Method and apparatus for monitoring optical path degradation in optical distribution network of passive optical network

Abstract

This invention relates to a method and apparatus for monitoring optical path degradation in a passive optical network (PON) optical distribution network, belonging to the field of passive optical network technology. The method includes: acquiring optical power time-series data of each ONU in the PON at a second-level cycle based on a telemetry protocol; performing trend degradation risk prediction and abnormal fluctuation detection on each ONU using a time-series prediction model and anomaly detection model; generating corresponding level degradation warning signals for ONUs experiencing degradation events based on a joint determination of the trend degradation risk prediction results and abnormal fluctuation detection results; and, in response to at least one degradation warning signal, determining the degradation segment in the ODN by statistically analyzing the degradation ratio of ONUs under each splitter node and their spatial distribution characteristics based on the resource tree topology of the PON, thereby achieving optical path degradation monitoring of the ODN. This invention can realize intelligent monitoring of optical paths in passive optical network optical distribution networks, from real-time monitoring and proactive warning to precise demarcation.

Description

Technical Field

[0001] This invention relates to the field of passive optical network technology, and in particular to a method and apparatus for monitoring optical path degradation in the optical distribution network of a passive optical network. Background Technology

[0002] Passive Optical Network (PON) is a point-to-multipoint (P2MP) single-fiber bidirectional optical access network. Its core characteristic is its passive nature; the Optical Distribution Network (ODN) contains no active electronic devices. Optical signals are distributed and aggregated through passive splitters, enabling a single backbone fiber to simultaneously provide high-speed data, voice, and video services to multiple end users. PON is the mainstream technology for achieving FTTx access, including Fiber to the Home (FTTH), Fiber to the Building (FTTB), Fiber to the Office (FTTO), and Fiber to the Room (FTTR).

[0003] A PON system consists of three parts: Optical Line Terminal (OLT), Optical Distribution Network (ODN), and Optical Network Unit / ONU / ONT. The OLT, located at the central office, is the core of the PON, responsible for connecting the service network and the ODN. The ODN is the optical channel connecting the OLT and ONU / ONT, composed of passive components (fiber optic cables, splitters, connectors, etc.). The ONU / ONT is located on the user side. In FTTH, FTTO, and FTTR networking scenarios, the ODN typically adopts a two-stage splitting structure, consisting of backbone optical cables, distribution optical cables, and drop cables sequentially from the central office to the user side.

[0004] However, current mainstream solutions for health monitoring of PON networks have significant shortcomings. First, existing monitoring systems primarily use Simple Network Management Protocol (SNMP) polling to collect information at the minute level, which cannot meet real-time monitoring requirements. For operator networks with a massive number of OLT and ONU devices, the minute-level collection cycle leads to discontinuous data and blind spots in network status monitoring. Second, existing monitoring systems can only detect and handle fault alarms triggered when the ODN optical path is interrupted or the received optical power is lower than the device's receiver sensitivity. However, when the optical power exhibits a continuous trend of degradation within the normal range, existing monitoring systems lack effective analytical models to detect it, thus failing to provide early warnings and proactive intervention before faults occur, often leading to unavoidable failures. Finally, even when optical path degradation is detected, existing monitoring systems cannot quickly and accurately determine whether the degradation point is located in the trunk cable, distribution cable, or drop cable, lacking the ability to spatially demarcate the degraded section. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method and device for monitoring optical path degradation in optical distribution networks in passive optical networks.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a method for monitoring optical path degradation in an optical distribution network in a passive optical network, employing the following technical solution: A method for monitoring optical path degradation in an optical distribution network in a passive optical network includes: Based on the telemetry protocol, optical power timing data of each optical network unit in the passive optical network are collected at a rate of seconds. Based on the optical power time-series data of each optical network unit, a time-series prediction model is used to predict the trend of degradation risk of each optical network unit; at the same time, based on the current optical power value of each optical network unit, an anomaly detection model is used to detect abnormal fluctuations. Based on the joint determination of trend-based degradation risk prediction results and abnormal fluctuation detection results, a corresponding level of degradation early warning signal is generated for the optical network unit where a degradation event has occurred. In response to at least one of the aforementioned degradation warning signals, based on the resource tree topology of the passive optical network, the degradation ratio of the optical network units under each splitter node is statistically analyzed and their spatial distribution characteristics are analyzed to determine the degradation segment in the optical distribution network, thereby realizing the monitoring of optical path degradation of the optical distribution network.

[0007] The beneficial effects of this invention are as follows: By employing a second-level telemetry protocol, continuous acquisition of optical power data from each optical unit in the network is achieved, overcoming the data lag inherent in traditional polling methods. Based on this, a hybrid intelligent analysis mechanism is constructed by combining time-series prediction models such as gated cyclic units with anomaly detection models such as isolated forests. This mechanism can proactively identify trend-based degradation risks and instantaneous abnormal fluctuations before optical power triggers a fault alarm, thus achieving a shift from passive alarm to proactive early warning. Furthermore, the method introduces a spatial clustering analysis algorithm based on resource tree topology. By statistically analyzing the degradation ratio and spatial distribution characteristics under the splitter node, it can accurately determine whether the degradation segment is located in the backbone, distribution cable, drop cable, or the splitter itself, achieving rapid fault location. Therefore, this invention can realize intelligent monitoring of the optical path of a passive optical network (PON) optical distribution network, from real-time monitoring and proactive early warning to precise demarcation.

[0008] Based on the above technical solution, the present invention can be further improved as follows.

[0009] Furthermore, the acquisition of optical power timing data of each optical network unit in the passive optical network at a second-level period based on the telemetry protocol includes: Send a subscription request based on a data modeling language definition to the optical line terminal in the passive optical network to establish a data acquisition channel; Through the data acquisition channel, the optical power timing data of each optical network unit actively pushed by the optical line terminal through at least one of its passive optical network ports at a second-level cycle is continuously received; wherein, the optical power timing data is encoded in binary encoding format and transmitted through a remote procedure call protocol that supports multiplexing, flow control and transport layer security encryption.

[0010] The beneficial effects of adopting the above-mentioned further solutions are as follows: By establishing a dedicated channel through subscription requests defined by the data modeling language, the required monitoring data content can be defined accurately and flexibly, avoiding redundant data transmission and improving the targeting and efficiency of data collection. Continuously receiving optical power time-series data from each optical network unit at a second-level cycle enables real-time, uninterrupted monitoring of the network status, allowing for timely detection of early signs of instantaneous anomalies or slow degradation, overcoming the monitoring blind spots of traditional minute-level polling. Simultaneously, using binary encoding to compress data significantly improves data transmission efficiency and reduces network bandwidth pressure. Furthermore, transmission based on a remote procedure call protocol that supports multiplexing, flow control, and transport layer security encryption not only ensures efficient and reliable data transmission during concurrent access of large-scale OLT devices, avoiding channel congestion and data loss, but also ensures the confidentiality and integrity of sensitive monitoring data during transmission through encryption mechanisms, preventing eavesdropping or tampering. This provides a high-quality, reliable data foundation for subsequent intelligent analysis and early warning decisions, fundamentally strengthening the real-time performance, stability, and security of the entire monitoring system.

[0011] Furthermore, the time-series prediction model is a gated cyclic unit model; The method of predicting the trend degradation risk of each optical network unit based on the optical power time series data of each optical network unit through a time series prediction model includes: For each optical network unit, the optical power time-series data of the optical network unit within the historical time window, which has undergone data cleaning and normalization, is input into the gated cyclic unit model to obtain the optical power prediction value of the optical network unit at a specified future time point. Calculate the decrease in the predicted optical power value relative to the current actual optical power value; If the decline exceeds a preset degradation threshold, the optical network unit is determined to have a trend of degradation risk.

[0012] The beneficial effects of adopting the above-mentioned further scheme are as follows: By inputting cleaned and normalized optical power time-series data reflecting historical variation patterns into the gated cyclic unit model, the model's ability to capture time-series dependencies can be leveraged to make scientific predictions of optical power values ​​at a specified future time point. By comparing the predicted optical power value with the current actual optical power value, the potential decline rate is calculated and judged against a preset degradation threshold. This allows for the proactive identification of optical network units at risk of continuous degradation even before the optical power value falls below the device's receiving sensitivity, i.e., when the network is still operating normally. This method overcomes the passive situation of traditional monitoring that only alarms after a fault occurs, achieving proactive detection of trend-based degradation of the optical path.

