Failure prediction device, failure prediction method, and failure prediction program

Abstract

In the present invention, upon detection of a predetermined optical path in which a quality value of an optical signal measured at a first time is less than a first threshold, the prediction server (SV) determines an abnormality level at the first time from a parameter measurement value of a photophysical property parameter that is measured at a point of passage of the predetermined optical path at the first time, and, when a correlation between the quality value of the optical signal obtained at the first time and the abnormality level obtained at the first time matches correlation data stored in a storage unit, the prediction server (SV) predicts that a sign of failure will occur in a sign detection part obtained from a point of measurement of the parameter measurement value at a time later than the first time.

Description

[Technical Field]

[0001] The present invention relates to a failure prediction device, a failure prediction method, and a failure prediction program. [Background technology]

[0002] In the maintenance and operation of optical transmission systems, there is a demand for early detection of faults and highly accurate identification of their locations. To address this demand, methods have been developed for automatically detecting faults in optical transmission systems and identifying their locations. Patent Document 1 discloses a method for narrowing down the range of suspected fault locations based on packet loss information in higher layers and the accommodation relationship of optical-channel data units (ODU) paths.

[0003] Patent Document 2 discloses a method for narrowing down the range of suspected locations on an optical path basis based on the optical signal characteristics of the receiving end of an optical path such as an Optical Multiplex Section (OMS) and the accommodation relationship of Optical Transport Unit (OTU) paths, and further identifying suspected locations in this optical path based on the optical signal characteristics. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2018-64160 [Patent Document 2] Japanese Patent Publication No. 2020-88628 Summary of the Invention [Problem to be solved by the invention]

[0005] Conventional technologies such as those in Patent Documents 1 and 2 do not use parameter measurements for each repeating section of the optical path, and therefore are insufficient in narrowing the range of fault detection and in detecting it with sufficient accuracy. Therefore, we consider detecting suspected fault locations from parameter measurements for each repeating section of the optical path.

[0006] For example, when a parameter measurement value that has experienced a fluctuation greater than the steady fluctuation range is judged to have a fluctuation, there may be many relay sections in which a fluctuation has been detected (for example, 90% of the relay sections have fluctuation). In such cases, it is difficult to identify the fault location from those relay sections. Even when no fault has occurred, fluctuations greater than the steady-state fluctuation range may occur. For example, if a maintenance technician temporarily touches the fiber, a disturbance occurs at the point of contact, causing the measured parameter values ​​to fluctuate momentarily, but then recover quickly. If such a false fault is mistakenly detected as the fault location, unnecessary warnings will hinder the maintenance technician's maintenance work.

[0007] Therefore, a main object of the present invention is to narrow down the range of fault detection in an optical transmission system with high precision. [Means for solving the problem]

[0008] In order to solve the above problems, the failure prediction device of the present invention has the following features. The present invention provides a failure prediction device having a storage unit and a control unit, the storage unit stores correlation data between a quality value of an optical signal measured at a receiving end point of an optical path in an optical transmission system and an abnormality degree for each type of optical physical property parameter measured at a passing point of the optical path; The control unit when detecting a predetermined optical path in which the quality value of the optical signal measured at a first time is less than a first threshold, calculating an anomaly degree at the first time from parameter measurement values ​​of the optical physical property parameters measured at a passing point of the predetermined optical path at the first time; When the correlation between the quality value of the optical signal at the first time and the degree of abnormality at the first time matches the correlation data stored in the memory unit, it is predicted that a sign of failure will occur at a time after the first time at the sign detection location determined from the measurement point of the parameter measurement value. [Effects of the Invention]

[0009] According to the present invention, the range of fault detection in an optical transmission system can be narrowed down with high precision. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 2 is a configuration diagram of an optical transmission system and a prediction server according to the present embodiment. [Figure 2] FIG. 2 is a hardware configuration diagram of the prediction server of FIG. 1 according to the present embodiment. [Figure 3] FIG. 2 is a configuration diagram of the transponder and relay transmission device of FIG. 1 according to the present embodiment. [Figure 4] 10 is a table showing the results of evaluating whether or not there is a fluctuation in the measured parameter values ​​of the optical transmission system according to the present embodiment. [Figure 5] 10 is a table showing the results of evaluating the deviation from the correlation line for the measured parameter values ​​of the optical transmission system according to the present embodiment. [Figure 6] 2 is a flowchart showing an example of a procedure for operating the optical transmission system of FIG. 1 according to the present embodiment. [Figure 7] 10 is a flowchart showing details of a failure prediction process of the prediction server according to the present embodiment. [Figure 8] 10 is a graph showing a correlation line between the degree of abnormality of a parameter x and a Q value according to the present embodiment. [Figure 9] 10 is a graph showing a correlation line between the degree of abnormality of parameter y and the Q value according to this embodiment. [Figure 10] 10 is a time series graph showing an error prediction line when an error occurs according to the present embodiment. [Figure 11]10 is a graph showing an error prediction line when a false error occurs according to the present embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, each embodiment of the present invention will be described in detail with reference to the drawings.

