ONU gateway fault intelligent repair method and device based on lightweight feature fusion

Abstract

The invention provides an ONU gateway fault intelligent repairing method and device based on lightweight feature fusion, and the method comprises the steps: obtaining the lightweight feature of a target ONU gateway in a current fusion period, carrying out fusion preprocessing according to a hardware state feature, a Wi-Fi state feature, a DNS state feature, an IoT correlation feature and an input requirement, and generating a feature vector; performing fault detection according to the feature vector and a lightweight neural network model, and generating a fault type label and a fault confidence coefficient; determining a current fault state according to the fault confidence, wherein the current fault state comprises a determined fault, a suspected fault and a fault-free state; and when the current fault state is the determined fault, matching the repair strategy library according to the fault type label, determining the repair feasibility of the current fault, and executing the fault repair program, thereby achieving the effects of remarkably improving the fault classification precision, innovatively repairing the fault local repair and realizing the fault tracing through the bidirectional association of the ONU gateway and the IoT equipment.

Description

Technical Field

[0001] This invention relates to the field of gateway fault detection, and specifically to an intelligent repair method and device for ONU gateway faults based on lightweight feature fusion. Background Technology

[0002] With the intelligent upgrade of home networks, ONU gateways have become the core hub connecting operator networks, home Wi-Fi, and IoT devices. However, current home ONU gateway fault handling faces many problems: it cannot identify complex faults, the fault types are complex and mutually influential, and traditional manual troubleshooting is difficult to find the root cause; it can only achieve simple fault identification and alarm, without utilizing the ONU gateway's own hardware and software capabilities to complete local repair, requiring users to manually report the fault and rely on professional technical personnel for manual repair, while on-site repair by technicians has long response times and high manual maintenance costs; existing technologies use complex models such as federated learning and transfer learning, and due to the limited hardware resources of home ONU gateways, the edge computing performance of ONUs is limited, making it difficult to find a balance between "classification accuracy" and "resource consumption"; moreover, existing technologies only focus on the faults of the ONU itself, without associating them with the abnormal states of IoT devices, easily misjudging "ONU faults" as "IoT device faults", resulting in low positioning accuracy. Summary of the Invention

[0003] In view of the aforementioned problems, this application is proposed to provide a method and apparatus for intelligent fault repair of ONU gateways based on lightweight feature fusion, which overcomes or at least partially solves the aforementioned problems, comprising: A method for intelligent fault repair of ONU gateway based on lightweight feature fusion, wherein the method performs fault detection based on lightweight features of the ONU gateway, wherein the lightweight features include hardware status features, Wi-Fi status features, DNS status features, and IoT association features; the method includes: Obtain lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors; Fault detection is performed based on the feature vector and a preset lightweight neural network model, generating fault type labels and fault confidence scores. The current fault status is determined based on the fault confidence level, and the current fault status includes confirmed fault, suspected fault, and no fault. When the current fault status is a confirmed fault, the feasibility of repairing the current fault is determined by matching the fault type label with the preset repair strategy library and the fault repair procedure is executed.

[0004] Specifically, the step of determining the current fault state based on the fault confidence level, wherein the current fault state includes the steps of confirmed fault, suspected fault, and no fault, includes: Based on the feature vector, the lightweight neural network model is used for inference and analysis to locate the current root cause of the fault as an ONU gateway fault, and the fault type label and the fault confidence level are generated. When the fault confidence level is greater than the first threshold of the fault determination threshold, the fault status is determined to be a confirmed fault. When the fault confidence level is between the first threshold and the second threshold of the fault determination threshold, the fault state is determined to be a suspected fault. If the fault confidence level is less than the second threshold of the fault determination threshold, then the fault state is determined to be fault-free.

[0005] Specifically, the step of determining the current fault state based on the fault confidence level, wherein the current fault state includes the steps of confirmed fault, suspected fault, and no fault, includes: When the fault status is confirmed fault, the subsequent fault repair program is started. Based on the fault type label, the preset repair strategy library is matched to determine the repair feasibility of the current fault and the corresponding processing program is automatically executed. When the fault status is a suspected fault, the feature collection frequency is doubled and data is collected and analyzed again. If the confidence level of the fault is greater than the first threshold for three consecutive times, it is determined to be a confirmed fault; otherwise, the suspected status is lifted. When the fault status is fault-free, the normal feature acquisition frequency is maintained.

[0006] Specifically, the step of performing fault detection based on the feature vector and a preset lightweight neural network model to generate fault type labels and fault confidence scores further includes: Based on the abnormal status signals uploaded by IoT devices and the pre-set association rule library, extract at least one candidate root cause fault that meets the conditions; Based on the candidate root cause faults, lightweight features of the ONU gateway are collected in real time, and corresponding feature data associated with the candidate root cause faults are extracted. The feature vector is generated based on the corresponding feature data and input into the lightweight neural network model for inference analysis to generate fault inference results and fault confidence. When the fault reasoning result is consistent with the candidate root cause fault and the fault confidence is greater than or equal to the first threshold, the current fault root cause is determined to be an ONU gateway fault and the fault status is a confirmed fault; otherwise, the current fault root cause is determined to be an IoT device fault.