[0013] Furthermore, the gated recurrent unit model is obtained through the following steps: Historical optical power time-series data of several optical network units under normal and degraded conditions are extracted from the historical database. Each historical optical power time-series data is divided into multiple samples according to a fixed time window. The actual optical power values ​​of each sample at multiple different future time points are used as the label set of the sample to form a labeled sample set. The labeled sample set is divided into a training set and a validation set; The training set is input into the initial gated recurrent unit model, and an adaptive learning rate optimizer is used to perform iterative training with the goal of minimizing the error between the predicted optical power value output by the model and the actual optical power value at the corresponding future time point. After each round of training, the validation loss is calculated using the validation set; The changes in the verification loss are monitored. When the verification loss no longer decreases within a preset number of consecutive rounds, an early stopping mechanism is triggered to terminate the training process and obtain the gated recurrent unit model.

[0014] The beneficial effects of adopting the above-mentioned further approach are as follows: By extracting samples from historical optical power data containing both normal and deteriorated states and constructing a multi-time-point label set, the model can learn the change patterns of optical power under various health states, including the complete evolution process before and after degradation. Dividing the dataset into training and validation sets avoids overfitting the model to the training data, ensuring its predictive stability on unseen data. Iterative training using an adaptive learning rate optimizer can dynamically adjust the learning step size, accelerate model convergence, and improve training efficiency. By calculating the loss using the validation set after each training round and monitoring its changes, an early stopping mechanism can be triggered in a timely manner when the model performance reaches its optimal level, preventing performance degradation caused by overtraining, thereby obtaining a final model with stronger generalization ability and more reliable predictions.

[0015] Furthermore, the anomaly detection model is an isolated forest model; The abnormal fluctuation detection based on the current optical power value of each optical network unit, using an anomaly detection model, includes: For each optical network unit, a multidimensional feature vector is constructed based on the optical power value of the optical network unit at the current moment and multiple derived features; wherein, the multiple derived features are used to characterize the instantaneous rate of change, fluctuation, and long-term trend of optical power; The multidimensional feature vector is input into the isolated forest model, and the anomaly score is calculated based on the average path length of the multidimensional feature vector in all isolated trees. If the abnormal score exceeds the abnormal detection threshold, it is determined that the optical network unit has abnormal fluctuations at the current moment.

[0016] The advantages of adopting the above-mentioned further approach are as follows: Utilizing the unsupervised learning model of Isolation Forest eliminates the need for training with a large number of labeled anomalous samples; a detection benchmark can be established using only normal historical data, reducing the difficulty and cost of model construction. By constructing a multi-dimensional feature vector for each optical network unit, containing the current optical power value and multiple derived features (such as instantaneous rate of change, fluctuation statistics, and long-term trend slope), the dynamic behavior of optical power can be comprehensively characterized from different dimensions, enabling the model to identify complex anomalous patterns that cannot be reflected by a single indicator. After inputting the feature vector into the Isolation Forest model, an anomalous score is calculated based on its average path length across all isolated trees. This score quantifies the degree of deviation of the current data from the "normal" cluster. By comparing the score with a preset threshold, the occurrence of instantaneous anomalous fluctuations can be objectively and automatically determined. This method effectively supplements the shortcomings of trend prediction, achieving rapid and accurate detection of optical path mutations.

[0017] Furthermore, the isolated forest model is trained through the following steps: Construct a multidimensional feature vector set for several optical network units under normal conditions; Multiple multidimensional feature vectors are randomly selected from the multidimensional feature vector set to form a positive sample set; Under the constraints of a preset number of subtrees and a preset number of samples extracted from each subtree, multiple isolated trees are generated based on the positive sample set, and the forest composed of the multiple isolated trees is used as the isolated forest model.

[0018] The beneficial effects of adopting the above-mentioned further scheme are as follows: By collecting a large amount of operational data of optical network units under normal conditions and extracting multi-dimensional features from it, a feature vector set that can accurately represent the baseline of network health is constructed. Samples are randomly drawn from this set to form a training set, ensuring that the model learns the typical statistical distribution of normal optical power behavior. During training, the complexity and detection capability of the model are controlled by a preset number of subtrees, and each subtree is generated based on a limited number of randomly drawn samples, introducing necessary randomness. This effectively captures the commonalities of normal data while preventing the model from being overly sensitive to specific noise. Finally, a forest composed of multiple such isolated trees constitutes an isolated forest model. This isolated forest model does not require anomaly labels and can learn to distinguish between normal and special data solely from normal data, enabling it to accurately identify abnormal fluctuations deviating from normal patterns in practical applications.

[0019] Furthermore, the multiple derived features include: first-order difference features, sliding window statistical features, and trend slope features; The multiple derived features are calculated using the following steps: The difference in optical power between adjacent time points is calculated to obtain the first-order difference feature; The mean and standard deviation are calculated within a preset sliding window to obtain the statistical characteristics of the sliding window; Linear fitting is performed on the data within the historical time window to obtain the trend slope characteristics.

[0020] The beneficial effects of adopting the above-mentioned further scheme are as follows: Calculating the difference in optical power between adjacent time points yields a first-order difference feature that directly quantifies the instantaneous rate of change of the signal, enabling it to respond sensitively to sudden jumps in optical power. Calculating the mean and standard deviation within a preset sliding window yields a sliding window statistical feature that statistically characterizes the average level and fluctuation range of the signal in the recent period, helping to detect amplitude anomalies or decreased stability. Linear fitting of data within historical time windows yields a trend slope feature that reveals the potential long-term evolution direction of the signal, identifying slow but continuous degradation trends. Constructing a multi-dimensional feature vector with these features and the current optical power value provides rich and structured input information for the isolated forest model. This allows the model to not only perceive anomalies in the current value but also to make a comprehensive judgment by combining the rate of change, fluctuation, and long-term trends, thereby more accurately and comprehensively identifying complex or hidden abnormal fluctuation patterns, greatly improving the accuracy and robustness of anomaly detection.

[0021] Furthermore, the joint determination based on the trend-based degradation risk prediction results and the abnormal fluctuation detection results generates a corresponding level of degradation early warning signal for the optical network unit where a degradation event has occurred, including: For each optical network unit that experiences a degradation event, the magnitude and / or duration of future optical power decrease are determined based on the trend degradation risk prediction results, and the number of times the abnormal score exceeds a preset abnormal score threshold is determined based on the abnormal fluctuation detection results. If the decrease exceeds a preset magnitude threshold or the abnormal score exceeds a preset abnormal score threshold, a first-level degradation warning signal is generated. If the duration exceeds a preset duration threshold or the number of occurrences reaches a first threshold within a first preset time period, a second-level warning signal is generated. If the future optical power is lower than the receiving sensitivity threshold or the number of times reaches the second threshold within the second preset time period, a third-level warning signal is generated. The warning signal level is directly proportional to the severity of degradation, the second preset time period is shorter than the first preset time period, and the second number threshold is lower than the first number threshold.

[0022] The beneficial effects of adopting the above-mentioned further scheme are as follows: The joint judgment and hierarchical early warning scheme enables refined and hierarchical intelligent early warning of optical path degradation events in passive optical networks, effectively improving the pertinence and response efficiency of operation and maintenance decisions. Specifically, by combining specific indicators such as the magnitude and duration of future optical power decline and the frequency of anomalies, degradation scenarios with different levels of urgency and impact can be distinguished: slight trend deviations or single anomalies trigger lower-level warnings, drawing attention; continuous degradation or frequent anomalies within a certain period trigger intermediate-level warnings, indicating risk accumulation; and predicted optical power falling below receiver sensitivity or dense anomalies within a very short period trigger the highest-level warning, indicating an impending or ongoing service interruption. This hierarchical mechanism ensures that early warning information is no longer simply a matter of presence or absence of an alarm, but rather includes rich semantics encompassing severity and development trends. Operations and maintenance personnel can quickly determine the priority of handling based on the warning level and allocate resources reasonably, thereby realizing the upgrade of the operation and maintenance mode from uniform alarms to differentiated responses. This can avoid over-responding to low-risk events and ensure the timely handling of high-risk events. While ensuring network reliability, it significantly improves the accuracy and economy of operation and maintenance work.

