Optical line protection method

Abstract

The invention relates to the technical field of telecommunication transmission, in particular to an optical line protection method. The optical line protection method comprises the following steps: dividing an optical line network into a plurality of sub-line segments; setting a first sliding time window, and obtaining an optical line index of a main optical line of each sub-line segment; respectively calculating to obtain a risk value of each sub-line segment; for each sub-line segment, executing the following steps: determining a plurality of associated sub-line segments of the current sub-line segment; establishing a first input matrix, wherein the first input matrix comprises a risk value of the current sub-line segment, adjacent features of each associated sub-line segment and the current sub-line segment, and a risk value of each associated sub-line segment; inputting the first input matrix into a risk prediction model to obtain a future maximum risk value of the current sub-line segment; and when the future maximum risk value of the current sub-line segment is greater than a preset first risk threshold value, enabling the standby optical line to execute an early warning action. According to the method, the optical line can be protected more efficiently.

Description

Technical Field

[0001] This application relates to the field of electric communication transmission technologies, and in particular, to an optical line protection method. Background Art

[0002] In existing communication networks, optical fiber communication is the most important information transmission method. To ensure the reliability and continuity of communication, a dual-redundancy design of a primary optical line and a backup optical line is usually adopted. During daily operation, the detection system monitors the status of the primary optical line in real time. Once it detects that the key indicators (such as fiber attenuation, power, or bit error rate) of the primary optical line exceed the preset threshold, the system will automatically switch the communication to the backup optical cable, thereby avoiding communication interruption.

[0003] However, since it is difficult to accurately predict the occurrence time of a fiber failure, to ensure an immediate switch when a failure occurs, the backup optical line and its transmission equipment usually need to be kept turned on all the time. This not only wastes energy but also shortens the service life of the backup system.

[0004] Therefore, the existing primary and backup optical line protection methods are not intelligent and efficient enough. Summary of the Invention

[0005] To solve the above technical problems or at least partially solve the above technical problems, this application provides an optical line protection method, which can protect the optical line more intelligently and efficiently.

[0006] This application proposes an optical line protection method, and the optical line protection method includes the following steps: According to the installation positions of each optical line switching device in the optical line network, the optical line network is divided into multiple sub-segments, and each sub-segment includes the primary optical line and the backup optical line controlled by itself; Set a first sliding time window, and obtain the optical line indicators of the primary optical line of each sub-segment within this time window; According to the optical line indicators of each sub-segment within the first sliding time window, calculate the risk value of each sub-segment respectively through a preset risk value calculation function; For each sub-segment respectively, perform the following steps: According to the optical line indicator fluctuations of the current sub-segment and the optical line indicator fluctuations of other sub-segments within the first sliding time window, determine multiple associated sub-segments of the current sub-segment; Establish a first input matrix, and the first input matrix includes the risk value of the current sub-segment, the adjacency characteristics between each associated sub-segment and the current sub-segment, and the risk value of each associated sub-segment; Input the first input matrix into a preset risk prediction model to obtain the future maximum risk value of the current sub-segment; When the future maximum risk value of the current sub-segment is greater than a preset first risk threshold, the standby optical line of the current sub-segment performs a warning action.

[0007] Optionally, the optical line protection method further includes the following steps: Obtain sub-segments whose current risk value is greater than a preset second risk threshold to obtain a plurality of hypothetical associated sub-segments; Combine each hypothetical associated sub-segment with the current sub-segment one by one to respectively establish a plurality of second data sets; The second data set includes the risk value of the current sub-segment, the adjacency feature between the current hypothetical associated sub-segment and the current sub-segment, and the risk value of the current hypothetical sub-segment; Input the second data set into a preset risk prediction model to obtain the future maximum risk value of the current sub-segment; When the future maximum risk value of the current sub-segment is greater than a third risk threshold, adjust the association threshold for determining that the current hypothetical associated sub-segment is an associated sub-segment of the current sub-segment according to the difference between the future maximum risk value and the third risk threshold, so that the greater the difference, the easier it is for the current hypothetical associated sub-segment to be determined as an associated sub-segment of the current sub-segment; The third risk threshold is less than the first risk threshold.

