Passive equipment state monitoring method, device, equipment, medium and product

Abstract

The embodiment of the invention provides a passive equipment state monitoring method, device and equipment, a medium and a product, which are applied to the technical field of optical communication. The method comprises the following steps: acquiring multi-source monitoring data of a target passive optical network based on a preset acquisition frequency; wherein the multi-source monitoring data comprises passive equipment state data, network performance data, service operation data and environment associated data in the target passive optical network; performing feature fusion processing based on the multi-source monitoring data to obtain a multi-dimensional feature vector; and performing multi-task analysis on the target passive optical network based on the multi-dimensional feature vector to obtain a passive equipment state reasoning set, a passive equipment fault root cause set and a passive equipment predictive maintenance set of the target passive optical network. The method achieves the technical effect of improving the fault management efficiency of the passive optical network.

Description

Technical Field

[0001] This application relates to the field of optical communication technology, and in particular to a passive device condition monitoring method, apparatus, equipment, medium and product. Background Technology

[0002] Passive optical networks (PONs), as the core architecture of fiber optic access technology, are widely used in signal network construction and digital economy infrastructure development. PONs support large-scale user access through a point-to-multipoint topology; however, due to complex access requirements, the network architecture has become increasingly complex. For PONs with complex architectures, appropriate fault detection methods are needed to detect equipment faults within the network, enabling technical fault diagnosis and handling, and preventing further escalation of problems.

[0003] In existing technologies, the main method for fault detection in passive optical networks is to collect the optical power in the passive optical network in real time, and then analyze the optical power data corresponding to each device node by combining it with rules set based on historical experience or trained models, so as to achieve the purpose of fault determination based on the current optical power data.

[0004] Because existing fault detection methods for passive optical networks rely on single optical power data, it is difficult to couple the effects of multiple factors on the fault, resulting in low efficiency in fault management of passive optical networks. Summary of the Invention

[0005] This application provides passive device status monitoring methods, apparatus, equipment, media, and products to achieve the technical effect of improving the efficiency of passive optical network fault management.

[0006] In a first aspect, embodiments of this application provide a passive device status monitoring method, including:

[0007] Based on a preset acquisition frequency, multi-source monitoring data of the target passive optical network is collected; the multi-source monitoring data includes passive device status data, network performance data, service operation data, and environmental correlation data in the target passive optical network.

[0008] Multi-dimensional feature vectors are obtained by performing feature fusion processing based on multi-source monitoring data.

[0009] Multi-task analysis of the target passive optical network is performed based on multi-dimensional feature vectors to obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

[0010] In one possible implementation, multi-task analysis is performed on the target passive optical network based on multi-dimensional feature vectors to obtain a set of passive device state inferences, a set of passive device fault root causes, and a set of passive device predictive maintenance, including:

[0011] Passive device state reasoning is performed based on multi-dimensional feature vectors to obtain the state reasoning results corresponding to each passive device in the target passive optical network, and a passive device state reasoning set is generated.

[0012] Based on the state reasoning results for each passive device, at least one target faulty device is identified, along with the fault type corresponding to each target faulty device, and a set of faulty devices is generated.

[0013] For each target faulty device in the faulty device set, root cause analysis and predictive maintenance analysis are performed to obtain the passive device fault root cause set and the passive device predictive maintenance set.

[0014] In one possible implementation, root cause analysis and predictive maintenance analysis are performed on each target faulty device in the faulty device set to obtain a passive device fault root cause set and a passive device predictive maintenance set, including:

[0015] For each target faulty device in the faulty device set, root cause analysis is performed to obtain the fault root cause chain corresponding to the fault type.

[0016] Based on each target faulty device, the fault type of the target faulty device, and the root cause chain of the fault type, a set of root causes of passive device faults is generated.

[0017] For each target faulty device, identify its corresponding other related devices, perform predictive maintenance analysis for each related device, obtain the predictive maintenance analysis results for each related device, and generate a set of predictive maintenance for passive devices based on each related device and its corresponding predictive maintenance analysis results.

[0018] Other associated devices include devices of the same type as the target faulty device, other passive devices under the route to which the target faulty device belongs, and other passive devices topologically associated with the target faulty device.

[0019] In one possible implementation, multi-source monitoring data of the target passive optical network is collected based on a preset acquisition frequency, including:

[0020] Multi-source monitoring data is collected by the optical line terminal side acquisition unit, optical network unit side acquisition unit, passive link sensing unit and environmental service acquisition unit deployed in the target passive optical network, according to the preset acquisition frequency.

[0021] The passive link sensing unit includes a distributed optical fiber temperature measurement system and an optical time domain reflectometer. The distributed optical fiber temperature measurement system is used to collect temperature data of the trunk optical cable, and the optical time domain reflectometer is used to collect data on optical cable link attenuation and breakpoints.

[0022] In one possible implementation, feature fusion processing is performed based on multi-source monitoring data to obtain a multi-dimensional feature vector, including:

[0023] Data preprocessing is performed on multi-source monitoring data to obtain standard multi-source monitoring data; the data preprocessing includes time-series alignment, outlier removal, and image data enhancement.

[0024] Temporal, spatial, and correlation features are extracted from standard multi-source monitoring data, and these features are then fused to obtain a multi-dimensional feature vector.

[0025] In one possible implementation, multi-task analysis of the target passive optical network based on multi-dimensional feature vectors further includes:

[0026] If the target passive optical network involves cross-regional operation and maintenance scenarios, a federated learning framework is used to perform multi-task analysis on the target passive optical network.

[0027] In one possible implementation, the method further includes:

[0028] Obtain operation and maintenance questions for the target passive optical network input by the target user; operation and maintenance questions include fault consultation, equipment status inquiry and operation consultation;

[0029] Semantic parsing is performed on operation and maintenance issues to extract key information, including the type of equipment involved, the type of fault, and the query dimensions.

[0030] Based on key information, the system retrieves corresponding data from the passive device state reasoning set, the passive device fault root cause set, and the passive device predictive maintenance set to generate a response result containing the fault root cause, handling steps, and operation instructions, and then pushes the response result to the target user's terminal.