[0007] Specifically, the step of obtaining lightweight features of the target ONU gateway within the current fusion cycle, performing fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors includes: The hardware status characteristics, Wi-Fi status characteristics, DNS status characteristics, and IoT association characteristics within the fusion cycle are aligned. Calculate the statistical values ​​of the hardware status features, Wi-Fi status features, DNS status features, and IoT association features during the fusion period to generate fusion feature data; The fused feature data is subjected to principal component analysis for dimensionality reduction, Min-Max standardization, and numerical vector transformation to generate the feature vector with fixed dimensions.

[0008] Specifically, when the current fault status is a confirmed fault, the steps of matching the fault type label with a preset repair strategy library to determine the repair feasibility of the current fault and automatically executing the corresponding processing procedure include: Based on the fault type label, the preset repair strategy library is matched to determine the feasibility of repairing the current fault; When the repair feasibility is determined to be an unrepairable fault, the unrepairable fault alarm process is automatically executed; When the repair feasibility is determined to be a repairable fault, the repair strategy corresponding to the fault type tag is extracted from the repair strategy library, the repair action is executed according to the step sequence of the repair strategy, and the execution log is recorded. After the repair action is completed, the repair effect verification process is immediately executed based on the feature vector and the preset verification rules.

[0009] Specifically, after the repair action is completed, the repair effect verification process is immediately executed based on the feature vector and preset verification rules, including: During the preset repair monitoring period, the feature vector in the current ONU gateway is collected in real time; Extract the corresponding feature data related to the fault type label from the feature vector; If the corresponding feature data remains below a preset normal threshold during the repair monitoring period, the repair is deemed successful; otherwise, the repair is deemed unsuccessful and an unrepairable fault alarm procedure is triggered.

[0010] A lightweight feature fusion-based intelligent fault repair device for ONU gateways includes: Feature acquisition and preprocessing module: Acquires lightweight features of the target ONU gateway within the current fusion cycle, performs fusion preprocessing based on the lightweight features and preset input requirements, and generates feature vectors; Fault detection module: performs fault detection based on the feature vector and a preset lightweight neural network model, and generates fault type labels and fault confidence scores; Fault Status Determination Module: Determines the current fault status based on the fault confidence level, whereby the current fault status includes confirmed fault, suspected fault, and no fault; Fault Processing Module: Triggered by the fault state, the system matches the fault type label with a preset repair strategy library to determine the feasibility of repairing the current fault and automatically executes the corresponding fault handling procedure.

[0011] A computer electronic device includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, it implements the steps of the ONU gateway fault intelligent repair method based on lightweight feature fusion as described above.

[0012] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the ONU gateway fault intelligent repair method based on lightweight feature fusion as described above.

[0013] This application has the following advantages: In the embodiments of this application, addressing the problems of difficulty in identifying and tracing complex faults, lack of local repair capabilities and automatic repair reporting functions, complexity of existing processing models making them difficult to adapt to home ONU gateways, and lack of fault detection and location for abnormal states of associated IoT devices in the prior art, this application provides a solution for deploying a lightweight, improved lightweight neural network model on the ONU gateway to intelligently analyze four types of multi-dimensional lightweight features (hardware, Wi-Fi, DNS, IoT) collected and fused in real time, thereby achieving automated fault diagnosis and repair. Specifically, it involves obtaining the lightweight features of the target ONU gateway within the current fusion cycle. The system performs a fusion preprocessing process based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate a feature vector. It then performs fault detection based on the feature vector and a preset lightweight neural network model, generating fault type labels and fault confidence scores. Based on the fault confidence scores, it determines the current fault state, which includes confirmed fault, suspected fault, and no fault. When the current fault state is confirmed fault, it matches the fault type label against a preset repair strategy library to determine the feasibility of repairing the current fault and executes the fault repair procedure. By collecting four types of lightweight features from the ONU gateway and combining them with a lightweight improved neural network model, this technology solves the problems of existing technologies where single features cannot identify complex faults and complex models are not suitable for edge computing power. The lightweight design of the model is fully adapted to the limited hardware resources of home ONU gateways. After training and continuous upgrades via OTA, it achieves ultra-high accuracy in fault diagnosis, classification, and location. Through a preset repair strategy library and a secondary detection closed-loop procedure, it realizes local automatic repair of ONU gateway faults, which not only reduces the waiting time for users to report repairs but also significantly reduces the manual operation and maintenance costs for operators. By associating the ONU gateway with the status of IoT devices and bidirectional verification, it enables reverse location of the ONU gateway fault root cause based on the status signals of IoT devices. This solves the problem of fault misjudgment caused by the lack of association with the status of IoT devices in existing technologies, avoids users blindly operating IoT devices due to misjudgment, and helps operation and maintenance personnel quickly locate the root cause of the fault, improving troubleshooting efficiency. It achieves the effects of significantly improved fault classification accuracy, innovative local fault repair, and fault tracing through bidirectional association between the ONU gateway and IoT devices. Attached Figure Description

