[0081] Figure 2This is a flowchart illustrating a fault diagnosis method according to another embodiment of the present invention;

[0082] Figure 3 This is a schematic diagram of the overall process of the fault diagnosis method described in the embodiments of the present invention;

[0083] Figure 4 This is a schematic diagram of the structure of the terminal device described in an embodiment of the present invention;

[0084] Figure 5 This is a schematic diagram of the network platform described in an embodiment of the present invention;

[0085] Figure 6 This is a schematic diagram of the fault diagnosis device according to one embodiment of the present invention;

[0086] Figure 7 This is a schematic diagram of the fault diagnosis device according to another embodiment of the present invention. Detailed Implementation

[0087] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0088] To address the issue that existing technologies require manual troubleshooting when equipment malfunctions and disconnects, which leads to untimely and ineffective solutions and incurs significant manpower and time costs, this invention provides a fault diagnosis method. This method involves data collection and fault detection by both the terminal device and the network platform, enabling collaborative analysis between the end and edge to identify the cause of network faults. This approach allows for effective resolution of network faults while reducing manpower and maintenance costs.

[0089] One embodiment of the present invention provides a fault diagnosis method, executed by a terminal device, such as... Figure 1 As shown, the method includes:

[0090] S110, in the event of a network connection failure, collects operational interaction data from terminal devices;

[0091] S120, perform fault analysis based on the operational interaction data to obtain fault analysis data;

[0092] S130: After the network returns to normal, report the fault analysis data to the network platform.

[0093] The fault diagnosis method described in this invention allows the terminal device to collect operational interaction data and perform fault analysis when a network connection failure occurs. After the network returns to normal, the terminal device reports the fault analysis data to the network platform, enabling the platform to establish anomaly sequence templates. After detecting network faults at the terminal, the detected abnormal data can be matched with the anomaly sequence templates to determine the cause of the network fault. This achieves the effect of collaborative analysis and determination of network faults between the terminal device and the network platform, providing assurance for effective resolution of network faults and reducing labor and maintenance costs.

[0094] In one embodiment, the fault analysis data may optionally include the correspondence between the fault diagnosis results based on the operational interaction data and the operational interaction data.

[0095] In one embodiment of the fault diagnosis method described in this invention, the method further includes:

[0096] When a terminal device detects an initial network access failure or a dropped connection, it determines that the network connection is faulty.

[0097] Using this implementation method, in the event of initial network access failure or network disconnection, the terminal device collects operational interaction data for fault analysis; after the terminal device's network is restored to normal, that is, after successful network connection, it reports fault indication information to the network platform.

[0098] In one embodiment, the operational interaction data includes AT commands and / or module interaction logs.

[0099] In this implementation method, when a network connection fails, the terminal device performs fault analysis by acquiring AT commands and / or module interaction logs. This includes analyzing SIM card faults, module faults, network faults, authentication problems, and invalid message problems, respectively, to identify the anomalies that caused the fault and obtain the final fault analysis data.

[0100] Optionally, in step S120, fault analysis is performed based on the operational interaction data to obtain fault analysis data, including:

[0101] Based on the operational interaction data, perform one or more of the following fault analyses to obtain fault analysis data:

[0102] SIM card fault analysis;

[0103] Module fault analysis;

[0104] Network coverage problem analysis;

[0105] Network connectivity problem analysis;

[0106] Authentication issue analysis;

[0107] Analysis of invalid message issues.

[0108] By employing this implementation method, and utilizing AT commands and / or module interaction logs to perform one or more of the aforementioned fault analyses, a comprehensive analysis of network faults can be ensured to obtain accurate fault analysis results.

[0109] Optionally, after performing fault analysis based on the operational interaction data to obtain fault analysis data, the method further includes:

[0110] The fault type of the network connection failure is obtained based on the fault analysis data;

[0111] The gateway indicator light controlling the terminal device flashes in a manner corresponding to the fault type to indicate the problem.

[0112] Specifically, after performing fault analysis based on operational interaction data and obtaining fault analysis data, the gateway indicator light is used to provide different colors and / or flashing patterns to indicate the different types of network connection faults indicated by the fault analysis data, so as to clearly indicate the type of fault that caused the network fault.

[0113] For example, in step S120, fault analysis is performed based on the operational interaction data to obtain fault analysis data, which may include one or more of the following analyses:

[0114] SIM card fault analysis: SIM card error codes are detected using AT commands. If error codes such as "SIM failure," "SIMbusy," or "SIM wrong" are detected, a SIM card fault is detected. One implementation method involves continuously illuminating a red indicator light on the gateway when a SIM card fault is detected.

[0115] Module fault analysis: Abnormal interaction is detected through AT commands. When an abnormal heartbeat occurs, the terminal device will attempt to repair it by restarting, thus detecting the abnormal AT interaction and determining that the module is abnormal or faulty. Among them, module abnormality can be indicated by the gateway indicator light flashing red.

[0116] Network coverage fault analysis: Obtain the Reference Signal Received Power (RSRP) value via AT commands. If the RSRP is less than a preset value, such as less than 105 dBm, a coverage anomaly can be determined. Network coverage anomalies can be indicated by a continuously illuminated yellow light on the gateway indicator.

[0117] Authentication Problem Analysis: AT commands are used to detect UE authentication-related error codes. If error codes such as "Illegal UE," "UE identity cannot be derived by the network," or "Implicitly deregistered" are detected, an authentication problem is confirmed. Terminal authentication problems can be indicated by a flashing yellow light on the gateway indicator.