[0023] Furthermore, the step of determining the degraded segments in the optical distribution network by statistically analyzing the degradation ratio of the optical network units under each splitter node and their spatial distribution characteristics includes: For each optical splitter node, the proportion of the number of degraded optical network units within its subordinate topology range to the total number of optical network units under the optical splitter node is calculated to obtain the degradation ratio. Based on the resource tree topology and the degradation ratio of each of the optical splitter nodes, the type of degradation segment in the optical distribution network is determined by the following method: If the degradation ratio of all secondary optical splitter nodes under a certain primary optical splitter node exceeds a preset ratio threshold, then the type of the degradation segment is determined to be the trunk optical cable connected upstream of the primary optical splitter node. If the degradation ratio of each of the secondary optical splitter nodes under a certain primary optical splitter node exceeds the preset ratio threshold, then the type of the degradation segment is determined to be the primary optical splitter node or the optical cable between the primary optical splitter node and the secondary optical splitter nodes; the secondary optical splitter nodes include multiple secondary optical splitter nodes. If the degradation ratio of a single secondary beam splitter node exceeds the preset ratio threshold, then the type of the degradation segment is determined to be that of the secondary beam splitter node. If only a single optical network unit under a single secondary splitter node degrades, the type of the degraded segment is determined to be the drop cable connected to that optical network unit or the optical network unit device.

[0024] The beneficial effects of adopting the above-mentioned further solutions are as follows: when all secondary optical splitters under the primary optical splitter are extensively degraded, the upstream common trunk optical cable can be identified as faulty; when only some secondary optical splitters are degraded, the problem may lie in the primary optical splitter itself or the distribution optical cable connecting these splitters; when the degraded problem is strictly limited to a single secondary optical splitter, the fault can be located in that splitter device; when the degraded problem occurs only in a very few users under a single secondary optical splitter, the fault can be further located in the drop cable connecting that user or the user equipment itself. This reasoning based on topology and statistics narrows the ambiguous fault range to specific segments or devices, providing clear and direct maintenance targets for operation and maintenance, avoiding blind troubleshooting, and greatly shortening the location and repair time.

[0025] Furthermore, after identifying the degraded segments in the optical distribution network, the method further includes: In the graphical representation of the resource tree topology, different colors are used to render the optical line terminal ports, splitter nodes and connecting optical cables involved in the degraded segments according to the type, location and corresponding degraded warning signal level of the degraded segments, so as to generate a topology heat map. For optical network units that have experienced degradation events, a comparison curve is generated between their historical optical power time-series data and the future optical power trend predicted by the time-series prediction model. The topological heatmap and the comparison curve are visualized.

[0026] The beneficial effects of adopting the above-mentioned further solutions are as follows: At the macro level, by generating a topology heatmap, maintenance personnel can clearly grasp the overall health status of the entire network in a single image, quickly locate the location of degradation and its severity, achieving a comprehensive view of the entire network from a single image, and providing a global perspective for resource scheduling and emergency command. At the micro level, by generating a comparison curve of historical and predicted optical power, specific and quantitative diagnostic basis can be provided for the degradation of individual optical network units, facilitating in-depth analysis of the root causes and development trends of problems. This presentation method combining macro and micro perspectives not only lowers the technical threshold for maintenance analysis but also makes fault location, cause analysis, and handling decisions more accurate and efficient, ultimately truly transforming the results of intelligent monitoring into an improvement in maintenance productivity.

[0027] Secondly, this invention provides a device for monitoring optical path degradation in a passive optical network's optical distribution network, employing the following technical solution: A device for monitoring optical path degradation in a passive optical network (PON) distribution network includes: The acquisition module is used to acquire optical power timing data of each optical network unit in the passive optical network at a rate of seconds, based on the telemetry protocol. The detection module is used to predict the trend of degradation risk of each optical network unit based on the optical power time series data of each optical network unit through a time series prediction model; at the same time, it detects abnormal fluctuations based on the current optical power value of each optical network unit through an anomaly detection model. The early warning module is used to generate a corresponding level of degradation early warning signal for the optical network unit that has experienced a degradation event, based on the joint determination of the trend degradation risk prediction results and the abnormal fluctuation detection results. The delimitation module is used to respond to at least one of the degradation warning signals, and based on the resource tree topology of the passive optical network, determine the degradation segment in the optical distribution network by statistically analyzing the degradation ratio of the optical network units under each splitter node and analyzing their spatial distribution characteristics, so as to realize the monitoring of optical path degradation of the optical distribution network.

[0028] Thirdly, the present invention provides an electronic device that adopts the following technical solution: An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the optical path degradation monitoring method for optical distribution networks in a passive optical network as described in any of the first aspects.

[0029] Fourthly, the present invention provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the optical path degradation monitoring method for optical distribution networks in a passive optical network as described in any of the first aspects.

[0030] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and will become apparent from the description or may be learned by practice of the invention. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the optical path degradation monitoring method for optical distribution networks in passive optical networks provided by the present invention. Figure 2 This is a schematic diagram of the optical path degradation monitoring device for optical distribution network in a passive optical network provided by the present invention. Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0032] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0033] Please refer to Figure 1 , Figure 1This is a flowchart illustrating the optical path degradation monitoring method for optical distribution networks in passive optical networks provided by the present invention. Figure 1 As shown, the method may include the following steps S101-S104.

[0034] S101. Based on the telemetry protocol, acquire the optical power timing data of each optical network unit in the passive optical network at a second-level cycle.

[0035] Specifically, a proactive telemetry protocol is used to collect optical power timing data of each ONU / ONT connected to the OLT in the PON at a high frequency with a second-level period (e.g., 1-5 seconds). The optical power timing data refers to a continuous data sequence composed of the real-time received optical power values ​​at the receiver (Rx) of each optical network unit arranged in chronological order, and is a core performance indicator for measuring the downlink optical path transmission quality of the ODN. By using a telemetry protocol to achieve second-level periodic acquisition, the monitoring blind spot problem caused by traditional SNMP polling is solved.

[0036] In one embodiment, step S101 may include: sending a subscription request based on a data modeling language definition to an optical line terminal in a passive optical network to establish a data acquisition channel; continuously receiving optical power timing data of each optical network unit actively pushed by the optical line terminal through at least one of its passive optical network ports at a second-level cycle through the data acquisition channel; wherein the optical power timing data is encoded in a binary encoding format and transmitted through a remote procedure call protocol that supports multiplexing, flow control and transport layer security encryption.

[0037] Specifically, a subscription request based on a data modeling language (such as YANG) is first sent to the OLT in the PON to explicitly specify the data items to be collected (such as the ONU / ONT device ID, the PON port it belongs to, the uplink / downlink real-time received optical power, the timestamp, and the optical module type (such as GPON, XGS-PON), etc.), thereby establishing a long-term connection channel for data collection. Subsequently, the OLT, acting as a data pusher, responds to this subscription request and actively and continuously pushes the optical power timing data of each ONU / ONT connected to it through one or more of its PON ports at a second-level period (e.g., 1-5 seconds). To ensure the efficiency, reliability, and security of massive, high-frequency data during transmission, the optical power timing data is serialized and encoded using an efficient binary encoding format (such as GPB) before transmission and transmitted based on a remote procedure call protocol (such as gRPC / HTTP2) that supports multiplexing, flow control, and transport layer security (TLS) encryption. By combining this "subscription-push" mechanism with a high-performance protocol stack, it is possible to stably and in real-time acquire second-level optical power data concurrently from a massive number of devices in a large-scale network.

[0038] S102. Based on the optical power time series data of each optical network unit, a time series prediction model is used to predict the trend of degradation risk of each optical network unit; at the same time, based on the current optical power value of each optical network unit, an anomaly detection model is used to detect abnormal fluctuations.

[0039] Specifically, based on the optical power time-series data of each optical network unit, a time-series prediction model can capture the long-term dependencies and variation patterns implicit in the optical power time-series data, thereby predicting whether each optical network unit has a trend of degradation risk. An anomaly detection model detects whether the current optical power value of each optical network unit experiences abnormal fluctuations. Through trend degradation risk prediction and abnormal fluctuation detection, a comprehensive and three-dimensional perception of the optical path health status, from long-term gradual degradation to instantaneous sudden fluctuations, can be achieved.