[0012] FIG. 1 is a configuration diagram of an optical transmission system NW and a prediction server (failure prediction device) SV. The optical transmission system NW includes transponders TS1, TS2, TR1, TR2, TR3, and TR4, and relay transmission devices WA, WB, WC, WD, WE, and WF. A transponder is a logical transmission path through which an optical signal passes, and is a device located at the start or end of an optical path (optical path P1 in FIG. 1) constructed as an optical-channel data unit (ODU) path and an optical transport unit (OTU) path. An optical path can also be constructed as a single ODU path by connecting multiple OTU paths using a 3R transponder, but in this case, fault isolation is also performed for each OTU path, which is the smallest unit. In this embodiment, transponders TS1 and TS2 (the second letter of the code is S) that are the start points of the optical paths, and transponders TR1, TR2, TR3, and TR4 (the second letter of the code is R) that are the end points of the optical paths are exemplified.

[0013] The relay transmission devices are optical cross connects (OXCs) that relay optical signals along optical paths. In this embodiment, six relay transmission devices WA, WB, WC, WD, WE, and WF are connected to adjacent relay transmission devices by links L1 to L6. For example, the optical path P1 is connected in the following order from the transmitting side: transponder TS2 → relay transmission device WA → link L1 → relay transmission device WB → link L4 → relay transmission device WC → link L6 → relay transmission device WD → transponder TR4 (receiving side).

[0014] The prediction server SV identifies signs of a possible failure occurring in the optical transmission system NW by combining information indicating the impact on communication services at the receiving endpoint with information indicating fluctuations in areas that are signs of a failure (sign detection areas) using the following procedure. (Step 1) Based on the quality information at the time of signal reception notified from transponder TR4, which is the receiving end point of optical path P1, a possible failure that may occur in optical path P1 is predicted. The quality information at the time of signal reception is information that indicates the impact on the communication service at the receiving end point, and for example, the Q value (Quality factor, unit: dB, the larger the value, the better the quality, and the smaller the value, the worse the quality) that indicates the optical signal quality of the optical path is used. (Step 2) Identify the precursor detection location from the parameter measurement values ​​for each relay section of each relay transmission device WA, WB, WC, and WD through which the optical path P1 predicted in (Step 1) passes. The parameter measurement values ​​are measurements taken for each measurement point and for each type of parameter of the optical physical properties (hereinafter referred to as "parameter type"). The parameter measurement values ​​are information that indicates fluctuations in the precursor detection location.

[0015] Therefore, the prediction server SV has a memory unit and a control unit, and the memory unit stores correlation data (hereinafter referred to as correlation lines) between the quality value of the optical signal measured at the receiving end point of the optical path within the optical transmission system NW and the degree of abnormality for each type of optical physical property parameter measured at the pass-through point of the optical path. Then, when the prediction server SV detects a specified optical path in which the quality value of the optical signal measured at the first time is less than a first threshold, it calculates the degree of abnormality at the first time from the parameter measurement values ​​of the optical physical property parameters measured at the passing points of the specified optical path at the first time. Furthermore, if the correlation between the quality value of the optical signal at the first time and the degree of abnormality at the first time matches the correlation data stored in the memory unit, the prediction server SV predicts that a sign of failure will occur at a time after the first time at the sign detection location determined from the measurement point of the parameter measurement value.

[0016] FIG. 2 is a hardware configuration diagram of the prediction server SV of FIG. The prediction server SV is configured as a computer 900 having a CPU (control unit) 901 , a RAM 902 , a ROM 903 , an HDD (storage unit) 904 , a communication I / F 905 , an input / output I / F 906 , and a media I / F 907 . The communication I / F 905 is connected to an external communication device 915. The input / output I / F 906 is connected to an input / output device 916. The media I / F 907 reads and writes data from a recording medium 917. Furthermore, the CPU 901 controls each unit by executing a program (also called an application, or an app for short) loaded into the RAM 902. This program can also be distributed via a communication line or recorded on a recording medium 917 such as a CD-ROM and distributed.