[0031] Secondly, embodiments of this application provide a passive device condition monitoring device, comprising:

[0032] The acquisition module is used to collect multi-source monitoring data of the target passive optical network based on a preset acquisition frequency. The multi-source monitoring data includes passive device status data, network performance data, service operation data, and environmental correlation data in the target passive optical network.

[0033] The first processing module is used to perform feature fusion processing based on multi-source monitoring data to obtain multi-dimensional feature vectors;

[0034] The second processing module is used to perform multi-task analysis on the target passive optical network based on multi-dimensional feature vectors, and obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

[0035] In one possible implementation, the second processing module is further configured to:

[0036] Passive device state reasoning is performed based on multi-dimensional feature vectors to obtain the state reasoning results corresponding to each passive device in the target passive optical network, and a passive device state reasoning set is generated.

[0037] Based on the state reasoning results for each passive device, at least one target faulty device is identified, along with the fault type corresponding to each target faulty device, and a set of faulty devices is generated.

[0038] For each target faulty device in the faulty device set, root cause analysis and predictive maintenance analysis are performed to obtain the passive device fault root cause set and the passive device predictive maintenance set.

[0039] In one possible implementation, the second processing module is further configured to:

[0040] For each target faulty device in the faulty device set, root cause analysis is performed to obtain the fault root cause chain corresponding to the fault type.

[0041] Based on each target faulty device, the fault type of the target faulty device, and the root cause chain of the fault type, a set of root causes of passive device faults is generated.

[0042] For each target faulty device, identify its corresponding other related devices, perform predictive maintenance analysis for each related device, obtain the predictive maintenance analysis results for each related device, and generate a set of predictive maintenance for passive devices based on each related device and its corresponding predictive maintenance analysis results.

[0043] Other associated devices include devices of the same type as the target faulty device, other passive devices under the route to which the target faulty device belongs, and other passive devices topologically associated with the target faulty device.

[0044] In one possible implementation, the acquisition module is further configured to:

[0045] Multi-source monitoring data is collected by the optical line terminal side acquisition unit, optical network unit side acquisition unit, passive link sensing unit and environmental service acquisition unit deployed in the target passive optical network, according to the preset acquisition frequency.

[0046] The passive link sensing unit includes a distributed optical fiber temperature measurement system and an optical time domain reflectometer. The distributed optical fiber temperature measurement system is used to collect temperature data of the trunk optical cable, and the optical time domain reflectometer is used to collect data on optical cable link attenuation and breakpoints.

[0047] In one possible implementation, the first processing module is further configured to:

[0048] Data preprocessing is performed on multi-source monitoring data to obtain standard multi-source monitoring data; the data preprocessing includes time-series alignment, outlier removal, and image data enhancement.

[0049] Temporal, spatial, and correlation features are extracted from standard multi-source monitoring data, and these features are then fused to obtain a multi-dimensional feature vector.

[0050] In one possible implementation, the second processing module is further configured to:

[0051] If the target passive optical network involves cross-regional operation and maintenance scenarios, a federated learning framework is used to perform multi-task analysis on the target passive optical network.

[0052] In one possible implementation, the device further includes a third processing module for:

[0053] Obtain operation and maintenance questions for the target passive optical network input by the target user; operation and maintenance questions include fault consultation, equipment status inquiry and operation consultation;

[0054] Semantic parsing is performed on operation and maintenance issues to extract key information, including the type of equipment involved, the type of fault, and the query dimensions.

[0055] Based on key information, the system retrieves corresponding data from the passive device state reasoning set, the passive device fault root cause set, and the passive device predictive maintenance set to generate a response result containing the fault root cause, handling steps, and operation instructions, and then pushes the response result to the target user's terminal.

[0056] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0057] The memory stores instructions that the computer executes;

[0058] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect above and various possible implementations of the first aspect.

[0059] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and various possible implementations thereof.

[0060] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and various possible implementations thereof.

[0061] This application provides a passive device status monitoring method, apparatus, device, medium, and product. The method utilizes a preset acquisition frequency to collect multi-source monitoring data, comprising passive device status data, network performance data, service operation data, and environmental correlation data in a target passive optical network. Feature fusion processing is performed on the multi-source monitoring data to obtain a multi-dimensional feature vector. Multi-task analysis is then performed on the target passive optical network using the multi-dimensional feature vector to obtain a status inference set corresponding to each passive device in the network, a set of device-related fault root causes, and a set of predictive maintenance parameters for the passive devices. In the data acquisition phase, this application collects multi-source monitoring data from the passive optical network, avoiding reliance on single data sources. Simultaneously, by utilizing feature fusion and multi-task analysis, it obtains not only status inference results for passive devices but also a set of fault root causes for device faults and a set of predictive maintenance results for the devices. This allows for accurate fault detection using multi-source monitoring data and comprehensive fault inference and final maintenance analysis using multi-task analysis, achieving the technical effect of improving the efficiency of passive optical network fault management. Attached Figure Description

[0062] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0063] Figure 1 Flowchart of the passive device condition monitoring method provided in this application Figure 1 ;

[0064] Figure 2 Flowchart of the passive device condition monitoring method provided in this application Figure 2 ;

[0065] Figure 3 Flowchart of the passive device condition monitoring method provided in this application Figure 3 ;

[0066] Figure 4 A schematic diagram of the passive equipment condition monitoring system provided in this application;

[0067] Figure 5 A schematic diagram of the passive equipment condition monitoring device provided in this application;

[0068] Figure 6 A schematic diagram of the structure of the electronic device provided in this application.

[0069] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0070] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0071] First, let me explain the terms used in this application:

[0072] Passive Optical Network (PON): refers to a point-to-multipoint fiber optic transmission and access technology that enables data transmission from a single point to multiple user terminals.

[0073] Optical Line Terminal (OLT): This refers to the core equipment in a passive optical network. It is usually deployed in the operator's equipment room and is mainly responsible for connecting the optical fiber trunk, converting electrical signals from the core network into optical signals, and transmitting them to the user-side equipment through the optical distribution network.