[0040] In one embodiment, the optical power timing data of each optical network unit is preprocessed through the following steps: 1) Data cleaning Outlier removal: The optical power time series data is scanned using statistical criteria (e.g., the improved 3σ criterion based on the mean and standard deviation); data points that exceed the mean ± 3 times the standard deviation are marked, as these marked data points are usually caused by instantaneous measurement noise, equipment malfunction, or transmission errors; if three consecutive sampling points are outliers, they are removed. Missing value handling: For single missing values ​​that may exist in the optical power time series data, linear interpolation is used to calculate and fill in the missing value using the effective optical power values ​​before and after the missing point to ensure the continuity and integrity of the time series data; if more than 5 consecutive sampling points are missing, the data for that time period is marked as invalid. Unit standardization: All optical power values ​​in the optical power time series data are uniformly converted to dBm units, with a power range typically from -40dBm to +8dBm; Time alignment: Based on the timestamps synchronized by the Network Time Protocol (NTP), the optical power data of different OLTs and different PON ports are aligned to a unified time axis; 2) Data normalization Since the nominal values ​​and dynamic ranges of optical power of optical network units from different manufacturers and models vary, in order to eliminate the influence of dimensions, accelerate model training convergence, and improve the model's generalization ability to different devices, the Min-Max normalization method is used to map each optical power value in the optical power time series data to a unified standard interval (e.g., the [0,1] interval): (1) in, Set to -40dBm (below the receiver sensitivity). Take +8dBm (overload optical power). This represents the optical power value to be normalized. This represents the optical power value after normalization. This step transforms all optical power data to the same numerical scale, enabling subsequent models to learn and capture the inherent patterns of change in the data more efficiently and stably, rather than being affected by the absolute magnitude of the original values. 3) Feature Engineering To enhance the anomaly detection model's recognition capability, the input to the anomaly detection model includes not only the current optical power value x_t, but also several derived features; among these, the derived features include: first-order difference features, sliding window statistical features, and trend slope features. Calculate multiple derived features using the following steps: The difference in optical power between adjacent time points is calculated to obtain the first-order difference feature, namely x_t-x_t-1, which is used to characterize the instantaneous rate of change of optical power. Within a preset sliding window, the mean (e.g., the average optical power over the past 10 seconds: mean(x_t-9,...,x_t)) and standard deviation (e.g., the standard deviation of optical power over the past 10 seconds: std(x_t-9,...,x_t)) are calculated to obtain the statistical characteristics of the sliding window, which are used to reflect the average level and fluctuation of optical power in recent times. Linear fitting is performed on data within a historical time window (such as the most recent 60 seconds, or a window configured according to the actual model) to obtain trend slope features. For example, by performing linear fitting on x_t-59 to x_t, the trend slope value k of the past 60 seconds is obtained, which is used to characterize the long-term evolution direction of optical power.

[0041] In one embodiment, the timing prediction model is a gated recurrent unit (GRU) model. Step S102, based on the optical power timing data of each optical network unit, uses the timing prediction model to predict the trend degradation risk of each optical network unit, including: for each optical network unit, inputting the optical power timing data of the optical network unit within the historical time window, which has undergone data cleaning and normalization, into the gated recurrent unit model to obtain the optical power prediction value of the optical network unit at a specified future time point; calculating the decrease of the optical power prediction value relative to the current actual optical power value; if the decrease exceeds a preset degradation threshold, it is determined that the optical network unit has a trend degradation risk.

[0042] Specifically, the time series prediction model selected in this embodiment is the GRU model. The GRU model is a special type of recurrent neural network. Because it has update gates and reset gates, it can effectively capture long-term dependencies in time series data, making it very suitable for processing continuous signals that change sequentially over time, such as optical power time series data.

[0043] The network architecture of the GRU model includes: • Input layer: The input of the GRU model is the optical power time series data of a fixed historical time window (e.g., 60 seconds), denoted as X={x_{t-59}, x_{t-58}, ..., x_t}. Each time point t corresponds to an optical power value (unit: dBm), and the data point interval is 1 second. Therefore, the length of the input sequence is 60. • GRU Hidden Layers: A two-layer stacked bidirectional GRU network is used. This structure allows the network to learn contextual features of time-series data simultaneously from both forward (past to future) and backward (future to past) directions, thereby capturing the patterns of optical power variation more comprehensively. The specific parameters are: the first bidirectional GRU layer contains 128 hidden units, and the second layer contains 64 hidden units. To prevent overfitting, a dropout layer with a dropout rate of 0.2 is connected after each GRU layer. • Output layer: The output of the last GRU layer is flattened and connected to a fully connected layer; this fully connected layer contains a neuron and uses a linear activation function, and its output is the predicted optical power value in the kth second of the future.

[0044] For each optical network unit (ONU), the optical power time-series data of that ONU within the most recent 60 seconds, after data cleaning and normalization, is input into the GRU model. Based on the historical patterns and variation laws it has learned, the GRU model outputs a predicted optical power value P_pred for that ONU at a specified future time point (e.g., 24 hours later). This predicted optical power value represents the likely level of future optical power assuming the current trend continues.

[0045] After obtaining the predicted optical power value, calculate the decrease in predicted optical power P_pred relative to the current actual optical power value P_current, ΔP. ΔP = P_current - P_pred. The magnitude of ΔP directly reflects the severity of the predicted deterioration in optical power.

[0046] The calculated degradation magnitude ΔP is compared with a preset degradation threshold. This preset degradation threshold is a threshold value (e.g., 5 dBm) set through business verification and model optimization. If the degradation magnitude exceeds the preset degradation threshold, the optical network unit is determined to have a trend degradation risk.

[0047] In this way, even if the current actual optical power value is completely within the normal operating range of the optical network unit and no traditional threshold alarm is triggered, the GRU model can predict that significant degradation will occur in the future based on the optical power time series data of the optical network unit. This allows the GRU model to determine that the optical network unit has a trend of degradation risk before physical failure or communication interruption actually occurs. This solves the technical problem that traditional monitoring methods cannot detect continuous degradation within the normal range.

[0048] In one embodiment, the gated recurrent unit model is trained through the following steps: extracting historical optical power time-series data of several optical network units under normal and degraded states from a historical database; dividing each historical optical power time-series data into multiple samples according to a fixed time window; and using the actual optical power values ​​of each sample at multiple different future time points as the label set of the samples to form a labeled sample set; dividing the labeled sample set into a training set and a validation set; inputting the training set into the initial gated recurrent unit model, and using an adaptive learning rate optimizer to perform iterative training with the goal of minimizing the error between the predicted optical power value output by the model and the actual optical power value at the corresponding future time point; calculating the validation loss using the validation set after each round of training; monitoring the change in the validation loss, and triggering an early stopping mechanism when the validation loss no longer decreases within a preset number of consecutive rounds to terminate the training process, thus obtaining the gated recurrent unit model.

[0049] Specifically, firstly, a large amount of long-sequence optical power data of optical network units under normal and degraded states is extracted from historical databases. This allows the GRU model to learn not only the fluctuation patterns under normal conditions but also the signs of impending degradation. Then, each long time-series data point is divided into sliding window segments according to a fixed time window (e.g., 60 seconds). For example, a 24-hour time-series data point, slid through a 60-second window with a 1-second step, can generate 86,341 independent samples. The input feature for each sample is the optical power time-series data of these 60 consecutive time points.

[0050] To enable the GRU model to predict the future, a "correct answer" (i.e., label) needs to be defined for each sample. This embodiment employs a multi-task learning or multi-step prediction strategy, defining multiple sets of actual optical power values ​​at different future time points as label sets for each sample. For example, for a 60-second sample ending at time t, its label set can be defined as the actual optical power values ​​at future times t+1, t+5, t+60, and up to t+86400 (24 hours later). In this way, one sample corresponds to multiple prediction tasks at future times, forcing the model to learn more general temporal evolution patterns. Ultimately, all samples and their corresponding label sets constitute a labeled sample set.

[0051] The labeled sample set is randomly divided into a training set (e.g., 80%) and a validation set (e.g., 20%). The training set is used to directly adjust the model parameters of the GRU model, while the validation set does not participate in model parameter updates and is only used to evaluate the generalization ability of the GRU model after each round of training to prevent overfitting.

[0052] An initial GRU model is constructed. At this stage, the parameters of the initial GRU model are randomly or initialized according to specific rules and lack predictive ability. A 60-second time-series optical power dataset is input into the initial GRU model to predict the optical power values ​​at each future time point. The difference between all optical power predictions and their corresponding true labels is calculated, commonly using mean squared error as the loss function. The gradient of the loss function with respect to all model parameters is calculated. Training is performed using an adaptive learning rate optimizer (e.g., the Adam optimizer), with an initial learning rate of 0.001, a batch size of 64, and 100 training epochs. The Adam optimizer dynamically adjusts the learning rate for each parameter based on the first and second moments of the gradient, thus updating the model parameters more efficiently and stably to reduce loss. This process is repeated on all training data (i.e., multiple epochs). After each training epoch, a validation loss is calculated using a validation set that has not participated in parameter updates. The validation loss reflects the model's performance on unseen data. The validation loss is continuously monitored. When the validation loss stops decreasing within a preset number of consecutive epochs (e.g., 10 consecutive epochs), the model's generalization ability is considered to have stopped improving. At this point, an early stopping mechanism is triggered, immediately terminating the training process. The model parameters at the time of training termination are saved as the final trained, optimal-performing GRU model. This ensures that the model is both adequately trained and avoids overfitting.