[0017] FIG. 3 is a diagram showing the configuration of the transponder TS2 and the relay transmission device WC shown in FIG. The transponder TS2 has a package 10 that accommodates an OTN (Optical Transport Network) framer 11, a DSP (Digital Signal Processor) 12, and an optical device 13. The OTN framer 11 is an LSI (Large Scale Integration) that converts a data signal to be transmitted into a signal frame format of the OTN, which is an optical transmission network. The DSP 12 is an LSI that performs digital signal processing for signal conversion and correction for long-distance, large-capacity transmission, and monitors various analog information data used for conversion and correction processing at the receiving end point. The optical device 13 transmits and receives optical signals.

[0018] The relay transmission device WC has three packages WC1, WC2, and WC3, and an optical physical property monitor 21 connected to a port of each package. Each of the packages WC1, WC2, and WC3 includes an AMP (Amplifier) ​​22 and a WSS (Wavelength Selective Switch) 23. The optical physical property monitor 21 measures time-series data of analog information related to the optical physical properties of the optical signal in the repeating section as parameter measurement values. The optical physical property monitor 21 measures, for example, optical spectrum information that expresses the optical signal as the light intensity relative to the wavelength or frequency, the waveform symmetry of the optical signal that can be confirmed from the optical spectrum information, and the signal quality OSNR (Optical Signal to Noise Ratio) that is expressed as the ratio of each optical signal to noise. The AMP 22 is an optical amplifier, and the WSS 23 is a wavelength selective switch.

[0019] An outline of the process for predicting a failure of the prediction server SV will be explained below with reference to FIGS. Figure 4 is a table showing the results of evaluating whether or not there is fluctuation in the measured parameter values ​​of the optical transmission system NW. The row elements of this table indicate each parameter (such as full-wave power), and the column elements indicate the relay transmission equipment (such as relay transmission equipment WA) at the measurement point. In Figure 4, all of the 16 parameter measurement values ​​(4 measurement locations x 4 parameter types) are in a state where fluctuations greater than the steady fluctuation range have occurred (shown as "fluctuations" in the figure). In this case, it is difficult for the prediction server SV to identify the failure location using a method that identifies the location where there is a change in the parameter measurement as the precursor detection location.

[0020] The measurement point may be a device unit such as a relay transmission device, or a component unit (such as a package part) within the relay transmission device. For example, the relay transmission device WC in Fig. 3 houses three packages WC1 to WC3, and the contact points between each package and the outside (circles in Fig. 3) are used as measurement points. This means that nine measurement points are set up within one relay transmission device WC. Furthermore, the symptom detection location may be a unit of a device such as a relay transmission device, or may be a unit of a component within a relay transmission device.

[0021] FIG. 5 is a table showing the results of evaluating the deviation from the correlation line for the measured parameter values ​​of the optical transmission system NW. The correlation line is a graph line (see Fig. 8 for details) that shows the correlation between the Q value of the optical path and the degree of anomaly of the parameter measurement (hereinafter referred to as "degree of anomaly"). For example, in Fig. 4, (4 measurement points) x (4 parameter types) = 16 correlation lines are prepared in advance. The degree of abnormality is an index in which the larger the value, the more the part at which the parameter measurement value is measured deviates from the normal state, and the greater the deviation of the parameter measurement value from the normal value, the higher the degree of abnormality. The degree of abnormality may be calculated not only from the deviation of each parameter from the normal value, but also from the frequency at which each parameter deviates from the normal value within a certain unit time. In addition, when a parameter measurement value is input, an abnormality evaluation function that outputs the degree of abnormality is prepared in advance for each parameter type, and the prediction server SV evaluates the degree of abnormality for each parameter measurement value by calculating the abnormality evaluation function.

[0022] First, the row elements and column elements of the table in Fig. 5 are the same as those in the table in Fig. 4. However, the value in the table for each combination of row and column elements, which was "presence or absence of fluctuation in the measured parameter values" in Fig. 4, is replaced in Fig. 5 with "whether the measured parameter values ​​deviate from or match the correlation line obtained in advance." For example, in the table of Fig. 5, the value in the table indicates "match" only for the parameter type "OSNR (Optical Signal to Noise Ratio)" measured by the relay transmission device WB. This allows the prediction server SV to evaluate the cause of failure (sign of failure) related to OSNR at the sign detection location = relay transmission device WB using a correlation line.