[0074] An Optical Network Unit (ONU) is a user-side device in a passive optical network, typically installed at the user end. The ONU receives optical signals transmitted from the optical line terminal (OLT) and converts them into electrical signals for use by the user-side terminal equipment.

[0075] An optical time-domain reflectometer (OTDR) is an instrument used for testing and diagnosing fiber optic links. This instrument measures fiber breakpoints, length, loss, and connection locations by sending light pulses into the fiber and analyzing the reflected signals.

[0076] In existing technologies, when performing fault detection on passive optical networks (PONs), it is first necessary to obtain the optical power of each connection point or passive device in the PON. The topology information and optical power data of the current PON are used as the basis for prediction. The basic data are then processed using a pre-set rule base or a model trained with historical data to obtain the fault prediction results in the PON.

[0077] However, current technologies for fault detection in passive optical networks (PONs) primarily rely on models based on historical experience or data, combined with real-time acquired optical power data for fault prediction. When PON optical power changes due to environmental factors or equipment modifications, the predictions from the rule base or model fail to account for these environmental variations. This results in a significant discrepancy between the detected fault and the actual fault, hindering effective targeted maintenance and leading to low efficiency in PON fault management.

[0078] To address the aforementioned technical problems, this application proposes the following technical concept: fault detection is performed using a multi-source data combination approach, simultaneously analyzing fault root causes and predictive maintenance. Specifically, multi-source monitoring data is collected from the target passive optical network (PON) based on a preset acquisition frequency; feature fusion processing is performed on the multi-source monitoring data to obtain multi-dimensional feature vectors. Multi-task analysis is then performed using these multi-dimensional feature vectors to obtain a passive device state inference set, a passive device fault root cause set, and a predictive maintenance set for passive devices, thereby achieving state inference, fault detection analysis, and predictive maintenance. Compared to existing technologies, this application utilizes multi-source monitoring data for state inference, fault detection analysis, and predictive maintenance, avoiding the low management efficiency caused by reliance on single data sources in existing technologies, and achieving the technical effect of improving the efficiency of PON fault management.

[0079] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0080] Figure 1 Flowchart of the passive device condition monitoring method provided in this application Figure 1 ,like Figure 1 As shown, the method includes:

[0081] S101. Based on the preset acquisition frequency, acquire multi-source monitoring data of the target passive optical network.

[0082] In this step, the multi-source monitoring data includes passive device status data, network performance data, service operation data, and environmental correlation data in the target passive optical network (PON). Passive device status data refers to the physical or operational status data of passive devices in the target PON, including but not limited to backbone cable temperature, fiber optic link attenuation and breakpoints, and splitter port status. Network performance data refers to the core performance indicators of network transmission and equipment operation in the target PON, including but not limited to optical power, bit error rate, and bandwidth utilization of optical line terminals, PON ports, or optical network units. Service operation data refers to relevant data during actual user service usage in the target PON, including but not limited to user service records, service interruption records, and traffic fluctuations. Environmental correlation data refers to external environmental data or auxiliary information data affecting the operation of the target PON, including but not limited to construction records, user complaints, and equipment runtime. The above multi-source monitoring data includes both fault-related anomaly data and baseline data during normal operation of passive devices.

[0083] Alternatively, one possible implementation method for multi-source monitoring data acquisition is:

[0084] Multi-source monitoring data is collected by the optical line terminal-side acquisition unit, optical network unit-side acquisition unit, passive link sensing unit, and environmental service acquisition unit deployed in the target passive optical network, according to the preset acquisition frequency.

[0085] For example, the method for obtaining multi-source monitoring data based on each acquisition unit can be as follows:

[0086] a1. Data is collected from the optical line terminal equipment itself and its downlink using the acquisition unit on the optical line terminal side. For example, optical power and bit error rate at the passive optical network port.

[0087] a2. Collect the operating status and service data of the user-end optical network unit based on the acquisition unit on the optical network unit side. For example, such as received optical power and service interruption records.

[0088] a3. Collect data on trunk optical cable temperature, optical cable link attenuation, and breakpoints based on passive link sensing units.

[0089] a4. Based on the environmental business data collection unit, connect to the user complaint system and construction management system to collect external related data.

[0090] In this step, the passive link sensing unit includes a distributed fiber optic temperature measurement system and an optical time-domain reflectometer (OTDR). The distributed fiber optic temperature measurement system is used to collect temperature data of the trunk optical cable, and the OTD is used to collect data on optical cable link attenuation and breakpoints. The preset acquisition frequency refers to the pre-set acquisition period for various types of data.

[0091] For example, the preset acquisition frequency for network performance data can be once every 1 second; the preset acquisition frequency for service operation data can be once every 1 hour. When acquiring data from the above multi-source monitoring data, the preset acquisition frequency can be a uniform acquisition frequency value or a combination of acquisition frequencies for monitoring data from different sources. The acquisition accuracy of the distributed fiber optic temperature measurement system can be set to ±0.5℃, and the acquisition accuracy of the optical time domain reflectometer can be set to ±1 meter.

[0092] S102. Based on multi-source monitoring data, feature fusion processing is performed to obtain multi-dimensional feature vectors.

[0093] In this step, the feature fusion process involves first extracting features from multi-source monitoring data, and then fusing the extracted features to obtain a multi-dimensional feature vector with a specific dimension. The feature dimension of this multi-dimensional feature vector is a pre-set dimension value.

[0094] Alternatively, one possible implementation for generating multi-dimensional feature vectors is as follows:

[0095] S1021. Perform data preprocessing based on multi-source monitoring data to obtain standard multi-source monitoring data.