[0014] To more clearly illustrate the technical solution of this application, the drawings used in the description of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1This is a flowchart illustrating the steps of an ONU gateway fault intelligent repair method based on lightweight feature fusion, provided in one embodiment of this application. Figure 2 This is a structural block diagram of an ONU gateway fault intelligent repair device based on lightweight feature fusion, provided in one embodiment of this application; Figure 3 This is a flowchart of a fault detection and repair method for an ONU gateway based on lightweight feature fusion, provided in one embodiment of this application. Figure 4 This is a flowchart illustrating the training, deployment, and iterative upgrade process of the lightweight neural network model provided in one embodiment of this application. Figure 5 This is a flowchart illustrating the reverse localization process of an ONU gateway fault intelligent repair method based on lightweight feature fusion, provided in one embodiment of this application. Figure 6 This is a schematic diagram of the structure of a computer electronic device provided in an embodiment of the present invention; 1. Computer electronic device; 2. External device; 3. Processing unit; 4. Bus; 5. Network adapter; 6. I / O interface; 7. Display; 8. Memory; 9. Random access memory; 10. Cache memory; 11. Storage system; 12. Program / utility; 13. Program module. Detailed Implementation

[0016] To make the objectives, features, and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0017] The inventors, through analysis of existing technologies, discovered that: due to limited hardware resources in home ONU gateways, current home ONU gateways cannot identify complex faults. The complexity of fault types and their mutual influence make it difficult to find the root cause through traditional manual troubleshooting; they can only achieve simple fault identification and alarms, without utilizing the ONU gateway's own hardware and software capabilities to complete local repairs. Users need to manually report repairs and rely on professional technical intervention for repairs, which results in long response times and high manual maintenance costs for on-site repairs; existing technologies all use complex learning models, and the limited edge computing power of the ONU makes it difficult to meet the demands, leading to functional limitations and even making the ONU itself prone to failure; existing technologies only focus on ONU faults themselves, without considering the abnormal states of IoT devices, which can easily lead to misjudging "ONU faults" as "IoT device faults," resulting in low positioning accuracy.

[0018] It should be noted that, in any embodiment of the present invention, the method is based on a lightweight improved lightweight neural network model and lightweight features of the ONU gateway for fault detection and location, wherein the lightweight features include hardware status features, Wi-Fi status features, DNS status features and IoT association features.

[0019] Reference Figure 1 This application illustrates an embodiment of an intelligent fault repair method for ONU gateways based on lightweight feature fusion. S110. Obtain the lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features and preset input requirements to generate feature vectors; S120. Based on the feature vector and the preset lightweight neural network model, perform fault detection and generate fault type labels and fault confidence scores. S130. Determine the current fault status based on the fault confidence level, wherein the current fault status includes confirmed fault, suspected fault, and no fault; S140. When the current fault status is a confirmed fault, the feasibility of repairing the current fault is determined by matching the fault type label with the preset repair strategy library and the fault repair procedure is executed. In the embodiments of this application, by collecting four types of lightweight features from the ONU gateway and combining them with a lightweight improved lightweight neural network model, the problems of existing technologies, such as the inability of single features to identify complex faults and the incompatibility of complex models with edge computing power, are solved. The lightweight design of the model is fully adapted to the limited hardware resources of home ONU gateways, and after training and continuous upgrades via OTA, it achieves ultra-high accuracy in fault diagnosis, classification, and location. Through a preset repair strategy library and a secondary detection closed-loop procedure, local automatic repair of ONU gateway faults is achieved, which not only reduces the waiting time for users to report repairs, but also significantly reduces the manual operation and maintenance costs of operators. By associating the state of the ONU gateway with that of IoT devices and bidirectional verification, the root cause of ONU gateway faults can be located in reverse based on the state signals of IoT devices. This solves the problem of fault misjudgment caused by the lack of association with the state of IoT devices in existing technologies, avoids users blindly operating IoT devices due to misjudgment, and helps operation and maintenance personnel quickly locate the root cause of the fault, improving troubleshooting efficiency. It achieves the effects of significantly improved fault classification accuracy, innovative local fault repair, and fault tracing through bidirectional association between the ONU gateway and IoT devices.

[0020] The following will further explain an ONU gateway fault intelligent repair method based on lightweight feature fusion in this exemplary embodiment.

[0021] In one embodiment of the present invention, the specific process of step S110, which involves "obtaining the lightweight features of the target ONU gateway in the current fusion cycle, performing fusion preprocessing based on the hardware status features, the Wi-Fi status features, the DNS status features, the IoT association features, and preset input requirements, and generating a feature vector", can be further explained in conjunction with the following description.

[0022] As described in the following steps The hardware status characteristics, Wi-Fi status characteristics, DNS status characteristics, and IoT association characteristics within the fusion cycle are aligned. Calculate the statistical values ​​of the hardware status features, Wi-Fi status features, DNS status features, and IoT association features during the fusion period to generate fusion feature data; The fused feature data is subjected to principal component analysis for dimensionality reduction, Min-Max standardization, and numerical vector transformation to generate the feature vector with fixed dimensions.

[0023] It should be noted that the hardware status characteristics include the core chip temperature, the optical power received by the optical module, the power supply voltage fluctuation value, and the physical connection status of the ports; The Wi-Fi status characteristics include channel utilization, packet loss rate, latency jitter, number of connected terminals, and channel status information amplitude characteristics. The DNS state characteristics include DNS resolution success rate, average resolution latency, and types of domain names that failed to resolve. The IoT-related features include IoT device online rate, communication latency between ONU and IoT device, and device data upload failure rate.