[0118] Network connectivity problem analysis: Obtain error codes related to network congestion, insufficient resources, and faults through AT commands. For example, if error codes such as "Congestion", "Insufficient resources for a specific slice and DNN", or "Maximum number of PDU sessions reached" are detected, a network connectivity problem is confirmed. Among these, network connectivity problems can be indicated by the gateway indicator light continuously flashing blue.

[0119] Invalid message problem analysis: Error codes related to protocol incompatibility, semantic unrecognition, and incorrectness are obtained via AT commands. For example, detecting semantically incorrect messages, conditional IE errors, or messages not compatible with protocol states confirms an invalid message problem. Optionally, invalid message problems can be indicated by flashing blue lights on the gateway indicator.

[0120] Using the above implementation process, AT commands can be used to sequentially analyze the faults of the Subscriber Identity Module (SIM) card and the module. Furthermore, AT commands can also be used to collect error codes. Based on the collected error codes, the cause of network faults can be analyzed and determined. According to the different types of network faults identified, the gateway indicator lights can be used to provide corresponding flashing patterns and / or colors to promptly display alarms.

[0121] In this embodiment of the invention, optionally, in step S120, fault analysis is performed based on the operational interaction data to obtain fault analysis data, including:

[0122] The first fault analysis model is used to perform fault analysis on the operational interaction data to obtain fault analysis data.

[0123] The method further includes:

[0124] Obtain the first model update information sent by the network platform;

[0125] The first fault analysis model is updated based on the first model update information.

[0126] In this implementation method, the network platform matches the detected abnormal data with the abnormal sequence template. Based on the verification results, the accuracy of the fault analysis model used by the terminal device and the network platform for fault analysis is judged. The fault analysis model is updated in real time and sent to the terminal device to ensure the accuracy of fault detection by the fault analysis model used by the terminal device.

[0127] It should be noted that when a terminal device detects an initial network access failure or a dropped connection, the terminal device can use the above methods to determine the network fault.

[0128] In one embodiment of the fault diagnosis method described in this invention, the method may optionally further include:

[0129] Under normal network connectivity conditions, predictive analysis is performed based on the operational interaction data of the terminal devices to obtain a prediction result on whether a network failure will occur within a first preset time period.

[0130] When the prediction indicates that a network failure will occur, an alarm message is reported to the network platform. This implementation method, by predicting network failures in advance when the network connection is normal and promptly reporting alarm messages to the network platform, enables the network platform to handle network issues in a timely manner, thereby preventing network failures from occurring.

[0131] In one implementation method, optionally, predictive analysis is performed based on the operational interaction data of the terminal device to obtain a prediction result of whether a network failure will occur within a first preset time period, including:

[0132] The first fault prediction model is used to predict and analyze the operation interaction data to obtain a prediction result of whether a network fault will occur within a first preset time period.

[0133] The method further includes:

[0134] Obtain the second model update information sent by the network platform;

[0135] The first fault prediction model is updated based on the second model update information.

[0136] Using this implementation method, the network platform can construct fault prediction models for both terminal devices and the network platform, and update these models periodically to ensure the accuracy of fault prediction.

[0137] Optionally, the method further includes:

[0138] Save the runtime interaction data and / or the fault analysis data.

[0139] When the amount of data to be saved is large, the operational interaction data and / or the fault analysis data of 30 seconds to 1 minute before the network failure can be saved, and the saved data can be reported to the network platform after the network is restored to normal.

[0140] Another embodiment of the present invention also provides a fault diagnosis method, executed by a network platform, such as... Figure 2 As shown, the method includes:

[0141] S210, Obtain fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs;

[0142] S220, Based on the fault analysis data, establish an abnormal sequence template for the terminal device;

[0143] S230, when a network connection failure of the terminal device is detected, abnormal status detection is performed on the real-time network data to obtain abnormal data;

[0144] S240, Match the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure.

[0145] Using the fault diagnosis method described in this embodiment, the network platform can establish an abnormal sequence template based on the fault analysis data reported by the terminal device. After detecting network faults in the terminal, the detected abnormal data can be matched with the abnormal sequence template to determine the cause of the network fault. This achieves the effect of collaborative analysis and determination of network faults between the terminal device and the network platform, ensuring that network faults can be effectively resolved and reducing manpower and maintenance costs.

[0146] In one embodiment, the fault analysis data may optionally include the correspondence between the fault diagnosis results based on the operational interaction data and the operational interaction data.

[0147] Optionally, in one implementation, the fault analysis data includes a correspondence between fault diagnosis results and operational interaction data. This correspondence is used to clearly indicate the relationship between the fault type of the terminal device and the operational interaction data. Based on this correspondence, the network platform can directly establish an anomaly sequence template for the terminal device. The established anomaly sequence template includes multiple fault types and the operational interaction data corresponding to each fault type. Thus, using the established anomaly sequence template, the network platform performs anomaly detection on real-time network data using an anomaly detection algorithm. By matching the detected anomaly data with the operational interaction data corresponding to each fault type in the anomaly sequence template, the corresponding fault type is determined. This allows the network platform to combine the fault analysis data of the terminal device for fault detection, thereby reducing false positives and ensuring detection accuracy. Specifically, the network platform analyzes real-time network data, performs anomaly data detection, and matches the anomaly data obtained from the detection results with the anomaly sequence template. Based on the matching results, the fault type of the network fault is obtained, and the cause of the network anomaly is output.

[0148] Optionally, in step S220, the abnormal sequence template of the terminal device is established based on the fault analysis data. The established abnormal sequence template includes multiple fault types and corresponding operation interaction data for each fault type.

[0149] Optionally, the real-time network data includes SIM card data of the terminal device obtained by the SIM card management platform, and / or network status data collected by the network platform for a second preset time period before a network failure occurs. When a network failure of the terminal device is detected, the network platform uses the real-time network data, including the SIM card data and / or network status data of the terminal device, to perform abnormal state detection and obtain abnormal data.