[0096] In this step, data preprocessing refers to cleaning and optimizing the raw multi-source monitoring data to ensure its validity and consistency. Data preprocessing includes time-series alignment, outlier removal, and image data enhancement. Time-series alignment involves adjusting the timestamps of data from different sources to ensure that the time deviation of related data for the same event is within acceptable limits. Outlier removal filters out invalid data caused by signal interference and false alarms. Image data enhancement optimizes the clarity of image data such as port labels of optical splitters in the target passive optical network and Optical Time Domain Reflectometer (OTDR) curves, facilitating subsequent feature extraction. The OTDR curve refers to the test result curve from the optical time domain reflectometer, used to analyze the characteristics of the fiber optic link.

[0097] For example, time-series alignment can be achieved by using timestamp interpolation to supplement the timestamps of low-frequency data, based on high-frequency collected data. Outlier removal can be performed using the 3σ principle of normal distribution for anomaly detection, classifying data exceeding the mean ± 3σ as outliers, while also filtering invalid data using business thresholds. Here, σ refers to the standard deviation. Image enhancement can be achieved by using Gaussian noise reduction combined with histogram equalization to optimize the recognition accuracy of the splitter port label image and OTDR curve image.

[0098] For example, one possible way to preprocess the collected multi-source monitoring data to obtain standard multi-source monitoring data is as follows:

[0099] b1. Classify the raw multi-source monitoring data to obtain numerical data, image data, and text data.

[0100] b2. Perform timestamp interpolation on numerical data to ensure that the time deviation of data for the same event is less than 2 minutes.

[0101] b3. Apply the 3σ principle of normal distribution and business thresholds to numerical data to remove outliers that exceed the ±3σ range or do not meet the business thresholds.

[0102] b4. Perform Gaussian noise reduction and histogram equalization on image data to optimize clarity.

[0103] b5. Convert all types of data into a combined structure of device identifier, timestamp, data content, and status to obtain standard multi-source monitoring data.

[0104] S1022. Extract time-domain features, spatial features, and correlation features based on standard multi-source monitoring data, and fuse the time-domain features, spatial features, and correlation features to obtain a multi-dimensional feature vector.

[0105] In this step, time-domain features refer to features extracted based on the patterns of data changes over time, used to capture sudden faults and trend anomalies. For example, if the optical power drops sharply from -15dBm to -20dBm within 5 seconds, it can be identified as a feature corresponding to a sudden optical cable fault.

[0106] Spatial features refer to features extracted based on the physical location and network topology of passive devices, primarily used to locate faulty nodes. For example, if all faulty optical network units belong to the same passive optical network port with the splitter numbered 001, and the attenuation point is located on the main optical cable from the optical line terminal to the splitter, then the features of this part can be extracted as spatial features.

[0107] Correlation features refer to features extracted based on the correlation relationships between different passive devices and data from different dimensions. They are mainly used to identify link loss and fault correlations. For example, the difference between the transmitted optical power of an optical line terminal and the received optical power of an optical network unit can be regarded as a link loss feature, and the temporal correlation between construction records and optical cable attenuation can be regarded as a causal feature.

[0108] A multi-dimensional feature vector is a fixed-dimensional numerical vector formed by combining temporal features, spatial features, and correlation features. For example, if the multi-dimensional feature vector is set to a 512-dimensional numerical vector and the feature fusion method is concatenation fusion, then dimensions 1-128 are temporal features, dimensions 129-256 are spatial features, and dimensions 257-512 are correlation features.

[0109] For example, one possible implementation of extracting temporal features, spatial features, and correlation features from standard multi-source monitoring data and then performing feature fusion is as follows:

[0110] c1. The mean, variance, and abrupt change magnitude of the data are extracted using the sliding window method to obtain the time-domain features.

[0111] c2. Based on the network topology map, the device locations and topological relationships are converted into numerical codes to obtain spatial features.

[0112] c3. Calculate the data difference between devices and the correlation of event time to obtain the correlation characteristics.

[0113] c4. Using feature concatenation and normalization, the three types of features are allocated according to a fixed dimension and normalized to the [0,1] interval to generate a multi-dimensional feature vector.

[0114] S103. Perform multi-task analysis on the target passive optical network based on multi-dimensional feature vectors to obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

[0115] In this step, multi-task analysis refers to the analysis process that uses the same feature vector to perform three types of tasks in parallel: state reasoning, root cause analysis, and predictive maintenance. The passive device state reasoning set includes the current health status assessment results or failure probability assessment results for each passive device in the target passive optical network. The passive device failure root cause set includes the root cause chain of the failed device, its intermediate nodes, and the final phenomenon, as well as the confidence level of each root cause chain. The passive device predictive maintenance set refers to the set of potential failure risk warnings for other passive devices associated with the failed passive device, including the failure type and failure probability of each related device.

[0116] Optionally, if the target passive optical network involves cross-regional operation and maintenance scenarios, a federated learning framework is used to perform multi-task analysis on the target passive optical network.

[0117] In this step, the federated learning framework refers to a privacy-preserving shared analysis framework that interacts only with model parameters rather than raw data during cross-regional operations.

[0118] For example, a federated learning framework can be used for multi-task analysis as follows: First, edge nodes in each region perform real-time multi-task analysis and model training based on their local multi-dimensional feature vectors, and exchange model training parameters with the cloud. The cloud aggregates the model training parameters from each region for global model optimization, and then distributes the optimized parameters to the edge nodes in each region. The edge nodes in each region then use the optimized parameters for further multi-task analysis.

[0119] It should be noted that the specific implementation of the multi-task analysis in this application is as follows: Figure 2 Further explanation will be provided in the embodiments shown, and will not be repeated here.

[0120] The passive device status monitoring method provided in this application collects multi-source monitoring data, consisting of passive device status data, network performance data, service operation data, and environmental correlation data, from a target passive optical network using a preset acquisition frequency. Feature fusion processing is performed on the multi-source monitoring data to obtain a multi-dimensional feature vector. Multi-task analysis is then performed on the target passive optical network using this multi-dimensional feature vector to obtain a status inference set corresponding to each passive device in the network, a set of device-related fault root causes, and a set of predictive maintenance parameters for the passive devices. In the data acquisition phase, this application collects multi-source monitoring data from the passive optical network, avoiding reliance on single data sources. Simultaneously, by utilizing feature fusion and multi-task analysis, it obtains not only status inference results for passive devices but also a set of fault root causes for device faults and a set of predictive maintenance results for the devices. This allows for accurate fault detection using multi-source monitoring data and comprehensive fault inference and final maintenance analysis using multi-task analysis, achieving the technical effect of improving the efficiency of passive optical network fault management.