[0053] In one embodiment, the anomaly detection model is an isolated forest model; step S102, based on the current optical power value of each optical network unit, performs anomaly fluctuation detection through the anomaly detection model, including: for each optical network unit, constructing a multidimensional feature vector based on the optical power value of the optical network unit at the current moment and multiple derived features; wherein, the multiple derived features are used to characterize the instantaneous rate of change, fluctuation, and long-term trend of optical power; inputting the multidimensional feature vector into the isolated forest model, calculating anomaly score based on the average path length of the multidimensional feature vector in all isolated trees; if the anomaly score exceeds the anomaly detection threshold, it is determined that the optical network unit has anomaly fluctuation at the current moment.

[0054] Specifically, in order to more comprehensively characterize the dynamic behavior of optical power, a multidimensional feature vector is constructed for each optical network unit at the current time t. The multidimensional feature vector includes the optical power value x_t of the optical network unit at the current time t and multiple derived features.

[0055] Among them, multiple derived features may include: First-order difference feature characterizing instantaneous change rate: Calculate the first-order difference feature, i.e., x_t-x_{t-1}; this feature directly reflects the instantaneous change of optical power between adjacent sampling points. Positive values ​​indicate an increase, negative values ​​indicate a decrease, and the absolute value represents the degree of drastic change, which can effectively capture sudden jumps. Sliding window statistical characteristics characterizing fluctuations: Select a recent time window (e.g., the past 10 seconds) and calculate the mean and standard deviation of optical power within that window; the mean reflects the recent average level, while the standard deviation quantifies the fluctuation range around that level; stable optical power has a smaller standard deviation, while frequently fluctuating optical power has a larger standard deviation; Trend slope feature characterizing long-term trends: A trend line is obtained by linearly fitting the light power values ​​over a relatively long historical time window (e.g., the past 60 seconds). The slope of this trend line is the trend slope feature. A positive slope indicates a long-term upward trend, while a negative slope indicates a long-term downward trend. The absolute value of the slope reflects the strength of the trend, which helps to identify slow but persistent deviations.

[0056] For example, suppose that at time t, the current optical power of a certain ONU is -15.3 dBm. Simultaneously calculate: the difference over the past second (e.g., -0.2 dBm), the mean (e.g., -15.1 dBm) and standard deviation (e.g., 0.5 dBm) over the past 10-second window, and the linear fitting slope of the data over the past 60 seconds (e.g., -0.01 dBm / second). These five values ​​(current optical power, first-order difference, mean, standard deviation, and slope) together constitute a 5-dimensional feature vector, which serves as the input to the isolated forest model.

[0057] The constructed multidimensional feature vector is input into the isolated forest model, which consists of a large number (e.g., 100) of randomly generated isolated trees. Each isolated tree is constructed during the training phase by randomly selecting features and split points.

[0058] On each isolated tree, starting from the root node, the value of a feature in the feature vector is randomly compared with the split threshold of the tree node to determine whether it is assigned to the left or right subtree. This process is repeated recursively until the feature vector is isolated to a leaf node. The number of edges traversed from the root node to that leaf node is called the path length. Outliers, because their feature values ​​deviate from the normal cluster, are usually isolated with only a few comparisons, and therefore have shorter path lengths.

[0059] Based on the average path length of this feature vector across all isolated trees, it is mapped to an outlier score s between 0 and 1 using a normalization formula. The closer s is to 1, the shorter the average path length of the sample across all trees, meaning it is more likely to be isolated and therefore more probable as an outlier.

[0060] The calculated anomaly score is compared with a preset anomaly detection threshold (e.g., s_th = 0.6). If the anomaly score exceeds this threshold, it is determined that the optical network unit has an abnormal fluctuation at the current time t, and instantaneous sudden fluctuations of the optical network unit can be detected.

[0061] In one embodiment, the isolated forest model is trained by the following steps: constructing a multidimensional feature vector set of several optical network units under normal conditions; randomly extracting multiple multidimensional feature vectors from the multidimensional feature vector set as a positive sample set; under the constraints of a preset number of subtrees and a preset number of samples extracted from each subtree, generating multiple isolated trees based on the positive sample set, and using the forest composed of multiple isolated trees as the isolated forest model.

[0062] Specifically, historical optical power time-series data of several optical network units under known normal conditions (i.e., no alarms, no faults) are extracted from the historical operational database. For each time-series data point under normal conditions, a corresponding multidimensional feature vector is constructed according to the method described in the previous embodiment (calculating the current power value, first-order difference, sliding window statistics, trend slope, etc.). The original one-dimensional optical power sequence can be transformed into a multidimensional feature vector that can characterize its dynamic changes. The feature vectors of all normal moments of all normal optical network units are summarized to form a multidimensional feature vector set. This set represents the data distribution pattern that the network should have under healthy conditions. From the above multidimensional feature vector set, a certain number (e.g., 10,000) of multidimensional feature vectors are randomly selected to form the positive sample set used in this training. Random sampling ensures the representativeness of the training data, and by controlling the number of samples, the training efficiency and generalization ability of the model can be balanced. Under the constraints of a preset number of subtrees (e.g., n_estimators=100) and a preset number of samples extracted from each subtree (e.g., max_samples=256), the following loop is executed: 1) Subsampling: Randomly extract max_samples feature vectors from the positive sample set (with or without replacement); 2) Tree building: Using this batch of sampled data as the root node, recursively build an isolated tree; the tree building rule is: randomly select a feature dimension, then randomly select a split value to divide the data into left and right subtrees; recursively continue until the stopping condition is met (e.g., the tree reaches its maximum depth, or only one sample remains in a node); 3) Aggregation: Repeat the above process n_estimators times to generate multiple independent isolated trees. Finally, the set of these isolated trees constitutes the trained isolated forest model.

[0063] Through the above process, the model learns the clustered distribution of normal data in the feature space. Since the training data is all normal, any data point that deviates from this clustered region in the feature space and is easily isolated by a few random splits (i.e., short path lengths) will be judged as anomaly during the inference phase.

[0064] S103. Based on the joint determination of trend-based degradation risk prediction results and abnormal fluctuation detection results, generate corresponding level degradation warning signals for optical network units that have experienced degradation events.

[0065] In one embodiment, step S103 includes the following sub-steps S1031-S1034.

[0066] S1031. For each optical network unit that has experienced a degradation event, determine the magnitude and / or duration of the future decrease in optical power based on the trend degradation risk prediction results, and determine the number of times the abnormal score exceeds the preset abnormal score threshold based on the abnormal fluctuation detection results.

[0067] Specifically, two core quantitative indicators are extracted from the analysis results of the time-series prediction model. The first is the magnitude of the future optical power decline, i.e., the amount by which the predicted future optical power value decreases relative to the current actual optical power value (e.g., a predicted decrease of 5 dBm after 24 hours). This is a spatial indicator for assessing the severity of the potential impact. The second is the duration, i.e., the length of time the optical power decline trend identified by the model has continued (e.g., a continuous decline exceeding 12 hours). This is a time indicator for assessing the degradation process. These two indicators together characterize the intensity and inertia of the trend-based degradation.

[0068] A frequency index is extracted from the analysis results of the anomaly detection model. This involves counting the total number of times the anomaly score of an optical network unit exceeds a preset anomaly detection threshold (e.g., s > 0.6) within a specified historical time window. This number reflects the frequency of drastic or irregular fluctuations in optical power in the recent period and serves as a dynamic indicator for assessing network stability.

[0069] S1032. If the decrease exceeds the preset magnitude threshold or the abnormal score exceeds the preset abnormal score threshold, a first-level deterioration warning signal is generated.

[0070] Specifically, this sub-step defines the triggering conditions for the first-level degradation warning signal (i.e., the yellow warning signal). Its decision logic reflects the principle that triggering occurs when a single indicator exceeds its limit, thus attracting the initial attention of operations and maintenance personnel with a relatively low threshold. There are two triggering conditions; either one needs to be met: Condition 1: Amplitude Exceeds Limit. When the predicted decrease in future optical power calculated from the trend forecast results (e.g., a predicted decrease of more than 5 dBm) exceeds the preset amplitude threshold, it means that even if the current network is still operational, there is a clear potential risk that may lead to a significant decline in future performance, and it needs to be included in monitoring.

[0071] Condition 2: Single Anomaly. If the anomaly score at the current moment exceeds the preset anomaly detection threshold, it indicates that the optical power has experienced a significant instantaneous fluctuation or abnormal pattern at that moment. Even without a clear trend of degradation, this instantaneous anomaly may be an early sign of potential network vulnerabilities.