[0024] Principal component analysis was performed to reduce the dimensionality of the amplitude characteristics of the channel state information. The core chip temperature, the optical module received optical power, the power supply voltage fluctuation value, the channel utilization rate, the packet loss rate, the latency jitter, the number of connected terminals, the DNS resolution success rate, the average resolution latency, the IoT device online rate, the communication latency, and the device data upload failure rate are all processed using Min-Max normalization. The domain name types that failed to resolve are converted into numerical vectors using one-hot encoding; In one specific implementation, the edge processing chip of the ONU gateway collects lightweight features of the ONU gateway within a fusion cycle of 2 minutes. After fusion preprocessing, the lightweight features are used to generate the feature vector. A single feature vector occupies less than 128 bytes, which is compatible with the memory of the edge processing chip of the home ONU gateway.

[0025] As an example, the core chip temperature can be monitored by reading data from the chip's built-in temperature sensor or by an external temperature sensor. Both methods can meet the requirements for fault diagnosis. The packet loss rate can be calculated not only based on ICMP ping packets but also by statistically analyzing the loss ratio of TCP protocol packets, which can also accurately reflect the packet loss situation of the Wi-Fi link. The types of domain name resolution failures can be classified not only by domain name suffix but also by application type (such as video applications and office applications), without affecting the judgment of DNS failure scenarios. The device communication latency can be calculated not only by TCP handshake latency but also by UDP (User Datagram Protocol) round-trip time. Both methods can reflect the communication latency between the ONU and the IoT device.

[0026] In one embodiment of the present invention, the specific process of "determining the current fault state based on the fault confidence level, wherein the current fault state includes confirmed fault, suspected fault and no fault" in step S130 can be further described in conjunction with the following description.

[0027] As described in the following steps Based on the feature vector, the lightweight neural network model is used for inference and analysis to locate the current root cause of the fault as an ONU gateway fault, and the fault type label and the fault confidence level are generated. When the fault confidence level is greater than the first threshold of the fault determination threshold, the fault status is determined to be a confirmed fault. When the fault confidence level is between the first threshold and the second threshold of the fault determination threshold, the fault state is determined to be a suspected fault. If the fault confidence level is less than the second threshold of the fault determination threshold, then the fault state is determined to be fault-free. It should be noted that this specifically refers to the situation where the root cause of the fault is determined to be the ONU gateway itself after inference and analysis based on the feature vector of the ONU gateway through the lightweight neural network model, without involving IoT devices or external networks.

[0028] In one specific implementation, the lightweight neural network model performs inference analysis on the feature vectors within a certain fusion cycle to determine the occurrence of a fault, identifies the current fault as an ONU gateway fault, and outputs the fault type and fault confidence level. The fault confidence level is between the first and second thresholds of the fault judgment threshold. At this time, the fault state is judged as a suspected fault. The feature collection frequency is doubled, and data is collected again and inference analysis is performed in the next three collection cycles. If all three fault confidence levels are greater than the first threshold, the fault state is judged as a confirmed fault.

[0029] As an example, for the training and deployment of the lightweight neural network, at least 100,000 home ONU gateway fault samples are collected. Each sample contains the lightweight features and the fault type label, which is double-labeled by operation and maintenance experts according to "fault type - feature threshold". The dataset is divided into training set, validation set and test set in an 8:1:1 ratio. Data augmentation is performed by adding random noise, feature value shifting, etc., to improve the model's generalization ability and build a lightweight improved lightweight neural network model. Training is carried out on a cloud server, and the accuracy of the validation set is monitored in real time. Training is stopped when the accuracy of the validation set does not improve for 3 consecutive rounds. The training objective is that the accuracy of the validation set is not less than 95%, the inference time of a single sample is less than 0.05ms, and the model size is less than 3.2MB. If the accuracy of the validation set is ≥95%, then TensorFlow is used. The Lite tool quantizes the model (INT8) and converts it to TFLite format, removing training-related layers such as dropout layers, retaining only the inference computation graph; if the requirements are not met, it returns to adjust the model parameters and retrains; the trained lightweight neural network model is deployed on the ONU gateway and upgraded and iterated remotely via OTA (Over-the-Air Technology) from the operator, completing the lightweight improvement, training, and deployment of the lightweight neural network model, fully adapting to the limited hardware resources of the home ONU gateway.

[0030] In one embodiment of the present invention, the specific features of the "intelligent repair method for ONU gateway faults based on lightweight feature fusion" can be further described in conjunction with the following description.

[0031] When the fault status is a suspected fault, the feature collection frequency is doubled and data is collected and analyzed again. If the confidence level of the fault is greater than the first threshold for three consecutive times, it is determined to be a confirmed fault; otherwise, the suspected status is lifted. When the fault status is fault-free, the normal feature acquisition frequency is maintained.

[0032] As an example, the first threshold of the fault judgment threshold can be set to 90%, and the second threshold can be set to 60%; the "feature acquisition frequency doubling" can be changed from once every 10 seconds to once every 5 seconds.

