A method for optical port degradation analysis
By analyzing the optical port degradation of fiber optic storage switches based on the Bayesian method, a dynamic model was established, which solved the problems of fiber optic port degradation modeling and lifetime prediction, and achieved high-precision fault diagnosis and lifetime estimation.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-17
- Publication Date
- 2026-04-10
AI Technical Summary
Existing technologies have not yet statistically modeled and analyzed the degradation of fiber optic ports in fiber optic storage switches, resulting in low accuracy in fault diagnosis and an inability to predict the lifecycle of fiber optic port power degradation in fiber optic communication relay equipment.
A Bayesian-based approach is adopted to acquire data from fiber optic storage switches, perform optical port degradation analysis, establish a degradation model, and use a dynamic Bayesian algorithm to update parameters, accurately fit the degradation trend of optical port power, and estimate the remaining service life.
It enables accurate fault diagnosis and estimation of remaining service life, reduces costs, improves diagnostic accuracy, and provides a new approach to fiber optic port degradation modeling and lifetime prediction.
Smart Images
Figure CN115643506B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of fiber storage switches, in particular to a kind of optical port degradation analysis method. BACKGROUND
[0002] In the prior art about optical fiber port anomaly detection, one kind is to measure the length of optical fiber using a tester, and in the process of communication between the transmitting end and the receiving end through optical fiber, the current transmitting optical power and the current receiving optical power are read to obtain the optical fiber transmission distance between devices, and the measurement result is used for fault determination. If the optical fiber transmission distance is not within the threshold range, an abnormal prompt information is output. This kind of method can omit the process of manual intervention, shorten the fault positioning time and reduce the communication loss, but the accuracy is limited. Another kind is to convert the experience and knowledge of experts in solving fault problems into computer language to establish an expert knowledge base, and update the knowledge base in real time. Real-time power data obtained by a power measurer is input into a PC end and simulated to quickly locate the fault point. This kind of method has the advantages of real-time and scalability, but it is too dependent on expert knowledge and has a certain subjectivity. The existing technology has not yet statistically modeled and analyzed the optical fiber port degradation data of the fiber storage switch, and the fault diagnosis accuracy is not high. Moreover, the existing technology has not yet predicted the life cycle of the optical fiber port power degradation of the optical fiber communication relay device.
[0003] Therefore, the skilled in the art is committed to developing an optical port degradation analysis method, which solves the problem of optical fiber port power degradation analysis of the optical fiber communication relay device based on Bayesian, fully extracts the information contained in the existing data, accurately fits the actual degradation path, so that the model well describes the degradation trend of optical port power, thereby obtaining accurate fault diagnosis and estimation of remaining service life, and provides a new idea for degradation modeling and prediction of remaining service life for such problems. SUMMARY
[0004] To achieve the above-mentioned purpose, the present application provides an optical port degradation analysis method, comprising:
[0005] Step 1: obtaining fiber storage switch data, including the range of devices that need to perform degradation analysis, the type and parameters of optical port used, and the alarm threshold of different optical ports;
[0006] Step 2: obtaining monitoring data of optical port;
[0007] Step 3: caching optical port data;
[0008] Step 4: performing degradation analysis on each optical port based on dynamic Bayesian algorithm, and then obtaining the time remaining for each optical port to the occurrence of fault;
[0009] Step 5: outputting analysis data.
[0010] Further, the step 3 comprises: saving the data of the step 1 and the step 2 into a variable, a temporary file or a database in the way of a data frame, a dictionary, a multi-dimensional array or a tensor.
[0011] Further, the step 4 comprises:
[0012] Step 4.1 pre-processes the optical port data to obtain optical port degradation data X n ;
[0013] Step 4.2 establishes a degradation model and performs parameter updating based on a Bayesian method, wherein the degradation model is:
[0014] X n =A n b+ε(t)
[0015] wherein X n represents optical port degradation data; A n is a time-dependent parameter coefficient item matrix;
[0016] b is a model parameter to be estimated; and ε(t) represents an error of the model.
[0017] Step 4.3 estimates the remaining useful life:
[0018] T={t:X(t;b)≤L}
[0019] wherein T represents device life, L represents a failure threshold, and X(t;b) represents device degradation data at time t.