[0121] Figure 2 Flowchart of the passive device condition monitoring method provided in this application Figure 2 ,like Figure 2 As shown, the method includes:

[0122] S201. Perform passive device state reasoning based on multi-dimensional feature vectors to obtain the state reasoning results corresponding to each passive device in the target passive optical network, and generate a passive device state reasoning set.

[0123] In this step, the state reasoning result refers to the determination result of the current operating state of a single passive device, which can be either normal or faulty. When a fault exists, the corresponding fault type must also be included. The passive device state reasoning set refers to the structured set of all passive device state reasoning results.

[0124] Optionally, the passive device state inference set can be generated by: calculating the similarity between the input 512-dimensional feature vector and the normal state feature templates and various fault feature templates stored in the model. A similarity score greater than or equal to 85% is considered a match for the corresponding state, and the fault type is determined according to the template with the highest similarity. The data is then structured and stored according to the patterns of device identifier, state, and fault type, thus generating the passive device state inference set.

[0125] For example, the similarity between the input feature vector and the attenuation fault template of the splitter port is 92%, which is greater than 85%, and the similarity with the normal template is 15%.

[0126] The similarity between the input feature vector and the optical network unit insufficient optical power fault template is 88%, which is greater than 85%, while the similarity with the normal template is 20%.

[0127] The similarity between the input feature vector and the normal template of the optical line terminal is 90%, which is greater than 85%, while the similarity with the fault template is 12%.

[0128] The corresponding state reasoning result is:

[0129] OLT-PON3, Status: Normal, Fault Type: None. Here, OLT-PON3 refers to the passive optical network port numbered 3 of the optical line terminal.

[0130] Splitter 001, Status: Fault, Fault Type: Port Attenuation.

[0131] ONU-001~008, Status: Fault, Fault Type: Insufficient Optical Power. Here, ONU-001~008 refers to optical network units numbered 001 to 008.

[0132] S202. Based on the state reasoning results corresponding to each passive device, determine at least one target faulty device and the fault type corresponding to each target faulty device, and generate a set of faulty devices.

[0133] In this step, the target faulty device refers to the passive device identified as faulty or suspected of being faulty in the state reasoning result set. The fault type refers to the specific fault type corresponding to the faulty device. The faulty device set refers to the association set of the target faulty device and its corresponding fault type.

[0134] Alternatively, the set of faulty devices can be generated in the following ways:

[0135] e1. Select passive devices with a faulty state from the passive device state inference set and identify them as target faulty devices.

[0136] e2. Group the multiple selected target fault devices according to the fault type corresponding to the target fault device to obtain at least one target fault device group.

[0137] e3. If multiple target fault devices are caused by the same root device, then the root device is retained as the core target fault device, and the other devices are marked as associated fault devices.

[0138] The purpose of this step is to determine the correlation between equipment failures, which will facilitate subsequent root cause reasoning.

[0139] S203. Perform root cause analysis and predictive maintenance analysis on each target faulty device in the faulty device set to obtain the passive device fault root cause set and the passive device predictive maintenance set.

[0140] In this step, root cause analysis refers to tracing the complete logical chain of fault causes, intermediate fault nodes, and final fault phenomena based on the target faulty equipment, fault type, and multi-dimensional feature vectors. Predictive maintenance analysis refers to predicting potential future faults of related equipment by combining the root causes of core faulty equipment and historical fault patterns. The passive equipment fault root cause set refers to the relational set of faulty equipment, fault type, and fault root cause chains of all core target faulty equipment. The passive equipment predictive maintenance set refers to the structured set of predictive maintenance analysis results for all related equipment.

[0141] Alternatively, one possible implementation of root cause analysis and predictive maintenance analysis is as follows:

[0142] S2031. Perform root cause analysis on the fault type of each target faulty device in the faulty device set to obtain the fault root cause chain corresponding to the fault type.

[0143] In this step, the root cause chain refers to the structured root cause description arranged logically according to the above-mentioned fault causes, intermediate nodes and final phenomena. Each root cause chain also includes its confidence level.

[0144] Alternatively, one possible implementation of root cause analysis is as follows: Identify the direct causes of the failure based on the correlation features in a multi-dimensional feature vector. Complete the failure propagation path based on network transmission principles. Calculate the confidence level based on the matching degree between the propagation path and historical cases. Output a root cause chain consisting of the failure cause, intermediate nodes, final phenomenon, and confidence level.

[0145] For example, the associated features in the multi-dimensional feature vector are: the optical splitter has been running for 3.5 years, there is no record of construction interference, and the port attenuation of the optical splitter is 7dB. The cause of the failure is determined to be equipment aging. The intermediate nodes obtained based on the cause of the failure are: aging caused the attenuation of port 1 of the optical splitter to increase to 7dB; port attenuation caused a decrease in the received optical power of the 8 downstream optical network units. The final phenomenon is confirmed to be that the bandwidth of a certain leased line drops from 100Mbps to 20Mbps. The calculated matching degree with historical cases is 95%, that is, the confidence level is 95%.

[0146] S2032. Based on each target faulty device, the fault type of the target faulty device, and the root cause chain of the fault type, generate a set of root causes for passive device faults.

[0147] In this step, the passive device fault root cause set is generated as follows: the target faulty device and its corresponding fault type are bound to the generated fault root cause chain to obtain the root cause chain record corresponding to that device. If there is a core target faulty device whose fault causes other associated faulty devices to malfunction, then that device is designated as the core target faulty device, and only the root cause chain record corresponding to the core faulty device is retained. The structured set consisting of the root cause chain records of multiple faulty devices is then used as the passive device fault root cause set.