[0023] For this reason, the prediction server SV creates a correlation line in advance based on the relationship between the Q value of the optical path and the parameter type. The process of creating the correlation line is performed by the prediction server SV, for example, by acquiring data in advance from a failure simulation system or using a QoT simulator (such as Gnpy). Acquiring data from a failure simulation system means, for example, acquiring parameter measurement data obtained by simulating the failure simulation patterns shown below. - Reduced output of AMP22 or WSS23 (reduction in single wavelength power or total wavelength power) Increase the output of AMP22 or WSS23 (increase the power of one wavelength or the power of all wavelengths) -WSS23 filter abnormality (waveform symmetry disturbance) AMP22 noise abnormality (OSNR decrease)

[0024] FIG. 6 is a flowchart showing an example of a procedure for operating the optical transmission system NW of FIG. A telecommunications carrier prepares a core / metro network system to which an optical cross connect (OXC) configuration is applied as the optical transmission system NW in FIG. 1 (S101). The optical transmission system NW detects optical paths with degraded signal quality and notifies the detection result to the prediction server SV (S102). S102 is, for example, a process in which the transponder TR4 detects degradation of the Q factor of the optical path P1, as described in (Procedure 1) of FIG. As explained in (Step 1) and (Step 2) of FIG. 1, the prediction server SV predicts a failure in the optical transmission system NW from the optical path detected in S102 (S103, see FIG. 7 for details). The prediction server SV notifies the failure prediction result (the part where the sign was detected and the time of occurrence of the failure) obtained in S103 to a maintenance terminal operated by a maintenance worker of the telecommunications carrier (S104). Upon receiving the notification in S104, the maintenance personnel performs preventive maintenance by switching the route of the faulty section or by replacing the faulty section (S105).

[0025] 7 is a flowchart showing details of the failure prediction process (S103) of the prediction server SV. This flowchart starts from a normal state where no failure has occurred in the optical transmission system NW. The prediction server SV determines whether the Q value of the optical path notified in S102 is less than the precursor detection threshold (FIG. 8) (S11, details in FIG. 8 and FIG. 9). If the answer is Yes in S11, proceed to S12. If the answer is No in S11, the Q value of the optical path is good, so return to the determination in S11 again. The prediction server SV checks whether the parameter measurement values ​​measured at the measurement points through which the optical path has passed match the correlation line for that parameter type in a brute force manner, as shown in the table in Fig. 5, and extracts the precursor detection part based on the parameter measurement values ​​of the matched parts (S12). For example, in Fig. 5, only the relay transmission device WB "matches", so the precursor detection part = relay transmission device WB.

[0026] On the other hand, when there are multiple matching locations in S12, the prediction server SV identifies the precursor detection location based on the most accurate one from the multiple locations and parameters. Of the multiple matching locations in S12, the upstream (starting point side) has the highest accuracy. For example, if the order of passage of optical path P1 is relay transmission device WA "deviation" → relay transmission device WB "deviation" → relay transmission device WC "match" → relay transmission device WD "match", the prediction server SV will determine the most upstream (starting point side) position (relay transmission device WC) of the matching locations (relay transmission devices WC, WD) as the precursor detection location. In addition, some parameter changes have a causal relationship, and there is a high degree of accuracy in identifying the location where the parameter that is the cause of the parameter change has changed. For example, if the full-wave power or single-wave power fluctuates, the OSNR in the subsequent stage will fluctuate, so if both parameters match, the location where the full-wave power or single-wave power has fluctuated is considered to be the location where the symptom was detected. As a result, the prediction server SV evaluated the areas where signs of failure could be detected based on the distribution of agreement or deviation between the parameter measurement values ​​and the correlation line.

[0027] The prediction server SV identifies the optical path with the smallest margin that passes through the error detection area among the optical paths detected by Yes in S11 (S21). The "margin" is the difference between the Q value of the measured optical path and the error occurrence boundary (FEC (Forward Error Correction) Limit) that is smaller than that Q value. That is, when there are multiple optical paths on the same route or sharing a portion of a route where Q-factor degradation has been detected, the prediction server SV focuses on the Q-factor of the optical path with the smallest margin that is expected to generate an error first. Note that optical paths change over time due to additions and deletions, so it is desirable to execute the process (S21) of identifying the optical path with the smallest margin every time there is a change in the optical path.

[0028] The prediction server SV acquires, for each measurement period (t seconds), the Q value for the optical path with the minimum margin in S21 and the parameter measurement values ​​at the precursor detection location extracted in S12 as a measurement pair (S22).The prediction server SV predicts the time of failure occurrence by predicting the trend of deterioration over time as an error prediction line from the measurement pair acquired in S22 (S23, see Fig. 10 for details).

[0029] The prediction server SV acquires a measurement pair for the optical path with the minimum margin, as in S22 (S31). The prediction server SV determines whether the impact of the failure indicated by the measurement pair acquired in S31 matches the error prediction line (S32, details in Figure 11). If the answer is Yes in S32, proceed to S33; if the answer is No in S32, cancel the notification of S34 as a false error and return to the judgment in S11. A false error is an event in which a failure was initially predicted, but the state in which the failure was predicted subsequently recovered. This allows the prediction server SV to prevent erroneous detections due to the match judgment with the error prediction line (S32).