[0020] Further, the step 4.2 comprises:
[0021] Assuming that the error obeys a Gaussian distribution, ε(t) ~ N(0, σ 2 I n ), the optical port degradation data X n also obeys a Gaussian distribution, that is, X n ~ N(A n b, σ 2 I n ), each row of the time parameter matrix A n represents a time coefficient at each time; and the optical port degradation data X1, X2,..., X n at input t1, t2,..., t n , then X n has an expanded expression as follows:
[0022]
[0023] wherein tk The parameter vector b at time k = 1, 2, ..., n is b tk Let b be at time t1. t1 The prior distribution of the parameters is Mean vector Sum of variance matrix parameters All of these can be obtained by fitting the results of historical data, b tk The posterior probability distribution remains Gaussian; the parameter vector b is updated based on the dynamic Bayesian method; the output t is... n Time b tn The posterior probability distribution is in:
[0024]
[0025] Furthermore, the failure threshold is stored in a configuration file, a database, or transmitted from the outside via an interface.
[0026] Further, step 4.3 includes:
[0027] Input t n time: variance t n+l time: variance Based on the definition of remaining equipment lifespan and the natural constraint that equipment lifespan cannot be negative, the distribution expression for the remaining usable lifespan of equipment is obtained:
[0028]
[0029] in This represents the standard normal distribution function.
[0030] Further, step 5 includes:
[0031] If the remaining time obtained from the analysis in step 4 is less than the safe operating time width set by the system pre-configuration or configuration parameters, then the action or notification shall be executed according to the pre-configured strategy.
[0032] Furthermore, the action includes:
[0033] Conduct risk analysis and automate the migration of risky applications, middleware, and databases.
[0034] Furthermore, the notification includes:
[0035] Send notification emails; or record data in a database and generate reports.
[0036] Furthermore, the preprocessing includes:
[0037] First, the optical port data is smoothed using a smoothing function, then further filtered using a moving average filtering method, and finally normalized using a maximum-minimum normalization method.
[0038] The technical effects of the present application are as follows: based on the Bayesian method, the optical fiber port power degradation analysis problem of the optical fiber communication relay equipment is solved, the degradation problem of the optical port can be modeled from the statistical point of view, the information contained in the existing data is fully extracted, the actual degradation path is accurately fitted, the model well describes the degradation trend of the optical port power, so as to obtain accurate fault diagnosis and estimation result of the remaining service life, and based on the dynamic Bayesian method, real-time parameter updating is carried out, the problem of low fault diagnosis accuracy in the existing method is solved, and manual intervention is not required for operation, which is a low-cost, high-precision and operable method, and provides a new idea for degradation modeling and remaining service life prediction of such problems.
[0039] The concept, specific structure and technical effects of the present application will be further described below with reference to the accompanying drawings, so as to fully understand the purpose, features and effects of the present application. BRIEF DESCRIPTION OF DRAWINGS
[0040] Figure 1 is the analysis method flowchart of a preferred embodiment of the present application;
[0041] Figure 2 is the degradation method flowchart based on dynamic Bayesian of a preferred embodiment of the present application;
[0042] Figure 3 is the optical port power stable decline trend chart of the optical fiber storage switch;
[0043] Figure 4 is the optical port power accelerated decline trend chart of the optical fiber storage switch;
[0044] Figure 5 is the dynamic Bayesian idea flowchart;
[0045] Figure 6 is the fitting and prediction effect chart of the dynamic Bayesian method;
[0046] Figure 7 is the remaining service life estimation chart. DETAILED DESCRIPTION
[0047] The present application can be embodied in many different forms and the scope of protection of the present application is not limited to the embodiments described herein.
[0048] In the drawings, identical or similar components are denoted by identical or similar reference numbers throughout the several views. The size and thickness of each component shown in the drawings are arbitrary and are not intended to limit the scope of the present application. The thickness of some components is exaggerated in order to make the drawing clearer.
[0049] As shown in Figure 1 The present application provides a method for optical port degradation analysis, comprising the following steps:
[0050] Step 1: Obtain the data of fiber storage switch, including the range of equipment to be analyzed, the type and main parameters of optical port, such as manufacturer, port type, port rate, purchase batch, port mounting time, port device type, data center, etc. The obtained device information can be saved in the monitoring database or configuration management database for future use; the obtained optical port degradation data can be saved in a device, including database, file, etc. The data can be obtained and updated in a timed manner, or can be obtained each time the analysis is performed.
[0051] Step 2: Based on the equipment range of step 1, obtain the monitoring data of the corresponding optical port. The monitoring data can be obtained from the database. From the perspective of query performance optimization, the query process will perform batch queries, and a certain number of optical port monitoring data will be obtained in each cycle to perform subsequent analysis. The optical port data is collected by the monitoring system in a timely manner and saved to a device, such as a database, a file, etc. Preferably, such as a monitoring system database or a time series database that saves data through the monitoring system. This is a step that is executed continuously and in a timely manner, and is not limited by the application environment of the present application. The monitoring data enters the database, and the present application consumes the data in the database.