[0148] S2033. For each target faulty device, determine its corresponding other related devices, perform predictive maintenance analysis for each related device, obtain the predictive maintenance analysis results for each related device, and generate a set of predictive maintenance for passive devices based on each related device and its corresponding predictive maintenance analysis results.

[0149] In this step, other related devices include devices of the same type as the target faulty device, other passive devices under the route of the target faulty device, and other passive devices topologically associated with the target faulty device. Predictive maintenance refers to predicting potential future faults of related devices by combining the root causes and historical fault patterns of the target faulty device. The predictive maintenance analysis results refer to the details of potential faults of related devices, including device identifiers, potential fault types, fault probabilities, warning priorities, and maintenance measures.

[0150] Alternatively, one possible implementation of predictive maintenance analytics is as follows:

[0151] f1. Filtering associated devices:

[0152] Filter similar devices: Based on the device model database, match devices with the same model or core parameters as the target faulty device.

[0153] Filter devices on the same route: Based on the routing configuration information, extract other devices under the same transmission route as the target faulty device.

[0154] Filter topology-related devices: Based on the network topology map, filter devices that are directly or indirectly connected to the target faulty device.

[0155] f2. Predictive maintenance analysis:

[0156] Analyze the potential failure types of related equipment: Based on the failure type and root cause of the target faulty equipment, match the risk points of related equipment.

[0157] Failure probability analysis: The failure probability of related equipment is calculated based on historical failure statistics.

[0158] Determine the warning priority: The warning score is determined by the combination of failure probability and business importance. When the score is greater than or equal to 70, the warning priority is determined as high; when the score is greater than or equal to 50 and less than 70, the warning priority is determined as medium; when the score is less than 50, the warning priority is determined as low.

[0159] Determine maintenance strategy: Based on historical fault handling data, the device identifier of associated equipment and fault type, determine the corresponding maintenance strategy.

[0160] f3. Structure the data according to the associated equipment, potential fault types, fault probabilities, warning priorities, and maintenance strategies to obtain a set of predictive maintenance for passive equipment.

[0161] It should be noted that the analysis process of steps f1 to f3 above can be implemented based on a pre-trained machine learning model.

[0162] For example, the contents of the passive device predictive maintenance set can be: [Associated device: splitter 002, potential fault type: port attenuation, fault probability: 65%, warning priority: high, maintenance measure: measure A10}, {Associated device: optical cable segment A, potential fault type: construction damage, fault probability: 30%, warning priority: medium, maintenance measure: measure A12].

[0163] Based on the above embodiments, this application also provides a method for operation and maintenance of passive equipment failures. Figure 3 Flowchart of the passive device condition monitoring method provided in this application Figure 3 ,like Figure 3 As shown, the method includes:

[0164] S301. Obtain the operation and maintenance questions for the target passive optical network input by the target user.

[0165] In this step, maintenance issues include troubleshooting, device status inquiries, and operational inquiries. These issues can be obtained through a natural language interface that receives user input. This interface supports multiple input channels, including visual input boxes, a maintenance chatbot, and speech-to-text input. The user's natural language input is converted into standard text format, redundant information is removed, and core semantics are retained to obtain the user's maintenance issue.

[0166] S302. Perform semantic parsing on operation and maintenance issues to extract key information.

[0167] In this step, key information includes the device types, fault types, and query dimensions involved in the maintenance issue. Semantic parsing can be achieved by: using natural language processing algorithms for word segmentation, part-of-speech tagging, and intent recognition of the maintenance-related issues; accurately matching specialized terms during the parsing process using a passive optical network (PON) terminology dictionary to avoid semantic ambiguity; and combining device relationships and fault type labels from multi-dimensional feature vectors to pinpoint the target data range corresponding to the issue, ultimately extracting the device types, fault types, and query dimensions involved in the maintenance issue.

[0168] S303. Based on key information, call the corresponding data in the passive device state reasoning set, the passive device fault root cause set, and the passive device predictive maintenance set to generate a response result containing the fault root cause, processing steps, and operation instructions, and push the response result to the target user's terminal.

[0169] In this step, the response result refers to the natural language response generated by integrating three types of data sets in response to the user's operation and maintenance issues. Pushing to the user's terminal means pushing the response result to the user's terminal device through the interactive interface or the interface used for interaction.

[0170] Optionally, the response result can be generated and pushed out by: using a pre-trained model, combining the equipment type and fault type in the key information, matching corresponding data from the passive equipment status inference set, fault root cause set, and predictive maintenance set. The content is organized logically according to core conclusions, reasoning basis, practical steps, and risk warnings, and expressed in natural language. The core conclusions directly address the query dimensions; the reasoning basis references key data from the three sets; the practical steps are written based on application-layer operation and maintenance specifications; and the risk warnings reference early warning information from the predictive maintenance set. The response result is then pushed to the user terminal via an interactive device.

[0171] It should be noted that after the response results are pushed to the user's end, supplementary questions, confirmation of repair results, or feedback can be submitted through natural language interaction. The supplementary questions, confirmation of repair results, or feedback are then passed to the model's self-learning module. The self-learning module uses the feedback information as incremental training data to optimize the model's semantic parsing accuracy, response accuracy, and multi-task analysis parameters, thereby improving the natural language interaction closed loop of question input, response output, and feedback iteration.

[0172] In conjunction with the passive device condition monitoring method described in the above embodiments, this application also provides a passive device condition monitoring system. Figure 4 This is a schematic diagram of the passive device condition monitoring system provided in this application, as shown below. Figure 4 As shown, the system includes:

[0173] The perception layer 401 is used to collect multi-source monitoring data of the target passive optical network. Specifically, it includes: an OLT-side acquisition unit, an ONU-side acquisition unit, a passive link perception unit, and an environmental service acquisition unit. The OLT-side acquisition unit collects optical power, bandwidth, and bit error rate data from the OLT equipment in the operator's equipment room. The passive link perception unit collects backbone optical cable temperature detected by DTS sensors and cable link attenuation and breakpoint data detected by an OTDR tester. The ONU-side acquisition unit collects the operating status and service data of the user-end optical network unit. The environmental service acquisition unit collects externally related construction record data. Finally, the collected multi-source monitoring data is summarized and transmitted to the data processing layer 402 for processing.