[0072] Therefore, the purpose of the first-level warning is to alert and identify optical network units that have shown clear risk signals but have not yet posed an imminent threat.

[0073] S1033. If the duration exceeds the preset duration threshold or the number of occurrences reaches the first threshold within the first preset time period, a second-level warning signal is generated.

[0074] Specifically, this sub-step defines the triggering conditions for the second-level warning signal (i.e., the orange warning signal). Its decision logic upgrades from single-point triggering to continuous or frequent triggering, meaning the deteriorating state is intensifying or persisting. There are also two triggering conditions: Condition 1: The trend continues. When the duration of the trend of deterioration (e.g., it has lasted for more than 12 hours) exceeds the preset duration threshold, it indicates that the decrease in optical power is not a short-term fluctuation, but a continuous process that is increasing the likelihood of it evolving into a failure, and the risk level needs to be raised.

[0075] Condition 2: Frequent Anomalies. Within a first, relatively long time period (e.g., a continuous 30 minutes), the total number of times the anomaly score exceeds the threshold reaches a high initial threshold (e.g., 15 times). This means that the anomaly fluctuations are not isolated events, but occur repeatedly and frequently over a period of time, reflecting an increased degree of instability in the network state.

[0076] The purpose of the Level 2 warning is to identify optical network units where risks are accumulating, their condition remains poor, or they are fluctuating frequently, indicating the need to prepare for more in-depth inspections or develop contingency plans.

[0077] S1034. If the future optical power is lower than the receiving sensitivity threshold or the number of times reaches the second threshold within the second preset time period, a third-level warning signal will be generated.

[0078] Specifically, this sub-step defines the triggering conditions for the third-level warning signal (i.e., the red warning signal). Its decision-making logic targets the most urgent situations that are most likely to cause immediate business disruption. There are also two triggering conditions: Condition 1: Predicted Outage. This occurs when the predicted future optical power value (e.g., 24 hours later) is expected to fall below the receiver sensitivity threshold of the optical network unit. This is the most direct warning of service interruption risk, meaning that if the current trend continues, the equipment will be unable to function properly due to excessively low received optical power, requiring immediate intervention.

[0079] Condition 2: Frequent anomalies (high intensity). Within a shorter second preset time period (e.g., within a consecutive 10 minutes), the total number of times the anomaly score exceeds the threshold reaches a higher second threshold (e.g., 8 times). This combination of a shorter time period and a higher threshold indicates that extremely dense anomalies have occurred in a very short period of time, which is a typical characteristic of a network that is about to or is experiencing a severe failure (such as link interruption or strong interference).

[0080] The third-level warning serves as an emergency alert, identifying optical network units facing imminent business interruption risks or experiencing serious failures, requiring maintenance personnel to immediately carry out emergency repairs.

[0081] S104. In response to at least one degradation warning signal, based on the resource tree topology of the passive optical network, by statistically analyzing the degradation ratio of optical network units under each splitter node and their spatial distribution characteristics, the degradation segment in the optical distribution network is determined, so as to realize the monitoring of optical path degradation of the optical distribution network.

[0082] Specifically, the triggering conditions for step S104 may include the following two cases.

[0083] 1) Single Alarm Signal: When a degradation warning signal is generated in the network, meaning that only one optical network unit (ONU) is identified as having experienced a degradation event, the delimitation process can still be initiated immediately. In this case, due to the single sample, the system will focus on the subtree of the topology containing the degraded ONU. By calculating the degradation ratio under its splitter node (1 / N in this case) and combining it with its position at the end of the resource tree, the delimitation logic can usually provide the most direct and accurate preliminary location, tending to determine that the degraded segment is the secondary splitter to which the degraded ONU belongs, the drop cable connecting it, or the degraded ONU device itself.

[0084] 2) Multiple Alarm Signals: When multiple degradation warning signals are generated in the network, meaning the total number of optical network units (ONUs) determined to have experienced degradation events is multiple, these warning signals will form a specific spatial distribution pattern in the resource tree topology (e.g., concentrated under some or all branches of the primary splitter). By systematically statistically analyzing the number and proportion of these degraded ONUs under each splitter node, and deeply analyzing whether their distribution characteristics are concentrated, dispersed, or hierarchically related, the system can infer the common upstream fault source with higher confidence, such as the trunk optical cable, the primary splitter itself, or related distribution optical cables.

[0085] In this way, it can provide a quick location reference for isolated user-side faults, and also perform reliable cluster analysis and intelligent tracing when facing large-scale network segment degradation, thus comprehensively covering various operation and maintenance scenarios from point to surface.

[0086] In one embodiment, step S104 includes the following sub-steps S1041 and S1042.

[0087] S1041. For each optical splitter node, calculate the ratio of the number of degraded optical network units to the total number of optical network units under the optical splitter node within its connected topology range, and obtain the degradation ratio.

[0088] Specifically, a resource tree topology for the passive optical network is pre-constructed, namely a five-layer resource tree model consisting of optical line terminal -> OLT PON port -> primary optical splitter -> secondary optical splitter -> optical network unit. Based on this, the relative positions and connections of the backbone optical cable, distribution optical cable, drop cable, and optical splitters at each level in the topology are clarified.

[0089] For each splitter node in the resource tree model described above, within a specified statistical time window, the entire subtree of its topology is scanned. Within this subtree, the total number of optical network units (ONUs) determined to have experienced a degradation event (i.e., generated at least a certain level of warning signal) is counted, and then the ratio between this number and the total number of all ONUs connected to that splitter node is calculated; this is the degradation ratio. For example, in the subtree of a secondary splitter node (with 32 ONUs), if 8 ONUs trigger warnings, then its degradation ratio R = 8 / 32 = 25%. This degradation ratio is a key spatial density indicator that can quantify the health status of a local area of ​​the network.

[0090] S1042. Based on the resource tree topology and the degradation ratio of each optical splitter node, the type of degradation segment in the optical distribution network is determined through the following steps S10421-S10424.

[0091] S10421. If the degradation ratio of all secondary optical splitter nodes under a certain primary optical splitter node exceeds the preset ratio threshold, the type of the degradation segment is determined to be the trunk optical cable connected upstream of the primary optical splitter node.

[0092] Specifically, when the degradation ratio (R) of all secondary optical splitter nodes under a certain primary optical splitter node exceeds a preset threshold (e.g., R_th = 50%), the degradation segment is determined to be the trunk optical cable connected upstream of that primary optical splitter node. The principle is that if all branches downstream of the primary optical splitter (i.e., all secondary optical splitters and their subordinate ONUs) experience large-scale degradation, this indicates that the root cause of the degradation is highly likely located in the only common upstream part of these branches, namely the trunk optical cable connecting that primary optical splitter. This is because the degradation of the trunk optical cable will equally affect all its downstream branches.

[0093] S10422. If the degradation ratio of each of the secondary optical splitter nodes under a certain primary optical splitter node exceeds the preset ratio threshold, the type of the degradation segment is determined to be the primary optical splitter node or the distribution optical cable between the primary optical splitter node and some secondary optical splitter nodes; some secondary optical splitter nodes include multiple secondary optical splitter nodes.

[0094] Specifically, when multiple, but not all, of the secondary optical splitter nodes under a certain primary optical splitter node have a degradation rate exceeding a preset threshold, the degradation segment is determined to be either the primary optical splitter node itself or the distribution optical cable between the primary optical splitter node and some of the secondary optical splitter nodes. This implies that the degradation affects part of the downlink ports of the primary optical splitter, possibly due to two reasons: 1) Degradation of a splitting module or corresponding port of the primary optical splitter itself, affecting all secondary optical splitters connected to that port; 2) Degradation of the distribution optical cable bundle (or some of its strands) connecting the primary optical splitter to these specific secondary optical splitters. Outputting these two possibilities side-by-side provides a more precise direction for troubleshooting in the field.

[0095] S10423. If the degradation ratio of a single secondary beam splitter node exceeds the preset ratio threshold, the type of the degradation segment is determined to be that of the secondary beam splitter node.

[0096] Specifically, when only a single secondary optical splitter node in the system experiences a degradation rate exceeding a preset threshold, while all other secondary optical splitter nodes under its primary optical splitter are functioning normally (with degradation rates not exceeding the preset threshold), the degradation segment is determined to be that secondary optical splitter node. This pattern indicates that the degradation phenomenon is strictly limited to the subtree of a single secondary optical splitter. The most reasonable inference is that the source of degradation is the secondary optical splitter itself, because if the upstream distribution cable or primary optical splitter fails, it should affect other secondary optical splitters on the same route.