[0150] When a terminal device experiences a network failure due to a network outage, the network platform cannot receive information from the terminal device. Therefore, it analyzes the network status data within the second preset time period before the network failure occurs, including the status of the SIM card, network operation status, performance, network quality, traffic, and terminal device, to identify abnormal data.

[0151] Specifically, abnormal state detection is performed on real-time network data, including one or more of the following:

[0152] The SIM card data is used to detect abnormal states; optionally, the SIM card data is used to detect card balance, card status, card power on / off status, and SIM card separation status, etc.; wherein the SIM card data can be obtained through a SIM card management platform.

[0153] Anomaly detection is performed on the network quality indicator data in the network status data; optionally, the network quality indicator data includes RSRP, RSRQ, SINR, and RSSI, etc.

[0154] Anomaly detection is performed on the traffic indicator data in the network status data; optionally, the traffic indicator data includes uplink traffic, downlink traffic, video service traffic, and network service traffic, etc.

[0155] Anomaly detection is performed on the device indicator data in the network status data; optionally, the device indicator data includes CPU utilization, Flash utilization, memory utilization, and temperature, etc.

[0156] Optionally, in addition to the SIM card data mentioned above, network quality index data, traffic index data, and device index data are collected and obtained by the network management platform based on historical data.

[0157] When a terminal device experiences a network outage, the network platform obtains SIM card data by connecting to the SIM card management platform to determine if there is any abnormal information on the SIM card. If so, an alarm is triggered. If the SIM card is in normal condition, anomaly detection is performed based on historical data before the network outage to determine if there are any abnormalities in network quality, traffic, or the device. In addition to identifying abnormal situations such as network outages, this method can also be used for routine fault monitoring.

[0158] In one embodiment of the present invention, an anomaly detection algorithm based on Key Performance Indicator (KPI) of an independent forest can be used to detect abnormal states in real-time network data and obtain abnormal data.

[0159] Optionally, in step S230, when a network fault is detected in the terminal device, abnormal state detection is performed on the real-time network data to obtain abnormal data, including:

[0160] The real-time network data is preprocessed to obtain processed data of the real-time network data;

[0161] Feature extraction is performed on the processed data to obtain the feature matrix of the processed data;

[0162] Multiple detection sample data of the feature matrix are obtained by constructing a binary tree;

[0163] For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal data.

[0164] One implementation method involves preprocessing the real-time network data to obtain processed data of the real-time network data, including:

[0165] For the time-series data (such as traffic data and / or RSRP data) in the real-time network data, based on the time information of multiple real-time network data, the multiple real-time network data are sorted with a third preset duration as the period to obtain the time sequence of multiple real-time network data;

[0166] Optionally, a third preset duration is set to N. Based on the time information of multiple real-time network data, the multiple real-time network data are sorted with a period of N to obtain a time sequence of multiple real-time network data, such as {x1, x2, ... x...}. n};

[0167] Each time-series data in the time-series is standardized to obtain the processed data of the real-time network data.

[0168] Optionally, the time series data in the time series {x1, x2, ... xn} can be standardized in the following way:

[0169] ;

[0170] in, This is one of the time series data in the time series sequence. The mean of the time series data; Let Variance be the variance.

[0171] In another embodiment, the real-time network data is preprocessed to obtain processed data of the real-time network data, including:

[0172] The discrete data (such as card status data and / or registration status data) in the real-time network data are processed by one-hot encoding to obtain the processed data of the real-time network data.

[0173] Optionally, for the time-series data in the real-time network data, feature extraction is performed on the processed data to obtain a feature matrix of the processed data, including:

[0174] The processed data are calculated using different mean calculation methods to obtain the predicted value corresponding to each mean calculation method;

[0175] A feature matrix of the processed data is generated based on the difference between each processed data and a different predicted value.

[0176] In one implementation method, different mean calculation methods are used to calculate the processed data to obtain a predicted value corresponding to each mean calculation method, including:

[0177] The processed data of the time series {x1, x2, ... xn} are analyzed using algorithms such as the difference algorithm, moving average algorithm, weighted moving average algorithm, exponential moving average (ewma), Holt-Winters algorithm, and Autoregression Integrity Moving Average (ARIMA) model to obtain the predicted value of the processed data of the time series for each mean algorithm, such as p;

[0178] Furthermore, based on the difference between each processed data point and different predicted values... Generate the feature matrix of the processed data for the time series {x1, x2,...xn}.

[0179] Optionally, for discrete data in the real-time network data, feature extraction is performed on the processed data to obtain a feature matrix of the processed data, including:

[0180] The processed data is encoded with weight of evidence (WOE) and the information value (IV) is calculated to obtain the feature matrix of the processed data.

[0181] Using the above implementation method, after obtaining the feature matrix of real-time network data, the feature matrix can be analyzed in the following way to determine whether the data corresponding to the corresponding features of the feature matrix is ​​abnormal data:

[0182] Multiple detection sample data of the feature matrix are obtained by constructing a binary tree;

[0183] For each detected sample data, an anomaly detection is performed to determine whether the corresponding detected sample data is abnormal. For example, multiple detected sample data of the feature matrix are obtained using a binary tree construction method, including:

[0184] Features are extracted from the feature matrix to construct multiple binary trees; the number of binary trees constructed depends on the number of features.

[0185] When constructing the binary tree, a feature from the feature matrix can be randomly selected as the starting node. Then, a value between the maximum and minimum values ​​of that feature is randomly chosen. Data in the corresponding detection sample data that is less than this value is assigned to the left branch, and data that is greater than or equal to this value is assigned to the right branch. This process is repeated in both branches until the following condition is met:

[0186] The data is indivisible, meaning it contains only one piece of data, or all data are identical; and / or

[0187] The binary tree reaches its maximum depth.