[0174] The data processing layer 402 is used for data cleaning and standardization of multi-source monitoring data, and for feature extraction and feature fusion of standard multi-source monitoring data to obtain multi-dimensional feature vectors. Simultaneously, the multi-dimensional feature vectors and standardized multi-dimensional monitoring data can be stored in different databases using distributed storage, serving as subsequent feature computation data and historical data to facilitate subsequent model prediction and iterative model training. Specifically, distributed storage involves storing data in different databases according to data type; for example, time-series features or data are stored in a time-series database, image features or data are stored in an image database, and object and structural features or data are stored in an object database.

[0175] Layer 403, the large model layer, is used to perform multi-task analysis on multi-dimensional feature vectors using a pre-trained large model to obtain a set of passive device state inferences, a set of passive device fault root causes, and a set of passive device predictive maintenance. Specifically: It utilizes the large model for multimodal fault diagnosis, performing text parsing, image recognition, and temporal inference based on multi-dimensional feature vectors to obtain fault diagnosis results or state inference results for passive devices, thereby determining the fault state of each passive device. It also utilizes the large model for topology awareness and fault localization to obtain the fault root cause chain corresponding to each fault. Furthermore, it utilizes the large model for fault solution generation and verification to obtain solutions for each fault. A federated learning framework is used to perform federated learning in a cross-regional target passive network to optimize model parameters. Finally, it utilizes the large model for predictive maintenance analysis, performing predictive maintenance analysis on associated devices of faulty passive devices to obtain corresponding fault prediction conclusions and maintenance measure analyses. Finally, a self-learning module receives feedback data from the natural language interaction module of the interaction layer and a third-party model training platform, and optimizes and trains the model based on this feedback data.

[0176] Application layer 404 includes a real-time monitoring center, an intelligent fault management module, a network optimization toolkit, and a rule-based decision support module. The real-time monitoring center monitors the health of passive devices and analyzes passive links based on fault diagnosis information and root cause chains generated by the large model layer 403. The intelligent fault management module generates work orders and provides remote guidance for repair based on the root cause chains and fault solutions. The network optimization toolkit stores fault solutions, which can include split-ratio adjustment schemes, ONU channel planning, and energy-saving schemes. The planning decision support module stores the results of predictive maintenance analysis, where maintenance schemes can include OLT expansion schemes, fiber optic cable routing configuration schemes, and equipment replacement schemes.

[0177] The interaction layer 405 includes a natural language interaction module, a visual operation interface, and a system integration interface. The natural language interaction module receives work orders and remote guidance solutions output by the intelligent fault management module and provides feedback to maintenance or management personnel; it can also transmit user feedback data to the self-learning module of the large model layer 403. The visual operation interface displays information output by the application layer 404 and provides feedback to maintenance or management personnel. The system integration interface outputs fault solutions and predictive maintenance analysis results to third-party large model training platforms and operator management systems.

[0178] Figure 5 This is a schematic diagram of the passive equipment condition monitoring device provided in this application, as shown below. Figure 5 As shown, the passive device status monitoring device provided in this embodiment includes:

[0179] The acquisition module 501 is used to acquire multi-source monitoring data of the target passive optical network based on a preset acquisition frequency; wherein, the multi-source monitoring data includes passive device status data, network performance data, service operation data and environmental correlation data in the target passive optical network.

[0180] The first processing module 502 is used to perform feature fusion processing based on multi-source monitoring data to obtain multi-dimensional feature vectors.

[0181] The second processing module 503 is used to perform multi-task analysis on the target passive optical network based on multi-dimensional feature vectors, and obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

[0182] Alternatively, in one possible implementation, the second processing module 503 is further configured to:

[0183] Passive device state reasoning is performed based on multi-dimensional feature vectors to obtain the state reasoning results corresponding to each passive device in the target passive optical network, and a passive device state reasoning set is generated.

[0184] Based on the state reasoning results for each passive device, at least one target faulty device is identified, along with the fault type corresponding to each target faulty device, and a set of faulty devices is generated.

[0185] For each target faulty device in the faulty device set, root cause analysis and predictive maintenance analysis are performed to obtain the passive device fault root cause set and the passive device predictive maintenance set.

[0186] Alternatively, in one possible implementation, the second processing module 503 is further configured to:

[0187] For each target faulty device in the faulty device set, root cause analysis is performed to obtain the fault root cause chain corresponding to the fault type.

[0188] Based on each target faulty device, the fault type of the target faulty device, and the root cause chain of the fault type, a set of root causes for passive device faults is generated.

[0189] For each target faulty device, identify its corresponding other related devices, perform predictive maintenance analysis for each related device, obtain the predictive maintenance analysis results for each related device, and generate a set of predictive maintenance for passive devices based on each related device and its corresponding predictive maintenance analysis results.

[0190] Other associated devices include devices of the same type as the target faulty device, other passive devices under the route to which the target faulty device belongs, and other passive devices topologically associated with the target faulty device.

[0191] Alternatively, in one possible implementation, the acquisition module 501 is further configured to:

[0192] Multi-source monitoring data is collected by the optical line terminal-side acquisition unit, optical network unit-side acquisition unit, passive link sensing unit, and environmental service acquisition unit deployed in the target passive optical network, according to the preset acquisition frequency.

[0193] The passive link sensing unit includes a distributed optical fiber temperature measurement system and an optical time domain reflectometer. The distributed optical fiber temperature measurement system is used to collect temperature data of the trunk optical cable, and the optical time domain reflectometer is used to collect data on optical cable link attenuation and breakpoints.

[0194] Optionally, in one possible implementation, the first processing module 502 is further configured to:

[0195] Data preprocessing is performed on multi-source monitoring data to obtain standard multi-source monitoring data; the data preprocessing includes time-series alignment, outlier removal, and image data enhancement.

[0196] Temporal, spatial, and correlation features are extracted from standard multi-source monitoring data, and these features are then fused to obtain a multi-dimensional feature vector.