[0030] The prediction server SV determines whether or not it is now z days (z=notification threshold) before the error occurs (S33). If Yes in S33, proceed to S34, and if No in S33, return to S31. The prediction server SV notifies the maintenance terminal of the maintenance technician of the failure prediction result at the symptom detection location (S34). The prediction server SV may also update the correlation line used in S12 and the error prediction line used in S32 based on feedback (correct or incorrect) of the prediction result from the maintenance terminal of the maintenance technician that notified it of the failure prediction result (S35). That is, the prediction server SV notifies the maintenance technician's maintenance terminal of a prediction of a failure at the precursor detection location and a prediction of the time of the failure before the time of the failure. The prediction server SV then obtains the correctness of the prediction from feedback information returned from the maintenance technician's maintenance terminal after the time of the failure, and updates the correlation data stored in the memory unit based on the correctness of the obtained prediction. The prediction server SV updates the error prediction data from information on the time-series change relationship between the measured and accumulated Q value and anomaly degree, which is stored in the memory unit together with the correlation data. This allows for improved accuracy of the correlation line and the error prediction line.

[0031] FIG. 8 is a graph showing the correlation line between the degree of abnormality of the parameter x and the Q value. The prediction server SV predicts the combination of the abnormality degree (E) of the parameter type = x (for example, full-wave power) and the Q value (Q) of the optical path.<E,Q> Let be a measurement pair measured at the same time. In Figure 8, the following measurement pairs are obtained by two measurements. First measurement pair p1<E1,Q1> However, Q1 = average value under normal conditions. Second measurement pair p2<E2,Q2> However, Q2=S11's sign detection threshold. Since both measurement pairs p1 and p2 of the prediction server SV are located on or near the correlation line Qf1, it is determined in S12 that the parameter measurement values ​​match the correlation line at this stage. Note that the proximity range that allows deviation from the correlation line that is determined to match may be adjusted as a threshold.

[0032] FIG. 9 is a graph showing the correlation line between the degree of abnormality of the parameter y and the Q value. The prediction server SV, as in Fig. 8, predicts the combination of the abnormality degree of the parameter type = y (for example, the power of one wave) and the Q value of the optical path.<E,Q> Let be a measurement pair measured at the same time. In Figure 9, the following measurement pairs are obtained by two measurements. First measurement pair p3<E3,Q3> However, Q3 is the average value under normal conditions. Second measurement pair p4<E4,Q4> However, Q4=S11's warning detection threshold. The prediction server SV is the measurement pair p4<E4,Q4> is a point on the correlation line Qf2<E5,Q4> Or, since the parameter measurement value is not located in the vicinity thereof, it is determined in S12 that the parameter measurement value deviates from the correlation line at this stage. Note that the vicinity range that allows deviation from the correlation line that is determined to match may be adjusted as a threshold value.

[0033] FIG. 10 is a time series graph showing an error prediction line when an error occurs. In Figure 10, the vertical axis of the upper graph is the Q value (the same as the vertical axis in Figures 8 and 9), and the horizontal axis of the lower graph is the degree of anomaly for each parameter (the same as the horizontal axis in Figures 8 and 9), with the upper and lower graphs corresponding to each other on the same time axis. Also, on the time axis, times t1, t2, t3, and t4 are past times when measurement pairs (shown by triangle icons) have already been measured, and time t5 is a future time when no measurement pair has yet been measured. For parameter measurement values ​​that match the correlation line as in Figure 8, the prediction server SV obtains measurement pairs multiple times from the third time onwards (a total of four times in Figure 10). Below, we will explain the processing of the prediction server SV, which uses two methods (a method using the upper graph in Figure 10 and a method using the lower graph) to predict deterioration and find the error occurrence boundary (error boundary threshold), and predict the time when the error occurrence boundary is reached as the time when a failure will occur.

[0034] In the graph at the top of Figure 10, the prediction server SV creates an error prediction line Ef1, for example, by extending the solid line of the graph with the same slope based on the trend of the measured Q value (in Figure 10, the solid line connecting the triangle icons).The prediction server SV then predicts the time of error occurrence according to the degree of quality degradation of the minimum-margin optical path passing through the symptom detection area. Specifically, the prediction server SV predicts that the time t5 when the error prediction line Ef1 falls below the Q value threshold (error boundary threshold Q9) at which an error bounds will be the time of failure occurrence.