[0033] In one embodiment of the present invention, the specific features of the "intelligent repair method for ONU gateway faults based on lightweight feature fusion" can be further described in conjunction with the following description.

[0034] Based on the abnormal status signals uploaded by IoT devices and the pre-set association rule library, extract at least one candidate root cause fault and fault confidence that meets the conditions; Based on the candidate root cause faults and the fault confidence, lightweight features of the ONU gateway are collected in real time, and corresponding feature data associated with the candidate root cause faults are extracted. The feature vector is generated based on the corresponding feature data and input into the lightweight neural network model to generate fault reasoning results; When the fault reasoning result is consistent with the candidate root cause fault and the fault confidence is greater than or equal to the first threshold, the current fault root cause is determined to be an ONU gateway fault and the fault status is a confirmed fault; otherwise, the current fault root cause is determined to be an IoT device fault.

[0035] It should be noted that this specifically refers to the situation where the fault is triggered by an abnormal state signal of an IoT device, and the root cause of the fault is determined to be the ONU gateway itself after comparing and verifying the feature vector of the ONU gateway and the abnormal state signal of the IoT device through the lightweight neural network model.

[0036] In one specific implementation, the ONU gateway receives an abnormal status signal uploaded by an IoT device (e.g., an abnormal status signal such as "video playback stuttering" or "slow response triggering" reported by a smart camera). By querying the association rule base, it matches a candidate root cause fault of "Wi-Fi channel congestion" with a confidence level of 92%. It then generates a fault inference result of the feature vector in the current ONU gateway through the lightweight neural network model and verifies whether the fault inference result is consistent with the candidate root cause fault and whether the fault confidence level is greater than or equal to a first threshold. If the conditions are met, the verification is successful. At this point, the current fault is determined to be an ONU gateway fault, and the fault detection procedure is triggered.

[0037] As an example, the "IoT device abnormal status signal" can be an alarm message actively reported by the IoT device, the device status actively accessed by the ONU gateway within a preset period, or a device abnormality notification pushed to the ONU gateway by the cloud platform. The candidate root cause fault can be one or more, for example, it may simultaneously match "Wi-Fi channel congestion" and "insufficient ONU uplink bandwidth".

[0038] As an example, the association rule base, in addition to being generated through cloud training, can also be updated through local incremental learning. Both methods can ensure the accuracy and timeliness of the association rules. Furthermore, the candidate root cause faults and the fault inference results, besides being input into the lightweight neural network model for inference analysis, can also be determined through feature threshold matching, which can similarly verify the accuracy of the candidate ONU faults. In one embodiment of the present invention, the specific process of step S140, "when the current fault state is a confirmed fault, the feasibility of repairing the current fault is determined by matching the preset repair strategy library according to the fault type label and executing the fault repair program," can be further explained in conjunction with the following description.

[0039] As described in the following steps Based on the fault type label, the preset repair strategy library is matched to determine the feasibility of repairing the current fault; When the repair feasibility is determined to be an unrepairable fault, the unrepairable fault alarm process is automatically executed; When the repair feasibility is determined to be a repairable fault, the repair strategy corresponding to the fault type tag is extracted from the repair strategy library, the repair action is executed according to the step sequence of the repair strategy, and the execution log is recorded. After the repair action is completed, the repair effect verification process is immediately executed based on the feature vector and the preset verification rules.

[0040] In one specific implementation, based on the fault type label of "Wi-Fi channel congestion", the processing chip queries the repair strategy library to determine if the fault is repairable, and extracts the repair actions corresponding to the fault type label from the repair strategy library as follows: "1. Scan and switch to the optimal channel; 2. Adjust the transmission power; 3. If there are too many terminals, enable rate limiting". The repair actions are executed in the order of the steps. The first step is to switch the channel and record "channel switched from 1 to 2" in the log. Then, the repair effect verification process is immediately executed according to the feature vector and the preset verification rules.

[0041] As an example, the number of repair strategies matches the number of fault type tags, and the repair actions include, but are not limited to, a combination of one or more actions such as restarting the service, switching configurations, adjusting parameters, and disabling functions.

[0042] In one embodiment of the present invention, the specific process of "after the repair action is completed, the repair effect verification process is immediately executed according to the feature vector and the preset verification rules" can be further described in conjunction with the following description.

[0043] As described in the following steps During the preset repair monitoring period, the feature vector in the current ONU gateway is collected in real time; Extract the corresponding feature data related to the fault type label from the feature vector; If the corresponding feature data remains below a preset normal threshold during the repair monitoring period, the repair is deemed successful; otherwise, the repair is deemed unsuccessful and an unrepairable fault alarm procedure is triggered. It should be noted that the repair effect verification process is automatically triggered after each repair action is completed, and verifies whether the currently completed repair action has successfully resolved the current fault based on the feature vector and preset verification rules.

[0044] In one specific implementation, after the self-healing action for "Wi-Fi channel congestion" is completed, a 30-second repair monitoring period is immediately entered. During the repair monitoring period, the feature vector is collected in real time, and the corresponding feature data related to "Wi-Fi channel congestion" is extracted from the feature vector. The corresponding feature data includes the channel utilization rate and the packet loss rate. If the channel utilization rate and the packet loss rate are continuously lower than the normal threshold (e.g., channel utilization rate < 60%, packet loss rate < 1%) during the repair monitoring period, the repair is determined to be successful, the normal collection frequency is restored, and a success log is recorded. In one specific implementation, if the corresponding feature data exceeds the normal threshold during the repair monitoring period, the repair is determined to have failed and an unrepairable fault alarm procedure is triggered, including pushing alarm information to the user, generating an alarm work order, and automatically reporting to the operation and maintenance platform.