[0008] Optionally, according to the optical line indicators of each sub-segment within a first sliding time window, the risk value of each sub-segment is calculated respectively through a preset risk value calculation function, including the following steps: The optical line indicators include: average optical power attenuation, port bit error rate, optical signal-to-noise ratio, optical line delay, and optical line jitter; The preset risk value calculation function is where R represents the risk value of the current sub-segment, ΔP represents the average optical power attenuation, BER represents the port bit error rate, OSNR represents the optical signal-to-noise ratio, Delay represents the optical line delay, Jitter represents the optical line jitter, 、 、 、 and are preset empirical parameters.

[0009] Optionally, according to the optical line index fluctuations of the current sub-segment and the optical line index fluctuations of other sub-segments within a first sliding time window, a plurality of associated sub-segments of the current sub-segment are determined, including the following steps: The fluctuations of the optical line indicators at least include one or more of the following time-varying data: the time-series change sequence of the optical power attenuation, the time-series change sequence of the port bit error rate, the time-series change sequence of the optical signal-to-noise ratio, the time-series change sequence of the optical line delay, and the time-series change sequence of the optical line jitter amount; Calculate the similarity between the optical line indicator fluctuation sequence of the current sub-segment within the first sliding time window and the optical line indicator fluctuation sequences of other sub-segments within the same time window respectively through the DTW algorithm, so as to obtain the similarity between each other sub-segment and the current sub-segment; When the similarity between an other sub-segment and the current sub-segment is greater than the associated threshold corresponding to the other sub-segment, determine the other sub-segment as the associated sub-segment of the current sub-segment.

[0010] Optionally, the preset risk prediction model is trained through the following steps: Step 1: Use a convolutional neural network as the model framework of the risk prediction model; Step 2: Collect multiple groups of first training data, and each group of first training data includes: The risk value of a target sub-segment, where the target sub-segment represents any sub-segment in the optical line network; The data of each sub-segment associated with the target sub-segment, including the adjacency features between each associated sub-segment and the target sub-segment and the risk values of each associated sub-segment; Step 3: For each group of first training data, collect a corresponding future maximum risk value, that is, the maximum risk value of the target sub-segment within a preset future time period; Step 4: Use the first training data as the input of the model framework and the future maximum risk value as the label of the model framework to train the final risk prediction model.

[0011] Optionally, when the future maximum risk value of the current sub-segment is greater than the preset first risk threshold, make the standby optical line of the current sub-segment perform a warning action, including the following steps: The warning action includes the standby optical line of the current sub-segment, and completes the signal quality test and traffic simulation switching test of the standby optical line.

[0012] The technical solution provided by this application has the following advantages compared with the prior art: One of its beneficial effects and the discussion of its working principle is that in the traditional protection scheme of the optical network, it is usually based on the obvious abnormality of certain monitored values in real time before switching. For example, when key indicators such as optical power attenuation or bit error rate deteriorate significantly, the switching of the standby line will be triggered passively. This method often lacks the forward-looking identification of potential risks and it is difficult to provide enough response time in a timely manner.

[0013] To address this problem, the method described in the present application divides the optical line network into multiple sub-segments, and continuously obtains the optical line indicators of each sub-segment (such as optical power attenuation, port bit error rate, delay, jitter, etc.) within a first sliding time window. After quantifying them into risk values, they are combined with the associated sub-segment information identified by "fluctuation similarity" and input into a preset risk prediction model in the form of a first input matrix, thereby predicting the future maximum risk value of the current sub-segment, and waking up the backup optical line in advance to perform early warning actions when the first risk threshold is exceeded.

[0014] The beneficial effects of this program are reflected in the following aspects: Different sub-segments often have certain connections or collateral effects. Through continuous data collection and model prediction mechanisms, the system can actively discover the correlation effects between sub-segments and capture potential risks before the signs of faults have fully deteriorated.

[0015] When the risk is low, the backup optical line can be in a state of low power consumption or partial dormancy. When the warning threshold is triggered, when the risk is high, the signal quality test and traffic simulation switching preparation are completed quickly, thereby effectively optimizing the backup resource scheduling strategy and taking into account both energy efficiency and cost-effectiveness.

[0016] Through the above collaboration, the present invention significantly improves the initiative and resilience of optical line protection, and achieves a balance between network energy efficiency and network robustness to the greatest extent.