[0188] Based on the multiple detection sample data obtained by constructing a binary tree using the above method, anomaly detection is performed on each detection sample data to determine whether the corresponding detection sample data is abnormal, including:

[0189] Calculate the path length of the detected sample data in the constructed binary tree;

[0190] Calculate the anomaly score of the detected sample data based on the path length;

[0191] When the abnormal score is greater than a preset judgment value, the corresponding detection sample data is determined to be abnormal data.

[0192] Alternatively, the outlier score for each test sample can be calculated in the following manner:

[0193] Calculate the path length h(xi) of the detected sample data xi corresponding to the feature in the binary tree iTree: h(xi) = e + C(T.size);

[0194] Where e is the number of edges traversed by x from the root node to a leaf node in iTree, and C(T.size) can be considered a correction value, representing the average path length in a binary tree constructed with T.size sample data. Where C(n) = 2H(n-1) - 2(n-1) / n; H(n-1) = ln(n-1) + 0.5772156649; and H(n-1) is the harmonic function used in outlier detection.

[0195] The anomaly score Score(xi) for the detected sample data xi is calculated by combining the results from multiple trees:

[0196] ;

[0197] The corresponding data xi is determined to be an outlier if the outlier score meets the following conditions:

[0198] Score(xi) > θ;

[0199] Where θ is a preset judgment value.

[0200] In this embodiment of the invention, optionally, matching the abnormal data with the abnormal sequence template to determine the fault type of the network fault includes:

[0201] Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template;

[0202] Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix;

[0203] Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters;

[0204] Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template;

[0205] If the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network fault corresponding to the abnormal data is determined according to the matching abnormal sequence template.

[0206] Specifically, determining whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost includes:

[0207] Determine whether the cumulative cost is less than a preset value;

[0208] If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

[0209] For example, let the abnormal sequence template be represented as Q{q1, q2,...qn}, and the abnormal data corresponding to the feature parameters be represented as C{c1, c2,...cm}. Construct a many-to-many distance mapping matrix. The size is n*m.

[0210] ;

[0211] Calculate the distance mapping matrix using dynamic programming algorithm. The minimum-cost path with feature parameters from (0,0) to (n,m) that satisfies monotonicity and continuity. This is a template for one of the abnormal sequences; This is one of the abnormal data;

[0212] Calculate the cumulative cost based on the path. :

[0213] ;

[0214] Specifically, if the cumulative cost is less than a preset value α, i.e., l < α, then the abnormal data is determined to match the corresponding abnormal sequence template, and the fault type of the network fault corresponding to the abnormal data is determined according to the matched abnormal sequence template.

[0215] Since the abnormal sequence template can record the abnormal data detected by the terminal device and the corresponding fault type, the fault type caused by the abnormal data can be determined by matching the abnormal data detected by the platform device with the abnormal sequence template.

[0216] Using the fault diagnosis method described in this embodiment, the terminal device can perform fault analysis after a network fault occurs to obtain fault analysis results, and report the fault analysis data to the network platform after the network returns to normal. This allows the network platform to establish an anomaly sequence template based on the actual disconnection anomaly data and fault results detected by the terminal device. After anomaly detection is performed on the network platform side using an anomaly detection algorithm, the detected anomaly data is matched with the anomaly sequence template to determine the fault type, thereby reducing misjudgments and ensuring detection accuracy.

[0217] Optionally, in this embodiment of the invention, the method further includes:

[0218] After receiving fault analysis data sent by the terminal device, the network platform constructs a first fault prediction model and a second fault prediction model. The first fault prediction model is sent to the terminal device, enabling the terminal device to use the first fault prediction model to perform predictive analysis on the operational interaction data and obtain a prediction result on whether a network fault will occur within a first preset time period. The network platform can also use the second fault prediction model to predict faults in advance using real-time network data when the terminal device's network connection is normal, thereby avoiding the occurrence of network faults.

[0219] Furthermore, the method also includes:

[0220] Based on the fault analysis data and / or the abnormal data obtained by detecting abnormal states in real-time network data, model update information is generated.

[0221] The model update information is sent to the terminal device; the model update information includes first model update information and / or second model update information; the first model update information is used to instruct the terminal device to update the first fault analysis model used for fault analysis, and the second model update information is used to instruct the first fault prediction model used for fault prediction to be updated.

[0222] Optionally, the model update information may further include third model update information for fault analysis updates by the network platform and / or fourth model update information for fault prediction updates by the network platform, used to update the network platform's model for fault analysis and the model for fault prediction, respectively.

[0223] By adopting this implementation method, the network platform can update the first fault analysis model and / or the first fault prediction model for fault analysis and prediction of terminal devices in real time to ensure the accuracy of fault analysis and fault prediction of the terminal.

[0224] The fault diagnosis method described in this invention involves data collection and fault detection by the terminal device and the network platform, respectively, to achieve end-edge collaborative analysis to obtain the cause of network faults, thereby effectively resolving network faults and reducing labor and maintenance costs.

[0225] like Figure 3 This is a schematic diagram of the overall process of the fault diagnosis method described in this embodiment of the invention. In the event of a communication failure between the terminal device and the network platform, i.e., a network failure in the terminal device, the terminal device performs the following implementation steps:

[0226] S3001, collects and processes interactive data;

[0227] S3002, The first fault analysis model is used to perform fault analysis on the operation interaction data;

[0228] S3003, based on the fault type obtained from the fault analysis, flashing prompts are made in a manner corresponding to the fault type;

[0229] S3004 reports fault analysis data to the network platform after the network is restored to normal.