[0045] As an example, the repair monitoring period can be adaptively set to different durations such as 30 seconds, 1 minute, or 2 minutes depending on the different fault types; "continuously below" can refer to all the corresponding feature data being below the normal threshold during the repair monitoring period.

[0046] As the apparatus embodiment is basically similar to the method embodiment, it is described in a relatively simple manner. For relevant details, please refer to the description of the method embodiment.

[0047] Reference Figure 2 This illustration shows an embodiment of an ONU gateway fault intelligent repair device based on lightweight feature fusion, specifically including the following modules: Specifically, it includes: The feature acquisition and preprocessing module 210 is used to acquire lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the lightweight features and preset input requirements to generate feature vectors. The fault detection module 220 is used to perform fault detection based on the feature vector and the preset lightweight neural network model, and generate fault type labels and fault confidence scores. The fault status determination module 230 is used to determine the current fault status based on the fault confidence level, wherein the current fault status includes confirmed fault, suspected fault, and no fault. The fault handling module 240 is triggered according to the fault status and is used to match the fault type label with a preset repair strategy library to determine the repair feasibility of the current fault and automatically execute the corresponding fault handling program.

[0048] In one embodiment of the present invention, the feature acquisition and preprocessing module 210 includes: The feature acquisition submodule is used to acquire lightweight features of the target ONU gateway within the current fusion cycle. The lightweight features include the hardware status features, the Wi-Fi status features, the DNS status features, and the IoT association features. The feature data fusion submodule is used to align the hardware status features, Wi-Fi status features, DNS status features and IoT association features within the fusion period, calculate the statistical values ​​of the hardware status features, Wi-Fi status features, DNS status features and IoT association features within the fusion period, and generate fused feature data. The data preprocessing submodule is used to perform one of the following on the fused feature data, based on the data type: principal component analysis dimensionality reduction, Min-Max standardization, and numerical vector transformation. The preprocessed feature data is then concatenated to generate the feature vector with a fixed dimension.

[0049] In one embodiment of the present invention, the fault detection module 220 includes: The fault classification submodule is used to perform inference analysis based on the feature vector through the lightweight neural network model to generate the fault type label; The fault confidence acquisition submodule is used to generate the fault confidence of the current fault type label by performing inference analysis through the lightweight neural network model based on the feature vector and the fault type label. The candidate root cause fault matching submodule is used to extract at least one candidate root cause fault that meets the conditions based on the abnormal status signals uploaded by IoT devices and the preset association rule library. The fault reasoning submodule is used to collect lightweight features of the ONU gateway in real time based on the candidate root cause faults, and extract corresponding feature data associated with the candidate root cause faults; generate the feature vector based on the corresponding feature data, and input it into the lightweight neural network model to generate fault reasoning results; The candidate root cause fault confidence submodule is used to perform reasoning analysis through the lightweight neural network based on the fault reasoning result and the candidate root cause fault, and generate the fault confidence that matches the current fault reasoning result and the candidate root cause fault. The reverse positioning submodule is used to compare and verify the fault inference results and the candidate root cause faults, and to trace the fault source in combination with the fault confidence to determine whether the current fault root cause is an ONU gateway fault or an IoT device fault itself.

[0050] In one embodiment of the present invention, the fault state determination module 230 includes: The fault status determination submodule is used to determine the current fault status based on the fault confidence level, wherein the fault status includes confirmed fault, suspected fault, and no fault. The fault handling procedure triggering submodule is used to trigger the corresponding fault handling procedure based on the determined fault state type.

[0051] In one embodiment of the present invention, the fault handling module 240 includes: The repair strategy matching submodule is used to match the preset repair strategy library according to the fault type label and generate a matching result; The repair feasibility judgment submodule is used to determine the repair feasibility of the current fault based on the matching result between the fault type label and the repair strategy library, including whether the fault is repairable or not, and to trigger the corresponding processing program. The repair action execution submodule is triggered when the repair feasibility is determined to be repairable. It is used to extract the repair strategy corresponding to the fault type tag in the repair strategy library, execute the repair action according to the step sequence of the repair strategy, and record the execution log. The repair effect verification submodule is triggered after the repair action is completed. It is used to collect the corresponding feature data related to the fault type tag in the current ONU gateway in real time during the preset repair monitoring period, and verify whether the current fault has been repaired based on the corresponding feature data and the preset normal threshold during the repair monitoring period, and trigger the corresponding subsequent program according to the repair status.

[0052] Reference Figure 3 This document illustrates a flowchart of a fault detection and repair method for an ONU gateway based on lightweight feature fusion, according to an embodiment of this application. The method includes fault detection based on the feature vector and a preset lightweight neural network model, and fault repair based on the fault state and a preset self-healing strategy library. This method enables the ONU gateway to automatically classify faults and perform local repairs.