[0097] S10424. If only a single optical network unit under a single secondary optical splitter node degrades, the type of the degraded segment is determined to be the drop cable connected to the optical network unit or the optical network unit equipment.

[0098] Specifically, when degradation is only manifested in a single secondary optical splitter node, with only a few (one or a very small number) optical network units (ONUs) experiencing degradation, and the overall degradation rate of that secondary optical splitter node does not exceed a threshold (i.e., most ONUs below it are normal), then the type of degradation segment is determined to be the drop cable connected to that optical network unit or the optical network unit itself. This indicates that the problem is not in the shared network equipment (secondary optical splitter) serving multiple users, but rather in the independent last link (drop cable) or the user terminal equipment itself connecting to a specific user terminal. This provides maintenance personnel with the most accurate on-site maintenance guidance.

[0099] After completing the spatial clustering and rule determination, a structured, complete delimitation result is generated that can directly drive subsequent automated operation and maintenance processes. This result is typically a structured data object or message, containing at least the following key information: 1) Identified deterioration segment type: Clearly indicate whether the root cause of the deterioration is the trunk optical cable, the primary splitter / distribution optical cable, the secondary splitter, or the drop cable / ONU; 2) Boundary confidence score: Based on the degradation ratio, the concentration of spatial distribution, and the clarity of rule matching, a quantitative confidence score is given to assist operation and maintenance personnel in assessing the reliability of the boundary results. 3) Detailed list of affected ONUs: Lists all specific ONUs identified as being at risk due to the degradation of this section, along with their corresponding warning levels, to facilitate accurate identification of the scope of user impact; 4) Preliminary handling and troubleshooting suggestions: Based on the paragraph type, the system can automatically link to the knowledge base to generate preliminary on-site troubleshooting steps, possible physical resources involved (such as optical distribution box number, port number), and suggested maintenance or further diagnostic measures.

[0100] By outputting structured early warnings with complete information, the results of intelligent analysis can be successfully transformed into actionable instructions that can directly guide operation and maintenance actions. This completes the closed loop from perception and early warning to location and delimitation and then to outputting decision-making basis, greatly improving the efficiency and accuracy of fault handling.

[0101] In one embodiment, after identifying the degraded segment in the optical distribution network, the method further includes: on a graphical representation of the resource tree topology, rendering the optical line terminal ports, splitter nodes, and connecting optical cables involved in the degraded segment using different colors according to the type, location, and corresponding degradation warning signal level of the degraded segment, generating a topology heatmap; for the optical network unit where the degradation event occurred, generating a comparison curve of its historical optical power time series data and the future optical power trend predicted by the time series prediction model; and visually displaying the topology heatmap and the comparison curve.

[0102] Specifically, after intelligently delimiting the degraded segments, this embodiment further transforms the analysis results into intuitive and actionable visualizations, significantly enhancing operations and maintenance personnel's overall perception of the network situation and their ability to deeply understand the root causes of problems. This is achieved by generating two types of core views.

[0103] Network topology heatmap: Based on the constructed resource tree topology, a graphical representation is generated. On this basis, according to the identified type of degraded segment, its precise location in the topology, and the level of associated degradation warning signals, a set of preset visual coding rules (e.g., green represents normal, yellow, orange, and red represent increasing warning levels) are used to dynamically render the network elements directly associated with the degraded segment. This includes affected optical line terminal ports, splitters that have failed or are upstream nodes of failures, and connected optical cables identified as degraded. Through this network topology heatmap, operations and maintenance personnel can clearly grasp the real-time health status distribution of the entire network, the severity of degradation (color), and the precise location of occurrence (graphical elements) on a global network topology map, achieving a comprehensive view of the entire network's health from a single image.

[0104] Optical Power Degradation Trend Comparison Chart: To conduct in-depth analysis of specific issues, this embodiment automatically generates a dedicated monitoring and analysis view for each optical network unit (ONU) experiencing a degradation event. The core of this view is a comparison chart, which plots the historical optical power time-series data of the ONU (as the factual baseline) alongside the optical power trend line predicted by a time-series prediction model (such as the GRU model) for a future period. Through this intuitive comparison of history and prediction, maintenance personnel can not only review the actual performance of the equipment over a past period but also clearly see the model's judgment on its future trends, thereby understanding the basis for the warning (e.g., the current value is normal, but the prediction curve shows a continued significant decline), and assisting in judging the urgency and speed of the fault.

[0105] Finally, the generated topology heatmap and optical power degradation trend comparison curve are pushed to a designated visualization monitoring interface for centralized display. This visualization method, which combines macroscopic situation (heatmap) and microscopic analysis (curve), constitutes a complete monitoring view from global positioning to single-point analysis, greatly enhancing the availability and decision support capabilities of the monitoring system. It upgrades maintenance work from viewing alarm lists to controlling the health status and drives intelligent work order dispatch, that is, converting the above-mentioned structured early warning information into maintenance work orders and pushing them to the relevant maintenance personnel in real time through DingTalk / WeChat robots, SMS, email, etc.

[0106] This invention provides a method for monitoring optical path degradation in a passive optical network (PON). Through a second-level telemetry protocol, it achieves continuous acquisition of received optical power data from each optical unit in the network, overcoming the data lag inherent in traditional polling methods. Based on this, a hybrid intelligent analysis mechanism is constructed by combining time-series prediction models such as gated cyclic units with anomaly detection models such as isolated forests. This mechanism can proactively identify trend-based degradation risks and instantaneous abnormal fluctuations before optical power triggers a fault alarm, thus achieving a shift from passive alarm to proactive early warning. Furthermore, the method introduces a spatial clustering analysis algorithm based on resource tree topology. By statistically analyzing the degradation ratio and spatial distribution characteristics under the splitter node, it can accurately determine whether the degraded segment is located in the backbone, distribution cable, drop cable, or the splitter itself, achieving rapid fault location. Therefore, this invention can realize intelligent monitoring of the optical path in a PON optical network's optical distribution network, from real-time monitoring and proactive early warning to precise demarcation.

[0107] Please refer to Figure 2 , Figure 2 This is a schematic diagram of the optical path degradation monitoring device for the optical distribution network in a passive optical network provided by the present invention. Figure 2 As shown, the device may include: The acquisition module 201 is used to acquire optical power timing data of each optical network unit in the passive optical network at a second-level cycle based on the telemetry protocol. The detection module 202 is used to predict the trend of degradation risk of each optical network unit based on the optical power time series data of each optical network unit through a time series prediction model; at the same time, based on the current optical power value of each optical network unit, it detects abnormal fluctuations through an anomaly detection model. The early warning module 203 is used to generate a corresponding level of early warning signal for optical network units that have experienced degradation events, based on the joint determination of trend degradation risk prediction results and abnormal fluctuation detection results. The delimiting module 204 is used to respond to at least one degradation warning signal, and based on the resource tree topology of the passive optical network, determine the degradation segment in the optical distribution network by statistically analyzing the degradation ratio of optical network units under each splitter node and analyzing their spatial distribution characteristics, so as to realize the monitoring of optical path degradation of the optical distribution network.

[0108] In some embodiments, the optical path degradation monitoring device for the optical distribution network in the passive optical network of the present invention can be implemented in a combination of hardware and software. As an example, the optical path degradation monitoring device for the optical distribution network in the passive optical network of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the optical path degradation monitoring method for the optical distribution network in the passive optical network of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0109] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0110] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the above-described methods for monitoring optical path degradation in a passive optical network (PON). In other words, an electronic device according to an embodiment of the present invention may include, but is not limited to: a processor and a memory; the memory is used to store the computer program; the processor is used to execute the method for monitoring optical path degradation in a PON as shown in any embodiment of the present invention by calling the computer program.

[0111] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3The illustrated electronic device 4000 includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0112] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0113] Bus 4002 may include a path for transmitting information between the aforementioned components. Bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 4002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus 4002 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0114] The memory 4003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0115] The memory 4003 stores application code (computer program) for executing the present invention, and its execution is controlled by the processor 4001. The processor 4001 executes the application code stored in the memory 4003 to implement the content shown in the foregoing method embodiments.

[0116] Among them, electronic devices can also be terminal devices, which can be any device that can install applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0117] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0118] An embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements any of the above-described methods for monitoring optical path degradation in a passive optical network.

[0119] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0120] In an exemplary embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned method for monitoring optical path degradation in a passive optical network.

[0121] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user'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 user'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).