[0035] That is, the prediction server SV measures the quality value of the optical signal at the receiving end point of the predetermined optical path multiple times in a period (times t3 to t4) after the first time t2. Based on the time-series change relationship of the measured quality value of the optical signal over time, the prediction server SV predicts the second time t5, at which the quality value of the optical signal is predicted to be less than the second threshold (error boundary threshold Q9), as the time of occurrence of a failure at the symptom detection site. In this way, there is a time-series change relationship between the change in the Q-factor and the change in time. Therefore, the time when a failure will occur can be predicted based on the time t5 when the quality (Q-factor) of the received signal of the optical path reaches its limit. For example, the time-series change relationship can be predicted by drawing an approximation line from the measured values ​​up to t4. Alternatively, it can be predicted from the time-series change relationship of the Q-factor of the same configuration that has been measured and accumulated in the past. Predictions from past measurement data can be made using statistical methods or machine learning. In this example, a case where the prediction is linear is described, but the time-series change relationship does not have to be linear.

[0036] In the graph at the bottom of Figure 10, the prediction server SV creates an error prediction line Ef2, for example, by extending the solid line of the graph at the same slope based on the trend of the measured abnormality level (in Figure 10, the solid line connecting the triangle icons).The prediction server SV then predicts the time of error occurrence according to the progression of the abnormality level in the precursor detection area. Specifically, the prediction server SV may predict the time t6 when the error prediction line Ef1 exceeds the abnormality level threshold (error boundary threshold E9) that bounds the error, as the time of failure occurrence. Note that E1 = the average normal value of the abnormality level, and E2 = the precursor detection threshold of the abnormality level.

[0037] That is, the prediction server SV measures the parameter measurements at the precursor detection location multiple times during the period (times t3 to t4) after the first time t2. Based on the time-series change relationship of the degree of abnormality over time determined from the parameter measurements at the precursor detection location, the prediction server SV predicts the third time t6, at which the degree of abnormality at the precursor detection location is predicted to fall below the third threshold (error boundary threshold E9), as the time t6 at which the failure will occur at the precursor detection location. For example, the time-series change relationship can be predicted by drawing an approximation line from the measurements up to t4. Alternatively, the prediction can be made from the time-series change relationship of the degree of abnormality for the same parameters and configuration that have been measured and accumulated in the past. Predictions from past measurement data can be made using statistical methods or machine learning. In this example, a case where a linear prediction is made is described, but the time-series change relationship does not have to be linear. As such, there is a time-series change relationship between the change in the anomaly level and the change in time. Therefore, as shown in Figure 8, the time of failure can be predicted based on the time t6 when the anomaly level measured at the predictive detection point that matches the correlation line reaches its limit. Furthermore, because the anomaly level is a factor directly linked to the failure that has occurred, it is possible to improve accuracy compared to predictions based solely on the Q value. Furthermore, while the Q value will deteriorate in line with the predicted time-series change relationship, it is possible that the anomaly level will not deteriorate in line with the time-series change relationship. In this case, it is possible that the point initially extracted based on the correlation match has deviated, and a different point now matches the correlation. Therefore, it is possible to switch back to checking the correlation between the measured values ​​of each parameter on the optical path and predicting a failure in another point.

[0038] Figure 11 is a graph showing the error prediction line when a false error occurs. A false error is when the time of a failure is predicted but the failure does not actually occur. In Figures 10 and 11, the axes of the two upper and lower graphs and the measurement pair up to time t4 are the same. On the other hand, in Figure 11, the condition of the measurement pair improves after time t4, the Q value of the measurement pair increases (the measured Q value line Ef1a slopes upward to the right), and the anomaly level of the measurement pair decreases (the measured anomaly level line Ef2a slopes downward to the right). The measured Q value line Ef1a indicates the relationship between the change in the Q value and the change in time. Furthermore, the measured anomaly level line Ef2a indicates the relationship between the change in the anomaly level and the change in time. One example of a cause of such an instantaneous improvement in the measurement pair is when a maintenance technician temporarily touches the optical fiber (fiber touch). On the other hand, examples of a failure that does not improve the Q factor include noise degradation in the AMP 22, a filter failure in the WSS 23, or an output control failure in the optical device 13.

[0039] In the upper graph of Figure 11, the prediction server SV withdraws the sign of failure (time of occurrence of failure) predicted at time t4 based on the deviation between the measured Q-value line Ef1a and the Q-value error prediction line Ef1 after time t4, and determines that no failure will occur. That is, after predicting the failure occurrence time, the prediction server SV continues to measure the quality value of the optical signal at the receiving end point of the specified optical path. If the time-series change relationship of the measured optical signal quality value over time (actual measurement line Ef1a) deviates from the time-series change relationship at the time when the failure occurrence time is predicted (error prediction line Ef1) (for example, if the difference in slope between the two lines is 60 degrees or more), or if the measured optical signal quality value recovers to or above the first threshold (Q2), the prediction server SV cancels the prediction of the failure at the precursor detection site and the prediction of the failure occurrence time at the second time t5.