[0052] Step 3: Cache the fiber port data: save the data obtained in steps 1 and 2 to a variable or temporary file or database using a data frame or dictionary or multi-dimensional array or tensor, etc. After obtaining the data from the storage medium, the data is saved in different data structures in the program. At the same time, add the alarm threshold and optical port model data obtained in step 1 to the corresponding optical port degradation data in the form of a label, so as to facilitate subsequent analysis. The monitoring data is time series data, while the alarm threshold is usually static and unchanging. In each monitoring data, if saved in a DataFrame data frame, a column is added to record the corresponding alarm threshold. In this way, the alarm threshold is repeatedly recorded, but it is more convenient to compare whether it breaks through the threshold later. The reason for adding a column instead of using the value in a fixed variable is that from the operation and maintenance history data, the threshold may change with the optical port maintenance and replacement, and the threshold of different models of optical ports is different. Therefore, the alarm threshold that can be used at different times for the optical port at the same location may be different due to maintenance actions. The alarm threshold is usually obtained separately from the monitoring system. This is set by humans or programs. The monitoring data is collected by the collection program.
[0053] Step 4: Based on the results of step 3, perform degradation analysis on each optical port based on the dynamic Bayesian algorithm, and compare with the threshold obtained in steps 1 and 2 to obtain the remaining time to failure for each optical port.
[0054] Step 5: Output the analysis data. If the remaining time in step 4 is less than the safe operation time width set by the system pre-configuration or configuration parameters, perform the corresponding action or notification according to the pre-configuration strategy. For example, carry out corresponding risk analysis, and automatically migrate the risk-related applications, middleware and databases, etc.; or send a notification email; or record it in the database and generate related reports, etc. Step 5 is an optional step, which is selected or set according to the specific application environment.
[0055] Among them, step 4 is to establish a degradation model based on the dynamic Bayesian method, and update the model parameters in real time to dynamically fit the actual degradation path, so as to accurately judge the remaining service life of the equipment and provide early warning for the failure of the equipment. Specifically, as shown in Figure 2 Step 4 includes:
[0056] Step 4.1 Data preprocessing: The electrical characteristics of the optical port in the fiber storage switch determine that it can only work within a certain power range to transmit data normally. Once it exceeds this normal range, the communication performance and stability will be affected. As shown in Figure 3 and Figure 4The shown is a fiber port power monitoring data curve graph of two different degradation trends. The pre-processing of the optical port degradation power data data of the optical fiber storage switch is as follows: first, the data is smoothed by using a smoothing function, then the data is further filtered by using a moving average filtering method, and finally the filtered data is normalized by using a maximum and minimum normalization method, so as to obtain the optical port degradation data X n .
[0057]
[0058] Step 4.2: Establish a degradation model and perform parameter updating based on the Bayesian method: analyze the structural characteristics and failure modes of the equipment, and establish a degradation model according to the degradation law of the power parameters over time
[0059] X n = A n b + ε(t)
[0060] Wherein, X n represents the optical port degradation data; A n is a time-dependent parameter coefficient item matrix;
[0061] b is the model parameter to be estimated, that is, the model parameter to be solved by the dynamic Bayesian method in the following; ε(t) represents the error of the model;
[0062] Assuming that the error obeys a Gaussian distribution, ε(t) ~ N(0, σ 2 I n ), the optical port degradation data X n also obeys a Gaussian distribution, and the specific distribution is: X n ~ N(A n b, σ 2 I n ), each row of the time parameter matrix A n represents the time coefficient at each time. When the optical port degradation data X1, X2,..., X n at t1, t2,..., t n are input, X n is expanded as follows (in the following formula, A i represents each row):
[0063]
[0064] Let the parameter vector b at t k ,k=1,2,...,n be b tk , and let the parameter prior distribution of b t1 at t1 be The mean vector and the variance matrix parameter All of these can be obtained by fitting the results of historical data, b t1 The posterior probability distribution is equivalent to N(A1b) t1 ,σ 2 )and Multiplying them together still results in a Gaussian distribution. Let the result be... According to the Gaussian distribution multiplication theorem, we can obtain:
[0065]
[0066] Parameter b at time t1 t1 The posterior probability distribution is the parameter b at time t2. t2 Prior probability distribution Right now Based on the idea of dynamic Bayes, when the observed value X2 at time t2 is known, then b t2 The posterior is equivalent to N(A2b) t2 ,σ 2 )and Multiply, let the result be Then we have:
[0067]
[0068] As time progresses, the parameter b based on dynamic Bayesian theory continues to be updated. tk The posterior probability distribution remains Gaussian. The parameter update process described above can be summarized as follows: Figure 5 Output t n Time b tn The posterior probability distribution is in:
[0069] noise variance σ 2 It can be obtained by comparing the variance between the actual degradation data and the fitted degradation data.