[0230] While the terminal device performs the above implementation steps, the network platform performs the following implementation steps:

[0231] S3011, SIM card data obtained from the SIM card management platform, performs abnormal status detection on the SIM card data;

[0232] S3012 detects network quality, traffic indicators, and device indicators before the disconnection based on network status data before the disconnection;

[0233] S3013, Match the detection results with the abnormal sequence template to determine the fault type of the network fault;

[0234] S3014, outputs the detection results of the network platform;

[0235] S3015, update the anomaly detection model of the terminal device and network platform, and send the first model update information to the terminal device to update the first fault analysis model of the terminal device;

[0236] S3016, construct an anomaly prediction database, update the anomaly prediction models of terminal devices and network platforms, send second model update information to terminal devices, and update the first fault preset model of terminal devices.

[0237] Under normal communication conditions between the terminal device and the network platform, the terminal device shall perform the following implementation steps:

[0238] S3021, The first fault prediction model is used for predictive analysis to obtain the prediction result of whether a network fault will occur within a first preset time period;

[0239] S3022, when a network failure is predicted, an alarm message is reported to the network platform;

[0240] Under normal communication conditions between the terminal device and the network platform, the network platform performs the following implementation steps:

[0241] S3031, The second fault prediction model is used for predictive analysis to obtain the prediction result of whether a network fault will occur within the fourth preset time period;

[0242] S3032, in the event of a predicted network failure, controls the terminal device according to the cause of the failure to prevent the terminal device from going offline.

[0243] It should be noted that the above implementation process can be performed even when the terminal device loses its network connection. When the device fails to connect to the network for the first time, the terminal device can primarily collect data, determine the fault type, and provide indication via indicator lights. Once the terminal device successfully connects to the network, it uploads the collected data and diagnostic results to the network platform for storage and verification of fault analysis.

[0244] Based on the above, when a network failure occurs on a terminal device, the terminal device collects data, performs a preliminary diagnosis, and then reports the collected data to the network platform. The algorithm database can be expanded based on the data reported by the terminal device. The network platform can perform auxiliary analysis based on the data reported by the terminal device and update the fault analysis model and fault prediction model of the terminal device and the network platform based on the operation and maintenance information.

[0245] The fault diagnosis method described in this invention addresses the most pressing issue in current industry networks: abnormal device connection drops. It analyzes the problem from both the terminal and platform sides, helping on-site operators to promptly identify and troubleshoot faults, reducing manpower and time costs for on-site maintenance, and shortening fault repair time.

[0246] This invention also provides a terminal device, such as... Figure 4 As shown, the terminal device 400 includes a processor 410 and a transceiver 420, wherein:

[0247] The processor 410 is used to collect operational interaction data of the terminal device in the event of a network connection failure; and

[0248] Fault analysis is performed based on the operational interaction data to obtain fault analysis data;

[0249] The transceiver 420 is used to report the fault analysis data to the network platform after the network returns to normal.

[0250] Optionally, in the terminal device, the processor 410 is further configured to:

[0251] Under normal network connectivity conditions, predictive analysis is performed based on the operational interaction data of the terminal devices to obtain a prediction result on whether a network failure will occur within a first preset time period.

[0252] The transceiver 420 is also used to report an alarm message to the network platform when the prediction result indicates that a network failure will occur.

[0253] Optionally, in the terminal device, the operational interaction data includes AT commands and / or module interaction logs.

[0254] Optionally, in the terminal device, the processor 410 performs fault analysis based on the operational interaction data to obtain fault analysis data, including:

[0255] Based on the operational interaction data, perform one or more of the following fault analyses to obtain fault analysis data:

[0256] SIM card problem analysis;

[0257] Module problem analysis;

[0258] Network coverage problem analysis;

[0259] Network connectivity problem analysis;

[0260] Authentication issue analysis;

[0261] Analysis of invalid message issues.

[0262] Optionally, in the terminal device, the processor 410 performs fault analysis based on the operational interaction data to obtain fault analysis data, including:

[0263] The first fault analysis model is used to perform fault analysis on the operational interaction data to obtain fault analysis data.

[0264] The processor 410 is further configured to:

[0265] Obtain the first model update information sent by the network platform;

[0266] The first fault analysis model is updated based on the first model update information.

[0267] Optionally, in the terminal device, the processor 410 performs predictive analysis based on the terminal device's operational interaction data to obtain a prediction result regarding whether a network failure will occur within a first preset time period, including:

[0268] The first fault prediction model is used to predict and analyze the operation interaction data to obtain a prediction result of whether a network fault will occur within a first preset time period.

[0269] The processor 410 is also used for:

[0270] Obtain the second model update information sent by the network platform;

[0271] The first fault prediction model is updated based on the second model update information.

[0272] Optionally, in the terminal device, after performing fault analysis based on the operational interaction data to obtain fault analysis data, the processor 410 is further configured to:

[0273] The fault type of the network connection failure is obtained based on the fault analysis data;

[0274] The gateway indicator light controlling the terminal device flashes in a manner corresponding to the fault type to indicate the problem.

[0275] This invention also provides a network platform, such as... Figure 5 As shown, the network platform 500 includes a transceiver 510 and a processor 520, wherein:

[0276] The transceiver 510 is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs;

[0277] The processor 520 is configured to: establish an abnormal sequence template for the terminal device based on the fault analysis data; perform abnormal state detection on real-time network data to obtain abnormal data when a network connection fault of the terminal device is detected; and match the abnormal data with the abnormal sequence template to determine the fault type of the network connection fault.