[0053] refer to Figure 4This document illustrates a flowchart of the training, deployment, and iterative upgrade process of the lightweight neural network model provided in an embodiment of this application. It also shows how a large number of historical fault samples are used to train the lightweight neural network model to improve its fault detection accuracy and deploy it in the ONU gateway. During use, new fault samples are continuously collected for retraining and iterative upgrades to continuously improve classification accuracy and ensure fault detection accuracy.

[0054] Reference Figure 5 This document illustrates a reverse location flowchart of an ONU gateway fault intelligent repair method based on lightweight feature fusion, provided in an embodiment of this application. By constructing an "ONU-IoT association model + bidirectional verification process," the method reverse-locates the root cause of the ONU fault from the IoT anomaly based on the abnormal status signal uploaded by the IoT device, thus solving the problem of "fault misjudgment" and further improving the accuracy of fault location.

[0055] Reference Figure 6 The illustration shows a computer electronic device for implementing a method for intelligent fault repair of an ONU gateway based on lightweight feature fusion according to the present invention, which may specifically include the following: The aforementioned computer electronic device 1 is manifested in the form of a general-purpose computing device. The components of a computer electronic device 1 may include, but are not limited to: one or more processors or processing units 3, memory 8, and a bus 4 connecting different system components (including memory 8 and processing unit 3).

[0056] Bus 4 represents one or more of several bus architectures, including memory buses or memory controllers, peripheral buses, graphics acceleration ports, processors, or local buses using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Audio / Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0057] A computer electronic device 1 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by a computer electronic device 1, including volatile and non-volatile media, removable and non-removable media.

[0058] Memory 8 may include computer system readable media in the form of volatile memory, such as random access memory 9 and / or cache memory 10. A computer electronic device 1 may further include other movable / non-movable, volatile / non-volatile computer system storage media. By way of example only, storage system 11 may be used to read and write non-movable, non-volatile magnetic media (commonly referred to as a "hard disk drive"). Although Figure 6As not shown, a disk drive for reading and writing to a removable non-volatile disk (such as a "floppy disk") and an optical disk drive for reading and writing to a removable non-volatile optical disk (such as a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 4 via one or more data media interfaces. The memory may include at least one program product having a set (e.g., at least one) of program modules 13 configured to perform the functions of the embodiments of this application.

[0059] A program / utility 12 having a set (at least one) of program modules 13 may be stored, for example, in memory. Such program modules 13 include—but are not limited to—an operating system, one or more application programs, other program modules 13, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 13 typically perform the functions and / or methods described in the embodiments of this application.

[0060] A computer electronic device 1 can also communicate with one or more external devices 2 (e.g., keyboard, pointing device, display 7, camera, etc.), and with one or more devices that enable an operator to interact with the computer electronic device 1, and / or with any device that enables the computer electronic device 1 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through I / O interface 6. Furthermore, the computer electronic device 1 can also communicate with one or more networks (e.g., local area network (LAN)), wide area network (WAN), and / or public networks (e.g., the Internet) via network adapter 5. Figure 6 As shown, network adapter 5 communicates with other modules of a computer electronic device 1 via bus 4. It should be understood that, although... Figure 6 As not shown, other hardware and / or software modules may be used in conjunction with a computer electronic device 1, including but not limited to: microcode, device drivers, redundant processing units 3, external disk drive arrays, RAID systems, tape drives, and data backup storage systems 11.

[0061] The processing unit 3 executes various functional applications and data processing by running programs stored in memory 8, such as implementing a method for intelligent repair of ONU gateway faults based on lightweight feature fusion provided in the embodiments of this application.

[0062] That is, when the above-mentioned processing unit 3 executes the above-mentioned program, it achieves the following: Obtain lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors; Fault detection is performed based on the feature vector and a preset lightweight neural network model, generating fault type labels and fault confidence scores. The current fault status is determined based on the fault confidence level, and the current fault status includes confirmed fault, suspected fault, and no fault. When the current fault status is a confirmed fault, the feasibility of repairing the current fault is determined by matching the fault type label with the preset repair strategy library and the fault repair procedure is executed.

[0063] In this application embodiment, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for intelligent repair of ONU gateway faults based on lightweight feature fusion as provided in all embodiments of this application.

[0064] That is, to implement the following when the program is executed by the processor: Obtain lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors; Fault detection is performed based on the feature vector and a preset lightweight neural network model, generating fault type labels and fault confidence scores. The current fault status is determined based on the fault confidence level, and the current fault status includes confirmed fault, suspected fault, and no fault. When the current fault status is a confirmed fault, the feasibility of repairing the current fault is determined by matching the fault type label with the preset repair strategy library and the fault repair procedure is executed.

[0065] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0066] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0067] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the operator's computer, partially on the operator's computer, as a standalone software package, partially on the operator's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the operator's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider). The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably.