[0017] The second beneficial effect and the discussion of its working principle is that in the actual operation of the optical network line, some sub-segments associated with the current sub-segment may have abnormalities, but the abnormality of the current sub-segment is not yet obvious. Since the similarity of its fluctuation characteristics with the current sub-segment has not yet reached the judgment standard, it may be ignored in the correlation analysis. However, once the potential failures of these high-risk sub-segments break out in full, and the current sub-segment has not yet completed the startup of the backup optical line, the severity of its chain reaction is often far beyond expectations. For such critical scenarios with unacceptable consequences, this application also proposes a set of dynamic optimization strategies for pre-assumed association thresholds.

[0018] Specifically, when this application monitors that the risk value of a sub-segment exceeds the second risk threshold (significantly higher than the normal level), even if it is not very similar to the indicator fluctuation of the current sub-segment, the system will first include it in a special "hypothesis association set" and conduct a forward-looking combined risk forecast based on it.

[0019] By comparing the gap between the future maximum risk value of the sub-segment and the third risk threshold, the system can determine whether the severity of a failure in the sub-segment is within a controllable range.

[0020] If the difference is large, that is, the risk significantly exceeds the warning line, it indicates that even if the correlation between this sub-segment and the current sub-segment is uncertain, it must be regarded as a "critical anomaly" in advance and given key attention. At this time, the system will automatically lower the correlation threshold of this sub-segment, so as to include it in the joint risk assessment scope with the current sub-segment with higher sensitivity.

[0021] This dynamic optimization of the predictive correlation threshold can identify hidden risks that, although not significantly correlated, may cause serious consequences once out of control as early as possible, and capture them in the global linkage disposal process in the first time, so as to capture these potential threats as early as possible.

[0022] If the difference is not large, it means that even if this sub-segment has an anomaly, there will not be overly serious consequences. The system can retain the original threshold judgment to avoid excessive consumption of resources.

[0023] In this way, this application can not only achieve more stringent prevention for high-risk sub-segments, but also will not bring frequent and meaningless warnings to the entire network, thus achieving an effective balance between warning sensitivity and global resource efficiency. Brief Description of the Drawings

[0024] Figure 1 It is a schematic flowchart of the optical line protection method provided by the embodiment of this application. Detailed Embodiment

[0025] Next, the technical solutions in this application will be described in conjunction with the drawings.

[0026] Many specific details are set forth in the following description in order to provide a thorough understanding of this application, but this application may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only a part of the embodiments of this application, rather than all the embodiments. It should be noted that, without conflict, the embodiments of this application and the features in the embodiments may be combined with each other.

[0027] This application proposes an optical line protection method, and the optical line protection method includes the following steps: S101: Divide the optical line network into multiple sub-segments according to the setting positions of each optical line switching device in the optical line network, and each sub-segment includes the main optical line and the standby optical line controlled by itself.

[0028] S102: Set a first sliding time window, and obtain the optical line indicators of the main optical line of each sub-segment within this time window.

[0029] Specifically, the first sliding time window here represents a data window that is periodically updated along the time axis. In the embodiments of the present application, the size of the first sliding time window is one week, and the update time step is 24 hours.

[0030] Specifically, the optical line indicators can be directly retrieved from the log files generated by the optical line detection function configured in the optical line control device. The optical line indicators include: average optical power attenuation, port bit error rate, optical signal-to-noise ratio, optical line delay, and optical line jitter.

[0031] S103: Calculate the risk value of each sub-segment respectively through a preset risk value calculation function according to the optical line indicators of each sub-segment within the first sliding time window; Specifically, it includes the following steps: Input the optical line indicators of the main optical line of each sub-segment into the risk value calculation function to obtain the risk value of each sub-segment.

[0032] The preset risk value calculation function is where R represents the risk value of the current sub-segment, ΔP represents the average optical power attenuation, BER represents the port bit error rate, OSNR represents the optical signal-to-noise ratio, Delay represents the optical line delay, Jitter represents the optical line jitter, 、 、 、 and are empirically preset parameters set by humans.

[0033] For each sub-segment respectively, perform the following steps: S104: Determine multiple associated sub-segments of the current sub-segment according to the fluctuations of the optical line indicators of the current sub-segment and the optical line indicators of other sub-segments within the first sliding time window.

[0034] Specifically, in the embodiments of the present application, it includes steps S201 - S202.

[0035] S201: Obtain the association threshold for each other sub-segment to be determined as an associated sub-segment of the current sub-segment.