[0278] Optionally, in the network platform, the real-time network data includes SIM card data of the terminal device obtained by the SIM card management platform, and / or network status data collected by the network platform within a second preset time period before a network connection failure occurs.

[0279] Optionally, in the network platform, the processor 520 performs abnormal state detection on real-time network data, including one or more of the following:

[0280] Perform abnormal status detection on the SIM card data;

[0281] Perform abnormal state detection on the network quality index data in the network status data;

[0282] Perform abnormal state detection on the traffic indicator data in the network status data;

[0283] Anomaly detection is performed on the device indicator data in the network status data.

[0284] Optionally, in the network platform, the processor 520 performs anomaly detection on real-time network data to obtain abnormal data, including:

[0285] The real-time network data is preprocessed to obtain processed data of the real-time network data;

[0286] Feature extraction is performed on the processed data to obtain the feature matrix of the processed data;

[0287] Multiple detection sample data of the feature matrix are obtained by constructing a binary tree;

[0288] For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal data.

[0289] Optionally, in the network platform, the processor 520 performs anomaly detection on each detection sample data, determining whether the corresponding detection sample data is abnormal data, including:

[0290] Calculate the path length of the detected sample data in the constructed binary tree;

[0291] Calculate the anomaly score of the detected sample data based on the path length;

[0292] When the abnormal score is greater than a preset judgment value, the corresponding detection sample data is determined to be abnormal data.

[0293] Optionally, in the network platform, the processor 520 preprocesses the real-time network data to obtain processed data of the real-time network data, including:

[0294] For the time-series data in the real-time network data, based on the time information of multiple real-time network data, the multiple real-time network data are sorted with a third preset duration as the period to obtain the time-series sequence of multiple real-time network data;

[0295] Each time-series data in the time-series is standardized to obtain the processed data of the real-time network data.

[0296] Optionally, in the network platform, the processor 520 preprocesses the real-time network data to obtain processed data of the real-time network data, including:

[0297] One-hot encoding is performed on the discrete data in the real-time network data to obtain the processed data of the real-time network data.

[0298] Optionally, in the network platform, the processor 520 performs feature extraction on the processed data to obtain a feature matrix of the processed data, including:

[0299] The processed data are calculated using different mean calculation methods to obtain the predicted value corresponding to each mean calculation method;

[0300] A feature matrix of the processed data is generated based on the difference between each processed data and a different predicted value.

[0301] Optionally, in the network platform, the processor 520 performs feature extraction on the processed data to obtain a feature matrix of the processed data, including:

[0302] The processed data is subjected to WOE encoding and IV value calculation to obtain the feature matrix of the processed data.

[0303] Optionally, in the network platform, the processor 520 matches the abnormal data with the abnormal sequence template to determine the fault type of the network fault, including:

[0304] Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template;

[0305] Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix;

[0306] Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters;

[0307] Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template;

[0308] If the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network fault corresponding to the abnormal data is determined according to the matching abnormal sequence template.

[0309] Optionally, in the network platform, the processor 520 determines whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost, including:

[0310] Determine whether the cumulative cost is less than a preset value;

[0311] If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

[0312] Optionally, in the network platform, the processor 520 is further configured to:

[0313] Based on the fault analysis data and / or the abnormal data obtained by detecting abnormal states in real-time network data, model update information is generated.

[0314] The transceiver 510 is further configured to send the model update information to the terminal device; the model update information includes first model update information and / or second model update information; the first model update information is used to instruct the terminal device to update the first fault analysis model used for fault analysis, and the second model update information is used to instruct the first fault prediction model used for fault prediction to be updated.

[0315] This invention also provides a fault diagnosis device applied to terminal devices, such as... Figure 6 As shown, the device includes:

[0316] The data acquisition module 610 is used to collect the operational interaction data of the terminal device in the event of a network connection failure.

[0317] Analysis module 620 is used to perform fault analysis based on the operation interaction data to obtain fault analysis data;

[0318] The reporting module 630 is used to report the fault analysis data to the network platform after the network returns to normal.

[0319] Optionally, in the fault diagnosis device, the analysis module 620 is further used for:

[0320] Under normal network connectivity conditions, predictive analysis is performed based on the operational interaction data of the terminal devices to obtain a prediction result on whether a network failure will occur within a first preset time period.

[0321] The reporting module 630 is also used to report alarm information to the network platform when the prediction result indicates that a network failure will occur.

[0322] Optionally, in the fault diagnosis device, the operational interaction data includes AT commands and / or module interaction logs.

[0323] Optionally, in the fault diagnosis device, the analysis module 620 performs fault analysis based on the operational interaction data to obtain fault analysis data, including:

[0324] Based on the operational interaction data, perform one or more of the following fault analyses to obtain fault analysis data:

[0325] SIM card problem analysis;

[0326] Module problem analysis;

[0327] Network coverage problem analysis;

[0328] Network connectivity problem analysis;

[0329] Authentication issue analysis;

[0330] Analysis of invalid message issues.

[0331] Optionally, in the fault diagnosis device, the analysis module 620 performs fault analysis based on the operational interaction data to obtain fault analysis data, including:

[0332] The first fault analysis model is used to perform fault analysis on the operational interaction data to obtain fault analysis data.

[0333] The analysis module 620 is further used for:

[0334] Obtain the first model update information sent by the network platform;

[0335] The first fault analysis model is updated based on the first model update information.

[0336] Optionally, in the fault diagnosis device, the analysis module 620 performs predictive analysis based on the operational interaction data of the terminal device to obtain a prediction result of whether a network fault will occur within a first preset time period, including:

[0337] The first fault prediction model is used to predict and analyze the operation interaction data to obtain a prediction result of whether a network fault will occur within a first preset time period.