[0040] In the graph at the bottom of Figure 11, the prediction server SV withdraws the sign of failure (time of occurrence of failure) predicted at time t4 depending on the deviation between the actual measurement line Ef2a of the abnormality degree after time t4 and the error prediction line Ef2 of the abnormality degree, and determines that a failure will not occur. That is, after predicting the time of failure occurrence, the prediction server SV continues to measure the parameter measurement values ​​at the precursor detection location. If the time-series change relationship of the degree of abnormality over time obtained from the measured parameter measurement values ​​(actual measurement line Ef2a) deviates from the time-series change relationship at the time of predicting the time of failure occurrence (error prediction line Ef2) (for example, if the difference in slope between the two lines is 60 degrees or more), the prediction server SV cancels the prediction of a failure at the precursor detection location and the prediction of the time of failure occurrence at the third time t6. The failure observation period before the prediction of the time of failure is canceled is, for example, the period from time t2, when the Q value falls below the precursor detection threshold in Figure 10, to the time when the failure occurrence time t5 or t6 is notified to the maintenance terminal of the maintenance personnel. This reduces the impact of false errors.

[0041] [effect] In the present invention, the prediction server SV has a storage unit and a control unit, The storage unit stores correlation data (correlation lines) between the quality values ​​of the optical signals measured at the receiving end points of the optical paths in the optical transmission system NW and the abnormality degrees of each type of optical physical property parameter measured at the passing points of the optical paths, The control unit When a predetermined optical path is detected in which the quality value of the optical signal measured at the first time is less than a first threshold, an abnormality degree at the first time is calculated from parameter measurement values ​​of optical physical property parameters measured at a passing point of the predetermined optical path at the first time; When the correlation between the quality value of the optical signal at the first time and the degree of abnormality at the first time matches the correlation data stored in the memory unit, it is predicted that a sign of failure will occur at a time after the first time at the sign detection location determined from the measurement point of the parameter measurement value.

[0042] This allows for the correlation between changes in optical signal quality and changes in measured parameters to be taken into account, thereby enabling the detection range of faults in optical transmission systems to be narrowed down to predictive detection areas with high accuracy compared to methods that only observe changes in measured parameters.

[0043] The present invention is characterized in that the control unit measures the quality value of an optical signal at the receiving end point of a specified optical path multiple times in a period after a first time, and based on the time series change relationship of the measured quality value of the optical signal over time, predicts a second time at which the quality value of the optical signal is predicted to be less than a second threshold as the time at which a failure will occur at the precursor detection site.

[0044] This makes it possible to predict the time at which the quality value (Q value) of the optical signal reaches its limit as the time at which a failure will occur.

[0045] The present invention is characterized in that, after predicting the time of failure occurrence, the control unit continues to measure the quality value of the optical signal at the receiving end point of a specified optical path, and if the time series change relationship of the measured quality value of the optical signal over time deviates from the time series change relationship at the time when the time of failure occurrence was predicted, or if the quality value of the measured optical signal recovers to or above the first threshold, the control unit cancels the prediction of a failure at the precursor detection site and the prediction of the time of failure occurrence at the second time.

[0046] This allows false errors such as fiber touch to be properly excluded from failure prediction, improving the accuracy of failure prediction.

[0047] The present invention is characterized in that the control unit measures parameter measurements at the precursor detection area multiple times in a period after the first time, and based on the time series change relationship of the degree of abnormality over time obtained from the parameter measurements at the precursor detection area, predicts a third time at which the degree of abnormality at the precursor detection area is predicted to be less than a third threshold as the time at which a failure will occur at the precursor detection area.

[0048] This makes it possible to predict the time at which the degree of abnormality at the symptom detection location reaches its limit as the time at which a failure will occur.

[0049] The present invention is characterized in that, after predicting the time of failure occurrence, the control unit continues to measure the parameter measurement values ​​at the precursor detection location, and if the time series change relationship over time of the abnormality degree obtained from the measured parameter measurement values ​​deviates from the time series change relationship at the time when the failure occurrence time was predicted, the control unit cancels the prediction of the failure at the precursor detection location and the prediction of the failure occurrence time at the third time.

[0050] This allows false errors such as fiber touch to be properly excluded from failure prediction, improving the accuracy of failure prediction.