[0036] Step 1: Set the association thresholds for all other sub-segments to be determined as associated sub-segments of the current sub-segment as a manually set association reference value.

[0037] Step 2: Obtain sub-segments with a current risk value greater than a preset second risk threshold to obtain multiple hypothetical associated sub-segments; Associate each assumed associated sub-segment with the current sub-segment one by one to establish multiple second data sets respectively in this way. The second data set includes the risk value of the current sub-segment, the adjacency feature between the current assumed associated sub-segment and the current sub-segment, and the risk value of the current assumed sub-segment. Input the second data set into a preset risk prediction model to obtain the future maximum risk value of the current sub-segment. When the future maximum risk value of the current sub-segment is greater than the third risk threshold, adjust the association threshold for determining that the current assumed associated sub-segment is the associated sub-segment of the current sub-segment according to the difference between the future maximum risk value and the third risk threshold, so that the greater the difference, the easier it is for the current assumed associated sub-segment to be determined as the associated sub-segment of the current sub-segment. The third risk threshold is less than the first risk threshold.

[0038] Specifically, in the embodiment of the present application, the following association threshold adjustment formula is used to adjust the association threshold for determining that the current assumed associated sub-segment is the associated sub-segment of the current sub-segment according to the difference between the future maximum risk value and the third risk threshold.

[0039] The association threshold adjustment formula is: Wherein, represents the adjusted association threshold, represents the artificially set association reference value, represents the future maximum risk value obtained through the risk prediction model, represents the third risk threshold, η represents the artificially set linear adjustment coefficient, and the linear adjustment coefficient should be a positive number.

[0040] S202: According to the association threshold for determining that each other sub-segment is the associated sub-segment of the current sub-segment obtained in S201, the similarity between each other sub-segment and the current sub-segment is obtained through the following steps, and a judgment is made on whether it is an associated sub-segment.

[0041] Specifically, the optical line index fluctuation at least includes one or more of the following data that change over time: the time series change sequence of the optical power attenuation amount, the time series change sequence of the port bit error rate, the time series change sequence of the optical signal-to-noise ratio, the time series change sequence of the optical line delay, and the time series change sequence of the optical line jitter amount. Calculate the similarity between the optical line index fluctuation sequence of the current sub-segment within the first sliding time window and the optical line index fluctuation sequences of other sub-segments within the same time window through the DTW algorithm to obtain the similarity between each other sub-segment and the current sub-segment. When the similarity between another sub-segment and the current sub-segment is greater than the associated threshold corresponding to the other sub-segment, the other sub-segment is determined as an associated sub-segment of the current sub-segment.

[0042] S105: Establish a first input matrix, where the first input matrix includes the risk value of the current sub-segment, the adjacency feature between each associated sub-segment and the current sub-segment, and the risk value of each associated sub-segment; Input the first input matrix into a preset risk prediction model to obtain the future maximum risk value of the current sub-segment.

[0043] Specifically, in the embodiment of the present application, the preset risk prediction model is trained through the following steps: Step 1: Use a convolutional neural network as the model framework of the risk prediction model; Step 2: Collect multiple groups of first training data, and each group of first training data includes: The risk value of a target sub-segment, where the target sub-segment represents any sub-segment in the optical line network; The data of each sub-segment associated with the target sub-segment, including the adjacency feature between each associated sub-segment and the target sub-segment and the risk value of each associated sub-segment; Step 3: For each group of first training data, collect a corresponding future maximum risk value, that is, the maximum risk value of the target sub-segment within a preset future time period; Step 4: Use the first training data as the input of the model framework and the future maximum risk value as the label of the model framework to train the final risk prediction model.

[0044] S106: When the future maximum risk value of the current sub-segment is greater than a preset first risk threshold, make the standby optical line of the current sub-segment perform a warning action.

[0045] Specifically, making the standby optical line of the current sub-segment perform a warning action includes the following steps: The warning action includes the standby optical line of the current sub-segment, and completes the signal quality test and traffic simulation switching test of the standby optical line.

[0046] In summary, the technical solution provided by the embodiment of the present application has the following advantages compared with the prior art: One of its beneficial effects and the discussion of its working principle is that in the traditional protection scheme of optical networks, it is usually based on the obvious abnormality of certain monitored values in real time before switching. For example, when key indicators such as optical power attenuation or bit error rate deteriorate significantly, the switching of the standby line will be triggered passively. This method often lacks the forward-looking identification of potential risks and it is difficult to provide enough response time in a timely manner.