[0338] The analysis module 620 is also used for:

[0339] Obtain the second model update information sent by the network platform;

[0340] The first fault prediction model is updated based on the second model update information.

[0341] Optionally, in the fault diagnosis device, after performing fault analysis based on the operational interaction data to obtain fault analysis data, the analysis module 620 is further configured to:

[0342] The fault type of the network connection failure is obtained based on the fault analysis data;

[0343] The gateway indicator light controlling the terminal device flashes in a manner corresponding to the fault type to indicate the problem.

[0344] This invention also provides a fault diagnosis device applied to a network platform, such as... Figure 7 As shown, the device includes:

[0345] The information acquisition module 710 is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operation interaction data when a network connection failure occurs;

[0346] The template creation module 720 is used to create an abnormal sequence template for the terminal device based on the fault analysis data.

[0347] The detection module 730 is used to detect abnormal states in real-time network data and obtain abnormal data when a network connection failure of the terminal device is detected.

[0348] The matching module 740 is used to match the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure.

[0349] Optionally, in the fault diagnosis device, the real-time network data includes SIM card data of the terminal device obtained by the SIM card management platform, and / or network status data collected by the network platform within a second preset time period before the network connection failure occurs.

[0350] Optionally, in the fault diagnosis device, the detection module 730 performs abnormal state detection on real-time network data, including one or more of the following:

[0351] Perform abnormal status detection on the SIM card data;

[0352] Perform abnormal state detection on the network quality index data in the network status data;

[0353] Perform abnormal state detection on the traffic indicator data in the network status data;

[0354] Anomaly detection is performed on the device indicator data in the network status data.

[0355] Optionally, in the fault diagnosis device, the detection module 730 performs abnormal state detection on real-time network data to obtain abnormal data, including:

[0356] The real-time network data is preprocessed to obtain processed data of the real-time network data;

[0357] Feature extraction is performed on the processed data to obtain the feature matrix of the processed data;

[0358] Multiple detection sample data of the feature matrix are obtained by constructing a binary tree;

[0359] For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal data.

[0360] Optionally, in the fault diagnosis device, the detection module 730 performs anomaly judgment on each detection sample data, determining whether the corresponding detection sample data is abnormal data, including:

[0361] Calculate the path length of the detected sample data in the constructed binary tree;

[0362] Calculate the anomaly score of the detected sample data based on the path length;

[0363] When the abnormal score is greater than a preset judgment value, the corresponding detection sample data is determined to be abnormal data.

[0364] Optionally, in the fault diagnosis device, the detection module 730 preprocesses the real-time network data to obtain processed data of the real-time network data, including:

[0365] For the time-series data in the real-time network data, based on the time information of multiple real-time network data, the multiple real-time network data are sorted with a third preset duration as the period to obtain the time-series sequence of multiple real-time network data;

[0366] Each time-series data in the time-series is standardized to obtain the processed data of the real-time network data.

[0367] Optionally, in the fault diagnosis device, the detection module 730 preprocesses the real-time network data to obtain processed data of the real-time network data, including:

[0368] One-hot encoding is performed on the discrete data in the real-time network data to obtain the processed data of the real-time network data.

[0369] Optionally, in the fault diagnosis device, the detection module 730 performs feature extraction on the processed data to obtain a feature matrix of the processed data, including:

[0370] The processed data are calculated using different mean calculation methods to obtain the predicted value corresponding to each mean calculation method;

[0371] A feature matrix of the processed data is generated based on the difference between each processed data and a different predicted value.

[0372] Optionally, in the fault diagnosis device, the detection module 730 performs feature extraction on the processed data to obtain a feature matrix of the processed data, including:

[0373] The processed data is subjected to WOE encoding and IV value calculation to obtain the feature matrix of the processed data.

[0374] Optionally, in the fault diagnosis device, the matching module 740 matches the abnormal data with the abnormal sequence template to determine the fault type of the network fault, including:

[0375] Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template;

[0376] Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix;

[0377] Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters;

[0378] Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template;

[0379] If the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network fault corresponding to the abnormal data is determined according to the matching abnormal sequence template.

[0380] Optionally, in the fault diagnosis device, the matching module 740 determines whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost, including:

[0381] Determine whether the cumulative cost is less than a preset value;

[0382] If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

[0383] Optionally, the fault diagnosis device further includes:

[0384] The model generation module 750 is used to generate model update information based on the fault analysis data and / or the abnormal data obtained by abnormal state detection of real-time network data;

[0385] The sending module 760 is used to send the model update information to the terminal device; the model update information includes first model update information and / or second model update information; the first model update information is used to instruct the terminal device to update the first fault analysis model used for fault analysis, and the second model update information is used to instruct the first fault prediction model used for fault prediction to be updated.

[0386] This invention also provides a terminal device, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the fault diagnosis method as described in any of the preceding embodiments.

[0387] In this embodiment of the invention, the specific implementation process of the fault diagnosis method executed by the processor on the terminal device can be found in the description of the method section, and will not be described in detail here.

[0388] An embodiment of the present invention provides a network platform, comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the fault diagnosis method as described in any of the preceding embodiments.

[0389] In this embodiment of the invention, the specific implementation process of the fault diagnosis method executed by the processor on the network platform can be found in the description of the method section, and will not be described in detail here.

[0390] In addition, specific embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the steps of the fault diagnosis method as described above.

[0391] Specifically, the computer-readable storage medium is applied to the aforementioned terminal device or network platform. When applied to the terminal device or network platform, the execution steps in the corresponding fault diagnosis method are described in detail above and will not be repeated here.

[0392] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0393] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can be physically comprised separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware or in the form of hardware plus software functional units.