[0197] Alternatively, in one possible implementation, the second processing module 503 is further configured to:

[0198] If the target passive optical network involves cross-regional operation and maintenance scenarios, a federated learning framework is used to perform multi-task analysis on the target passive optical network.

[0199] Optionally, in one possible implementation, the device further includes a third processing module 504, for:

[0200] Obtain operation and maintenance questions for the target passive optical network input by the target user; operation and maintenance questions include fault consultation, equipment status inquiry and operation consultation.

[0201] Semantic parsing is performed on the operation and maintenance issues to extract key information, including the type of equipment involved, the type of fault, and the query dimensions.

[0202] Based on key information, the system retrieves corresponding data from the passive device state reasoning set, the passive device fault root cause set, and the passive device predictive maintenance set to generate a response result containing the fault root cause, handling steps, and operation instructions, and then pushes the response result to the target user's terminal.

[0203] The apparatus provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0204] Figure 6 A schematic diagram of the structure of the electronic device provided in this application. Figure 6 As shown, the electronic device provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0205] In the specific implementation process, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to execute the above-mentioned passive device status monitoring method or approach.

[0206] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0207] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0208] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0209] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0210] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0211] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0212] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0213] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0214] The division of units is merely a logical functional division; 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 indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0215] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0216] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0217] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of 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.

[0218] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0219] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for monitoring the condition of passive equipment, characterized in that, include: Based on a preset acquisition frequency, multi-source monitoring data of the target passive optical network is acquired; wherein, the multi-source monitoring data includes passive device status data, network performance data, service operation data, and environmental correlation data in the target passive optical network; Based on the multi-source monitoring data, feature fusion processing is performed to obtain a multi-dimensional feature vector; Based on the multi-dimensional feature vectors, multi-task analysis is performed on the target passive optical network to obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

2. The method according to claim 1, characterized in that, The multi-task analysis of the target passive optical network based on the multi-dimensional feature vector yields a set of passive device state inferences, a set of passive device fault root causes, and a set of passive device predictive maintenance, including: Based on the multi-dimensional feature vector, passive device state reasoning is performed to obtain the state reasoning results corresponding to each passive device in the target passive optical network, and the passive device state reasoning set is generated. Based on the state reasoning results corresponding to each passive device, at least one target faulty device and the fault type corresponding to each target faulty device are determined, and a set of faulty devices is generated. For each target faulty device in the set of faulty devices, root cause analysis and predictive maintenance analysis are performed to obtain the set of root causes of passive device faults and the set of predictive maintenance for passive devices.

3. The method according to claim 2, characterized in that, The step of performing root cause analysis and predictive maintenance analysis on each target faulty device in the set of faulty devices to obtain the set of root causes of passive device faults and the set of predictive maintenance for passive devices includes: For each target faulty device in the set of faulty devices, root cause analysis is performed to obtain the fault root cause chain corresponding to the fault type. Based on each target faulty device, the fault type of the target faulty device, and the root cause chain of the fault type, the root cause set of the passive device fault is generated. For each target faulty device, other associated devices are identified, and predictive maintenance analysis is performed on each associated device to obtain the predictive maintenance analysis results for each associated device. Based on each associated device and its corresponding predictive maintenance analysis results, the passive device predictive maintenance set is generated. The other associated devices include devices of the same type as the target faulty device, other passive devices under the route to which the target faulty device belongs, and other passive devices topologically associated with the target faulty device.

4. The method according to claim 1, characterized in that, The multi-source monitoring data of the target passive optical network, collected based on a preset acquisition frequency, includes: The multi-source monitoring data is collected by the optical line terminal-side acquisition unit, optical network unit-side acquisition unit, passive link sensing unit and environmental service acquisition unit deployed in the target passive optical network, according to the preset acquisition frequency. The passive link sensing unit includes a distributed optical fiber temperature measurement system and an optical time domain reflectometer. The distributed optical fiber temperature measurement system is used to collect temperature data of the trunk optical cable, and the optical time domain reflectometer is used to collect optical cable link attenuation and breakpoint data.

5. The method according to claim 1, characterized in that, The feature fusion processing based on the multi-source monitoring data yields a multi-dimensional feature vector, including: Based on the multi-source monitoring data, data preprocessing is performed to obtain standard multi-source monitoring data; wherein, the data preprocessing includes time-series alignment processing, outlier removal, and image data enhancement processing; Based on the standard multi-source monitoring data, temporal features, spatial features, and correlation features are extracted, and the temporal features, spatial features, and correlation features are fused to obtain the multi-dimensional feature vector.

6. The method according to claim 1, characterized in that, The multi-task analysis of the target passive optical network based on the multi-dimensional feature vector also includes: If the target passive optical network involves cross-regional operation and maintenance scenarios, a federated learning framework is used to perform multi-task analysis on the target passive optical network.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: Obtain operation and maintenance questions for the target passive optical network input by the target user; the operation and maintenance questions include fault consultation, equipment status query and operation consultation; Semantic parsing is performed on the aforementioned operation and maintenance issues to extract key information; wherein, the key information includes the equipment type, fault type, and query dimension involved in the operation and maintenance issues; Based on the key information, the corresponding data in the passive device state reasoning set, the passive device fault root cause set, and the passive device predictive maintenance set are invoked to generate a response result containing the fault root cause, processing steps, and operation instructions, and the response result is pushed to the target user's terminal.

8. A passive equipment condition monitoring device, characterized in that, include: The acquisition module is used to acquire multi-source monitoring data of the target passive optical network based on a preset acquisition frequency; wherein, the multi-source monitoring data includes passive device status data, network performance data, service operation data and environmental correlation data in the target passive optical network; The first processing module is used to perform feature fusion processing based on the multi-source monitoring data to obtain a multi-dimensional feature vector. The second processing module is used to perform multi-task analysis on the target passive optical network based on the multi-dimensional feature vector to obtain the passive device state inference set, the passive device fault root cause set, and the passive device predictive maintenance set of the target passive optical network.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images