[0047] To address this issue, the method described in this application divides multiple sub-segments in the optical line network, continuously obtains the optical line indicators of each sub-segment (such as optical power attenuation, port bit error rate, time delay, jitter, etc.) within the first sliding time window, quantifies them into risk values, and then combines the associated sub-segment information identified by "fluctuation similarity" and inputs it in the form of a first input matrix into a preset risk prediction model to predict the future maximum risk value of the current sub-segment, and when it exceeds the first risk threshold, wakes up the standby optical line in advance to perform a warning action.

[0048] The beneficial effects of this solution are reflected in the following aspects: There are often certain associations or collateral effects among different sub-segments. Through continuous data collection and model prediction mechanisms, the system can actively explore the correlation effects among sub-segments and capture potential risks before the fault signs have fully deteriorated.

[0049] When the risk is low, the standby optical line can be in a state of low power consumption or partial dormancy. When the warning threshold is triggered and the risk is high, signal quality testing and traffic simulation switching preparation can be quickly completed, thus effectively optimizing the standby resource scheduling strategy and taking into account both energy efficiency and cost-effectiveness.

[0050] Through the above coordination, the present invention significantly improves the initiative and resilience of optical line protection and achieves a balance between network energy efficiency and network robustness to the greatest extent.

[0051] The discussion of the second beneficial effect and its working principle is that in the actual operation of the optical network line, some sub-segments associated with the current sub-segment may have already shown abnormalities, but the abnormalities of the current sub-segment are not obvious. Since the similarity of its fluctuation characteristics to the current sub-segment has not reached the judgment standard, it may be ignored in the correlation analysis. However, once the potential faults of these high-risk sub-segments fully break out and the current sub-segment has not completed the startup of the standby optical line, the severity of the chain effect is often far beyond expectations. For this critical scenario where the consequences are unacceptable, this application also proposes a dynamic optimization strategy for the pre-set association threshold.

[0052] Specifically, when this application monitors that the risk value of a certain sub-segment exceeds the second risk threshold (significantly higher than the normal level), even if its index fluctuation is not very similar to that of the current sub-segment, the system will first include it in a special "hypothetical association set" and conduct a forward-looking combined risk prediction based on this.

[0053] By comparing the gap between the future maximum risk value of this sub-segment and the third risk threshold, the system can judge whether the severity is within the controllable range once this sub-segment fails.

[0054] If the difference is large, that is, the risk significantly exceeds the warning line, it indicates that even if the correlation between this sub-segment and the current sub-segment is uncertain, it must be regarded as a "critical anomaly" in advance and given key attention. At this time, the system will automatically lower the correlation threshold of this sub-segment, so as to include it in the joint risk assessment scope with the current sub-segment with higher sensitivity.

[0055] This dynamic optimization of the predictive correlation threshold can identify hidden risks that, although not obviously correlated, may cause serious consequences once out of control as early as possible, and capture them into the global linkage handling process in the first time, so as to capture these potential threats as early as possible.

[0056] If the difference is not large, it means that even if this sub-segment has an anomaly, there will not be overly serious consequences. The system can retain the original threshold judgment to avoid excessive consumption of resources.

[0057] In this way, this application can not only achieve more stringent prevention for high-risk sub-segments, but also will not bring frequent and meaningless warnings to the entire network, thus achieving an effective balance between warning sensitivity and global resource efficiency.

[0058] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Additionally, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element. Moreover, in the description of the embodiments of this application, unless otherwise specified, " / " means "or". For example, A / B can represent A or B; "and / or" in this article is only a description of the association relationship of the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. And, in the description of the embodiments of this application, "multiple" means two or more than two.