[0394] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions that cause a computer device (which may be a personal computer, server, or network device, etc.) to execute some steps of the transmission and reception methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0395] The above describes the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A fault diagnosis method, characterized in that, The method, executed by a network platform, includes: Acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs; Based on the fault analysis data, an abnormal sequence template for the terminal device is established; wherein, the abnormal sequence template is established based on the correspondence between the fault types of the terminal device and the operational interaction data, and the abnormal sequence template includes multiple fault types and the operational interaction data corresponding to each fault type; When a network connection failure of the terminal device is detected, abnormal status detection is performed on the real-time network data to obtain abnormal data. The abnormal data is matched with the abnormal sequence template to determine the fault type of the network connection failure; The process of matching the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure includes: Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template; Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix; Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters; Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template; When the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network connection failure corresponding to the abnormal data is determined according to the matching abnormal sequence template. Specifically, determining whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost includes: Determine whether the cumulative cost is less than a preset value; If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

2. The fault diagnosis method according to claim 1, characterized in that, The real-time network data includes SIM card data of the terminal device obtained by the SIM card management platform, and / or network status data collected by the network platform within a second preset time period before the network connection failure occurs.

3. The fault diagnosis method according to claim 2, characterized in that, Anomaly detection of real-time network data includes one or more of the following: Perform abnormal status detection on the SIM card data; Perform abnormal state detection on the network quality index data in the network status data; Perform abnormal state detection on the traffic indicator data in the network status data; Anomaly detection is performed on the device indicator data in the network status data.

4. The fault diagnosis method according to claim 1, characterized in that, The process of detecting abnormal states in real-time network data and obtaining abnormal data includes: The real-time network data is preprocessed to obtain processed data of the real-time network data; Feature extraction is performed on the processed data to obtain the feature matrix of the processed data; Multiple detection sample data of the feature matrix are obtained by constructing a binary tree; For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal data.

5. The fault diagnosis method according to claim 4, characterized in that, For each test sample data, an anomaly detection is performed to determine whether the corresponding test sample data is abnormal, including: Calculate the path length of the detected sample data in the constructed binary tree; Calculate the anomaly score of the detected sample data based on the path length; When the abnormal score is greater than a preset judgment value, the corresponding detection sample data is determined to be abnormal data.

6. The fault diagnosis method according to claim 4, characterized in that, Feature extraction is performed on the processed data to obtain a feature matrix of the processed data, including: The processed data are calculated using different mean calculation methods to obtain the predicted value corresponding to each mean calculation method; A feature matrix of the processed data is generated based on the difference between each processed data and a different predicted value.

7. The fault diagnosis method according to claim 1, characterized in that, The method further includes: Based on the fault analysis data and / or the abnormal data obtained by detecting abnormal states in real-time network data, model update information is generated. The model update information is sent to the terminal device; the model update information includes first model update information and / or second model update information; the first model update information is used to instruct the terminal device to update the first fault analysis model used for fault analysis, and the second model update information is used to instruct the first fault prediction model used for fault prediction to be updated.

8. A network platform, comprising a transceiver and a processor, characterized in that: The transceiver is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs; The processor is configured to: establish an abnormal sequence template for the terminal device based on the fault analysis data; perform abnormal state detection on real-time network data to obtain abnormal data when a network connection fault of the terminal device is detected; and match the abnormal data with the abnormal sequence template to determine the fault type of the network connection fault. The abnormal sequence template is established based on the correspondence between the fault type of the terminal device and the operation interaction data, and the abnormal sequence template includes multiple fault types and the operation interaction data corresponding to each fault type. The processor matches the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure, including: Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template; Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix; Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters; Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template; When the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network connection failure corresponding to the abnormal data is determined according to the matching abnormal sequence template. The processor determines whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost, including: Determine whether the cumulative cost is less than a preset value; If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

9. A fault diagnosis device, characterized in that, The device, applied to a network platform, includes: The information acquisition module is used to acquire fault analysis data reported by the terminal device to the network platform; wherein, the fault analysis data is obtained by the terminal device after performing fault analysis based on the collected operational interaction data when a network connection failure occurs; The template creation module is used to create an abnormal sequence template for the terminal device based on the fault analysis data; wherein the abnormal sequence template is created based on the correspondence between the fault types of the terminal device and the operation interaction data, and the abnormal sequence template includes multiple fault types and the operation interaction data corresponding to each fault type. The detection module is used to detect abnormal states in real-time network data and obtain abnormal data when a network connection failure of the terminal device is detected. A matching module is used to match the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure. The matching module matches the abnormal data with the abnormal sequence template to determine the fault type of the network connection failure, including: Construct a distance mapping matrix based on the abnormal data and the abnormal sequence template; Based on the dynamic programming algorithm, calculate the minimum cost path among multiple feature parameters of the distance mapping matrix; Based on the path with the minimum cost, calculate the cumulative cost of the corresponding feature parameters; Based on the cumulative cost, determine whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template; When the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template, the fault type of the network connection failure corresponding to the abnormal data is determined according to the matching abnormal sequence template. The matching module determines whether the abnormal data corresponding to the feature parameter matches the corresponding abnormal sequence template based on the cumulative cost, including: Determine whether the cumulative cost is less than a preset value; If the cumulative cost is less than the preset value, then the abnormal data corresponding to the feature parameter is determined to match the corresponding abnormal sequence template.

10. A network platform, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the fault diagnosis method as described in any one of claims 1 to 7.

11. A readable storage medium, characterized in that, The readable storage medium stores a program that, when executed by a processor, implements the steps of the fault diagnosis method as described in any one of claims 1 to 7.

Images