[0051] The present invention is characterized in that the control unit notifies the maintenance staff's maintenance terminal of a prediction of a failure at the precursor detection site and a prediction of the time when the failure will occur before the time when the failure will occur, obtains the accuracy of the prediction from feedback information returned from the maintenance staff's maintenance terminal after the time when the failure will occur, and based on the accuracy of the obtained prediction, updates the correlation data stored in the memory unit, the time series change relationship of the quality value of the optical signal over time, and the time series change relationship of the degree of abnormality calculated from parameter measurement values ​​at the precursor detection site over time.

[0052] This improves the accuracy of correlation data (correlation line) and prediction of time-series changes in optical signal quality and abnormality level (error prediction line). [Explanation of symbols]

[0053] 10 packages 11 OTN Framer 12 DSP 13 Optical Devices 21 Photophysical property monitor 22 AMP 23 WSS L1~L6 Link NW optical transmission system SV prediction server (failure prediction device) TS1, TS1, TR1, TR2, TR3, TR4 transponders WA,WB,WC,WD,WE,WF Relay transmission equipment

Claims

1. The failure prediction device includes a storage unit and a control unit, the storage unit stores correlation data between a quality value of an optical signal measured at a receiving end point of an optical path in an optical transmission system and an abnormality degree for each type of optical physical property parameter measured at a passing point of the optical path; The control unit when detecting a predetermined optical path in which a quality value of an optical signal measured at a first time is less than a first threshold, calculating an abnormality degree at the first time from parameter measurement values ​​of the optical physical property parameters measured at a passing point of the predetermined optical path at the first time; When the correlation between the quality value of the optical signal at the first time and the degree of anomaly at the first time matches the correlation data stored in the storage unit, it is predicted that a sign of a failure will occur at a time after the first time in a sign detection portion determined from the measurement points of the parameter measurement values. Failure prediction device.

2. The control unit measures a quality value of the optical signal at a receiving end point of the predetermined optical path a plurality of times in a period after the first time, and predicts a second time at which the quality value of the optical signal is predicted to be less than a second threshold value as a time at which a failure will occur in the precursor detection portion, based on a time-series change relationship of the measured quality value of the optical signal over time. The failure prediction device according to claim 1 .

3. After predicting the failure occurrence time, the control unit continues to measure the quality value of the optical signal at the receiving end point of the predetermined optical path, and when a time-series change relationship of the measured quality value of the optical signal over time deviates from a time-series change relationship at the time when the failure occurrence time is predicted, or when the quality value of the measured optical signal recovers to or above the first threshold, cancels the prediction of the failure at the symptom detection site and the prediction of the failure occurrence time at the second time. The failure prediction device according to claim 2 .

4. The control unit measures parameter measurements at the precursor detection portion a plurality of times in a period after the first time, and predicts a third time at which the degree of abnormality at the precursor detection portion is predicted to become less than a third threshold based on a time-series change relationship over time of the degree of abnormality determined from the measured parameter measurements at the precursor detection portion, as a time at which a failure will occur at the precursor detection portion. The failure prediction device according to claim 1 .

5. After predicting the time of failure occurrence, the control unit continues to measure the parameter measurements at the symptom detection portion, and when a time-series change relationship of the degree of abnormality over time calculated from the measured parameter measurements deviates from the time-series change relationship at the time of predicting the time of failure occurrence, cancels the prediction of the failure at the symptom detection portion and the prediction of the time of failure occurrence at the third time. The failure prediction device according to claim 4 .

6. The control unit notifies a maintenance terminal of a prediction of a failure at the symptom detection site and a prediction of the time when the failure will occur before the time when the failure will occur, obtains whether the prediction is correct or incorrect from feedback information returned from the maintenance terminal after the time when the failure will occur, and updates the correlation data stored in the storage unit, the time-series change relationship of the quality value of the optical signal over time, and the time-series change relationship of the degree of abnormality calculated from parameter measurements at the symptom detection site over time, based on the obtained prediction. The failure prediction device according to claim 2 .

7. The failure prediction device includes a storage unit and a control unit, the storage unit stores correlation data between a quality value of an optical signal measured at a receiving end point of an optical path in an optical transmission system and an abnormality degree for each type of optical physical property parameter measured at a passing point of the optical path; The control unit when detecting a predetermined optical path in which a quality value of an optical signal measured at a first time is less than a first threshold, calculating an abnormality degree at the first time from parameter measurement values ​​of the optical physical property parameters measured at a passing point of the predetermined optical path at the first time; When the correlation between the quality value of the optical signal at the first time and the degree of anomaly at the first time matches the correlation data stored in the storage unit, it is predicted that a sign of a failure will occur at a time after the first time in a sign detection portion determined from the measurement points of the parameter measurement values. Failure prediction methods.

8. A failure prediction program for causing a computer to function as the failure prediction device according to any one of claims 1 to 6.

Images