[0059] The above are only specific embodiments of the present application, enabling those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments described herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An optical line protection method, characterized in that: The optical line protection method comprises the following steps: According to the setting position of each optical line switching device of the optical line network, the optical line network is divided into a plurality of sub-line segments, each sub-line segment includes a main optical line and a backup optical line controlled by itself; Setting a first sliding time window, and obtaining an optical line indicator of a main optical line of each sub-line segment within the time window; According to the optical line index of each sub-segment in the first sliding time window, a risk value of each sub-segment is calculated by using a preset risk value calculation function; For each sub-segment separately, perform the following steps: Determine a plurality of associated sub-segments of the current sub-segment according to fluctuations of optical line indicators of the current sub-segment and fluctuations of optical line indicators of other sub-segments within the first sliding time window; Establishing a first input matrix, wherein the first input matrix includes a risk value of a current sub-segment, adjacency features of each associated sub-segment and the current sub-segment, and a risk value of each associated sub-segment; Input the first input matrix into a preset risk prediction model to obtain the future maximum risk value of the current sub-segment; When the future maximum risk value of the current sub-segment is greater than a preset first risk threshold, the backup optical line of the current sub-segment is enabled to perform an early warning action.

2. The optical line protection method according to claim 1, characterized in that: The optical line protection method further comprises the following steps: Obtaining a sub-line segment whose current risk value is greater than a preset second risk threshold value to obtain a plurality of hypothesized associated sub-line segments; Combining each of the assumed associated sub-segments with the current sub-segment one by one, thereby establishing a plurality of second data sets respectively; The second data set includes the risk value of the current sub-segment, the adjacency feature between the current assumed associated sub-segment and the current sub-segment, and the risk value of the current assumed sub-segment; Inputting the second data set into a preset risk prediction model to obtain a future maximum risk value of the current sub-segment; When the future maximum risk value of the current sub-segment is greater than the third risk threshold, adjusting the association threshold at which the current assumed associated sub-segment is judged as the associated sub-segment of the current sub-segment according to the difference between the future maximum risk value and the third risk threshold, so that the larger the difference is, the easier it is for the current assumed associated sub-segment to be judged as the associated sub-segment of the current sub-segment; The third risk threshold is lower than the first risk threshold.

3. The optical line protection method according to claim 1, characterized in that: Calculating the risk value of each sub-segment according to the optical line index of each sub-segment in the first sliding time window by using a preset risk value calculation function includes the following steps: The optical line indicators include: average optical power attenuation, port bit error rate, optical signal-to-noise ratio, optical line delay and optical line jitter; The preset risk value calculation function is: Among them, R represents the risk value of the current sub-segment, ΔP represents the average optical power attenuation, BER represents the port bit error rate, OSNR represents the optical signal-to-noise ratio, Delay represents the optical line delay, and Jitter represents the optical line jitter. , , , and are preset empirical parameters.

4. The optical line protection method according to claim 2, characterized in that: Determining to obtain multiple associated sub-segments of the current sub-segment according to the optical line index fluctuation of the current sub-segment and the optical line index fluctuation of other sub-segments within the first sliding time window includes the following steps: The optical line indicator fluctuations include at least one or more of the following data that change with time: a time series change sequence of optical power attenuation, a time series change sequence of port bit error rate, a time series change sequence of optical signal-to-noise ratio, a time series change sequence of optical line delay, and a time series change sequence of optical line jitter; The optical line index fluctuation sequence of the current sub-segment in the first sliding time window is respectively calculated by the DTW algorithm with the optical line index fluctuation sequence of other sub-segments in the same time window, so as to obtain the similarity between each other sub-segment and the current sub-segment; When the similarity between another sub-segment and the current sub-segment is greater than the associated threshold corresponding to the other sub-segment, the other sub-segment is determined as an associated sub-segment of the current sub-segment.

5. The optical line protection method according to claim 1, characterized in that: The preset risk prediction model is trained by the following steps: Step 1: Use convolutional neural network as the model framework of risk prediction model; Step 2: Collect multiple sets of first training data, each set of first training data includes: a risk value of a target sub-segment, wherein the target sub-segment represents any sub-segment in the optical line network; The data of each sub-segment associated with the target sub-segment, including the adjacency characteristics of each associated sub-segment and the target sub-segment and the risk value of each associated sub-segment; Step 3: For each set of first training data, collect a corresponding future maximum risk value, that is, the maximum risk value of the target sub-segment in a preset future time period; Step 4: Use the first training data as the input of the model framework, use the future maximum risk value as the label of the model framework, and train to obtain the final risk prediction model.

6. The optical line protection method according to claim 1, characterized in that: When the future maximum risk value of the current sub-segment is greater than a preset first risk threshold, the backup optical line of the current sub-segment is enabled to perform an early warning action, including the following steps: The early warning action includes the backup optical line of the current sub-line segment, and completes the signal quality test and the flow simulation switching test of the backup optical line.

Images