[0068] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0069] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0070] The above provides a detailed description of the ONU gateway fault intelligent repair method and device based on lightweight feature fusion provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligent fault repair of ONU gateways based on lightweight feature fusion, characterized in that, The method performs fault detection based on lightweight features of the ONU gateway, wherein the lightweight features include hardware status features, Wi-Fi status features, DNS status features, and IoT association features; the method includes: Obtain lightweight features of the target ONU gateway within the current fusion cycle, and perform fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors; Fault detection is performed based on the feature vector and a preset lightweight neural network model, generating fault type labels and fault confidence scores. The current fault status is determined based on the fault confidence level; wherein, the current fault status includes confirmed fault, suspected fault, and no fault; When the current fault status is a confirmed fault, the feasibility of repairing the current fault is determined by matching the fault type label with the preset repair strategy library and the fault repair procedure is executed.

2. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 1, characterized in that, The step of determining the current fault status based on the fault confidence level, wherein the current fault status includes the steps of confirmed fault, suspected fault, and no fault, includes: Based on the feature vector, the lightweight neural network model is used for inference and analysis to locate the current root cause of the fault as an ONU gateway fault, and the fault type label and the fault confidence level are generated. When the fault confidence level is greater than the first threshold of the fault determination threshold, the fault status is determined to be a confirmed fault. When the fault confidence level is between the first threshold and the second threshold of the fault determination threshold, the fault state is determined to be a suspected fault. If the fault confidence level is less than the second threshold of the fault determination threshold, then the fault state is determined to be fault-free.

3. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 1, characterized in that, Also includes: When the fault status is a suspected fault, the feature collection frequency is doubled and data is collected and analyzed again. If the fault confidence is greater than the first threshold of the fault determination threshold for three consecutive times, it is determined to be a confirmed fault; otherwise, the suspected status is lifted. When the fault status is fault-free, the normal feature acquisition frequency is maintained.

4. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 1, characterized in that, Also includes: Based on the abnormal status signals uploaded by IoT devices and the pre-set association rule library, extract at least one candidate root cause fault that meets the conditions; Based on the candidate root cause faults, lightweight features of the ONU gateway are collected in real time, and corresponding feature data associated with the candidate root cause faults are extracted. The feature vector is generated based on the corresponding feature data and input into the lightweight neural network model for inference analysis to generate fault inference results and fault confidence. When the fault reasoning result is consistent with the candidate root cause fault and the fault confidence is greater than or equal to the first threshold, the current fault root cause is determined to be an ONU gateway fault and the fault status is a confirmed fault; otherwise, the current fault root cause is determined to be an IoT device fault.

5. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 1, characterized in that, The step of obtaining lightweight features of the target ONU gateway within the current fusion cycle, and performing fusion preprocessing based on the hardware status features, Wi-Fi status features, DNS status features, IoT association features, and preset input requirements to generate feature vectors includes: The hardware status characteristics, Wi-Fi status characteristics, DNS status characteristics, and IoT association characteristics within the fusion cycle are aligned. Calculate the statistical values ​​of the hardware status features, Wi-Fi status features, DNS status features, and IoT association features during the fusion period to generate fusion feature data; The fused feature data is subjected to principal component analysis for dimensionality reduction, Min-Max standardization, and numerical vector transformation to generate the feature vector with fixed dimensions.

6. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 1, characterized in that, When the current fault status is a confirmed fault, the steps of matching the fault type label with a preset repair strategy library, determining the repair feasibility of the current fault, and automatically executing the corresponding processing procedure include: Based on the fault type label, the preset repair strategy library is matched to determine the feasibility of repairing the current fault; When the repair feasibility is determined to be an unrepairable fault, the unrepairable fault alarm process is automatically executed; When the repair feasibility is determined to be a repairable fault, the repair strategy corresponding to the fault type tag is extracted from the repair strategy library, the repair action is executed according to the step sequence of the repair strategy, and the execution log is recorded. After the repair action is completed, the repair effect verification process is immediately executed based on the feature vector and the preset verification rules.

7. The intelligent fault repair method for ONU gateway based on lightweight feature fusion according to claim 6, characterized in that, After the repair action is completed, the repair effect verification process is immediately executed based on the feature vector and preset verification rules, including: During the preset repair monitoring period, the feature vector in the current ONU gateway is collected in real time; Extract the corresponding feature data related to the fault type label from the feature vector; If the corresponding feature data remains below a preset normal threshold during the repair monitoring period, the repair is deemed successful; otherwise, the repair is deemed unsuccessful and an unrepairable fault alarm procedure is triggered.

8. A lightweight feature fusion-based intelligent fault repair device for ONU gateways, characterized in that, include: Feature acquisition and preprocessing module: Acquires lightweight features of the target ONU gateway within the current fusion cycle, performs fusion preprocessing based on the lightweight features and preset input requirements, and generates feature vectors; Fault detection module: performs fault detection based on the feature vector and a preset lightweight neural network model, and generates fault type labels and fault confidence scores; Fault Status Determination Module: Determines the current fault status based on the fault confidence level, whereby the current fault status includes confirmed fault, suspected fault, and no fault; Fault handling module: Triggered by the fault status, the module matches the fault type label with a preset repair strategy library to determine the feasibility of repairing the current fault and automatically executes the corresponding fault handling program.

9. A computer electronic device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When executed by the processor, the computer program implements the steps of the ONU gateway fault intelligent repair method based on lightweight feature fusion as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, it implements the steps of the ONU gateway fault intelligent repair method based on lightweight feature fusion as described in any one of claims 1 to 7.

Images