[0122] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0123] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: 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 (EEPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0124] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0125] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0126] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0127] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0128] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for monitoring optical path degradation in an optical distribution network in a passive optical network, characterized in that, include: Based on the telemetry protocol, optical power timing data of each optical network unit in the passive optical network are collected at a rate of seconds. Based on the optical power time-series data of each optical network unit, a time-series prediction model is used to predict the trend of degradation risk of each optical network unit; at the same time, based on the current optical power value of each optical network unit, an anomaly detection model is used to detect abnormal fluctuations. Based on the joint determination of trend-based degradation risk prediction results and abnormal fluctuation detection results, a corresponding level of degradation early warning signal is generated for the optical network unit where a degradation event has occurred. In response to at least one of the aforementioned degradation warning signals, based on the resource tree topology of the passive optical network, the degradation ratio of the optical network units under each splitter node is statistically analyzed and their spatial distribution characteristics are analyzed to determine the degradation segment in the optical distribution network, thereby realizing the monitoring of optical path degradation of the optical distribution network.

2. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 1, characterized in that, The method of collecting optical power timing data of each optical network unit in a passive optical network at a second-level period based on a telemetry protocol includes: Send a subscription request based on a data modeling language definition to the optical line terminal in the passive optical network to establish a data acquisition channel; Through the data acquisition channel, the optical power timing data of each optical network unit actively pushed by the optical line terminal through at least one of its passive optical network ports at a second-level cycle is continuously received; wherein, the optical power timing data is encoded in binary encoding format and transmitted through a remote procedure call protocol that supports multiplexing, flow control and transport layer security encryption.

3. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 1, characterized in that, The time-series prediction model is a gated cyclic unit model; The method of predicting the trend degradation risk of each optical network unit based on the optical power time series data of each optical network unit through a time series prediction model includes: For each optical network unit, the optical power time-series data of the optical network unit within the historical time window, which has undergone data cleaning and normalization, is input into the gated cyclic unit model to obtain the optical power prediction value of the optical network unit at a specified future time point. Calculate the decrease in the predicted optical power value relative to the current actual optical power value; If the decline exceeds a preset degradation threshold, the optical network unit is determined to have a trend of degradation risk.

4. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 3, characterized in that, The gated recurrent unit model is obtained through the following steps: Historical optical power time-series data of several optical network units under normal and degraded conditions are extracted from the historical database. Each historical optical power time-series data is divided into multiple samples according to a fixed time window. The actual optical power values ​​of each sample at multiple different future time points are used as the label set of the sample to form a labeled sample set. The labeled sample set is divided into a training set and a validation set; The training set is input into the initial gated recurrent unit model, and an adaptive learning rate optimizer is used to perform iterative training with the goal of minimizing the error between the predicted optical power value output by the model and the actual optical power value at the corresponding future time point. After each round of training, the validation loss is calculated using the validation set; The changes in the verification loss are monitored. When the verification loss no longer decreases within a preset number of consecutive rounds, an early stopping mechanism is triggered to terminate the training process and obtain the gated recurrent unit model.

5. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 1, characterized in that, The anomaly detection model is an isolated forest model; The abnormal fluctuation detection based on the current optical power value of each optical network unit, using an anomaly detection model, includes: For each optical network unit, a multidimensional feature vector is constructed based on the optical power value of the optical network unit at the current moment and multiple derived features; wherein, the multiple derived features are used to characterize the instantaneous rate of change, fluctuation, and long-term trend of optical power; The multidimensional feature vector is input into the isolated forest model, and the anomaly score is calculated based on the average path length of the multidimensional feature vector in all isolated trees. If the abnormal score exceeds the abnormal detection threshold, it is determined that the optical network unit has abnormal fluctuations at the current moment.

6. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 5, characterized in that, The isolated forest model is trained through the following steps: Construct a multidimensional feature vector set for several optical network units under normal conditions; Multiple multidimensional feature vectors are randomly selected from the multidimensional feature vector set to form a positive sample set; Under the constraints of a preset number of subtrees and a preset number of samples extracted from each subtree, multiple isolated trees are generated based on the positive sample set, and the forest composed of the multiple isolated trees is used as the isolated forest model.

7. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 6, characterized in that, The multiple derived features include: first-order difference features, sliding window statistical features, and trend slope features; The multiple derived features are calculated using the following steps: The difference in optical power between adjacent time points is calculated to obtain the first-order difference feature; The mean and standard deviation are calculated within a preset sliding window to obtain the statistical characteristics of the sliding window; Linear fitting is performed on the data within the historical time window to obtain the trend slope characteristics.

8. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 1, characterized in that, The joint determination based on the trend-based degradation risk prediction results and the abnormal fluctuation detection results generates a corresponding level of degradation early warning signal for the optical network unit where a degradation event has occurred, including: For each optical network unit that experiences a degradation event, the magnitude and / or duration of future optical power decrease are determined based on the trend degradation risk prediction results, and the number of times the abnormal score exceeds a preset abnormal score threshold is determined based on the abnormal fluctuation detection results. If the decrease exceeds a preset magnitude threshold or the abnormal score exceeds a preset abnormal score threshold, a first-level degradation warning signal is generated. If the duration exceeds a preset duration threshold or the number of occurrences reaches a first threshold within a first preset time period, a second-level warning signal is generated. If the future optical power is lower than the receiving sensitivity threshold or the number of times reaches the second threshold within the second preset time period, a third-level warning signal is generated. The warning signal level is directly proportional to the severity of degradation, the second preset time period is shorter than the first preset time period, and the second number threshold is lower than the first number threshold.

9. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 1, characterized in that, The degradation rate of the optical network units at each splitter node is statistically analyzed, and their spatial distribution characteristics are examined. Identify degraded segments in the optical distribution network, including: For each optical splitter node, the proportion of the number of degraded optical network units within its subordinate topology range to the total number of optical network units under the optical splitter node is calculated to obtain the degradation ratio. Based on the resource tree topology and the degradation ratio of each of the optical splitter nodes, the type of degradation segment in the optical distribution network is determined by the following method: If the degradation ratio of all secondary optical splitter nodes under a certain primary optical splitter node exceeds a preset ratio threshold, then the type of the degradation segment is determined to be the trunk optical cable connected upstream of the primary optical splitter node. If the degradation ratio of each of the secondary optical splitter nodes under a certain primary optical splitter node exceeds the preset ratio threshold, then the type of the degradation segment is determined to be the primary optical splitter node or the optical cable between the primary optical splitter node and the secondary optical splitter nodes; the secondary optical splitter nodes include multiple secondary optical splitter nodes. If the degradation ratio of a single secondary beam splitter node exceeds the preset ratio threshold, then the type of the degradation segment is determined to be that of the secondary beam splitter node. If only a single optical network unit under a single secondary splitter node degrades, the type of the degraded segment is determined to be the drop cable connected to that optical network unit or the optical network unit device.

10. The method for monitoring optical path degradation in a passive optical network's optical distribution network according to claim 9, characterized in that, After identifying the degraded segments in the optical distribution network, the method further includes: In the graphical representation of the resource tree topology, different colors are used to render the optical line terminal ports, splitter nodes and connecting optical cables involved in the degraded segments according to the type, location and corresponding degraded warning signal level of the degraded segments, so as to generate a topology heat map. For optical network units that have experienced degradation events, a comparison curve is generated between their historical optical power time-series data and the future optical power trend predicted by the time-series prediction model. The topological heatmap and the comparison curve are visualized.

11. A device for monitoring optical path degradation in a passive optical network (PON), characterized in that, include: The acquisition module is used to acquire optical power timing data of each optical network unit in the passive optical network at a rate of seconds, based on the telemetry protocol. The detection module is used to predict the trend of degradation risk of each optical network unit based on the optical power time series data of each optical network unit through a time series prediction model. Meanwhile, based on the current optical power value of each optical network unit, abnormal fluctuations are detected using an anomaly detection model; The early warning module is used to generate a corresponding level of degradation early warning signal for the optical network unit that has experienced a degradation event, based on the joint determination of the trend degradation risk prediction results and the abnormal fluctuation detection results. The delimitation module is used to respond to at least one of the degradation warning signals, and based on the resource tree topology of the passive optical network, determine the degradation segment in the optical distribution network by statistically analyzing the degradation ratio of the optical network units under each splitter node and analyzing their spatial distribution characteristics, so as to realize the monitoring of optical path degradation of the optical distribution network.

12. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the optical path degradation monitoring method for optical distribution networks in passive optical networks as described in any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the optical path degradation monitoring method for optical distribution networks in passive optical networks as described in any one of claims 1 to 10.

Images