[0070] Step 4.3 Estimate Remaining Service Life: The remaining service life of the equipment is considered to be a failure when the equipment's health condition first degrades to a level not exceeding a pre-set failure threshold. Mathematically, this can be expressed as T = {t: X(t; b) ≤ L}, where T represents the equipment lifespan, L represents the failure threshold, and X(t; b) represents the equipment degradation data at time t. The failure threshold is determined based on expert experience or obtained from the manufacturer's equipment specifications for different application environments. The failure threshold may be stored in a configuration file, database, or passed from an external source via an application programming interface (API). Input t n time: variance t n+l time: variance According to the definition of the remaining life of the device and the natural restriction that the life of the device cannot be negative, the distribution expression of the remaining usable life of the device is obtained:
[0071]
[0072] Wherein represents a standard normal distribution function, and the optical port power degradation data and related parameters of the optical fiber storage switch are substituted, that is, the optical fiber port data fitting and degradation prediction based on dynamic Bayesian, etc., as shown in Figure 6 At the same time, the prediction result of the remaining usable life of the device can be obtained, see Figure 7 , so as to realize real-time monitoring of the device. Since step 4.1 performs normalization processing, corresponding processing is required here in order to perform threshold comparison and evaluate the distance from the remaining time of the optical port failure.
[0073]
[0074] The preferred embodiments of the application are described in detail above. It should be understood that those skilled in the art can make many modifications and changes to the application without creative labor on the basis of the concept of the application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment on the basis of the prior art according to the concept of the application shall be within the protection scope determined by the claims.
Claims
1. A method of optical port degradation analysis, the method comprising: The application relates to a method for analyzing the remaining life of an optical port, comprising the following steps: Step 1: obtaining optical fiber storage switch data, including the device range needing to perform degradation analysis, the optical port type and parameters used, and the alarm threshold of different optical ports; Step 2: obtaining monitoring data of the optical port; Step 3: buffering the optical port data; Step 4: performing degradation analysis on each optical port based on a dynamic Bayesian algorithm, and then obtaining the time remaining before failure of the optical port; Step 5: outputting analysis data; The step 4 comprises the following steps: Step 4.1 Pre-process the optical port data to obtain optical port degradation data ; Step 4.2: establishing a degradation model and performing parameter updating based on the Bayesian method, wherein the degradation model is as follows: wherein, represents optical port degradation data; is a time-dependent parameter coefficient term matrix; are model parameters to be estimated; denotes the error of the model; Step 4.3: estimating the remaining service life: wherein, represents a device lifetime, represents a failure threshold, represents device degradation data at a time instant; The step 4.2 comprises the following steps: Assume that the error obeys Gaussian distribution, The optical port degradation data also obeys Gaussian distribution, which is: Each row of the time parameter matrix represents the time coefficient at each time; the input optical port degradation data at each time Therefore, The expanded expression is: Wherein, record Parameter vector at time t For , set Parameter prior distribution at time t Is Mean vector And variance matrix parameter Can be obtained by using the result of fitting historical data, The posterior probability distribution of t Is still Gaussian distribution; based on dynamic Bayesian method, parameter updating is carried out on the parameter vector Output The posterior probability distribution of t Is , ; The failure threshold is stored in a configuration file, a database or is transmitted from the outside through an interface; The step 4.3 comprises the following steps: Input Time: , variance ; Time: , variance , according to the definition of the remaining life of the device and the natural restriction that the life of the device cannot be negative, the distribution expression of the remaining usable life of the device is obtained: wherein denotes the standard normal distribution function.
2. The optical port degradation analysis method of claim 1, wherein, The step 3 comprises the following steps: the data of the step 1 and the step 2 are stored in variables, temporary files or databases in the form of data frames, dictionaries, multidimensional arrays or tensors.
3. The optical port degradation analysis method of claim 1, wherein, The step 5 comprises the following steps: If the time remaining obtained according to the step 4 is less than the safe operation time width set by system preparation or configuration parameters, an action or a notification is performed according to the preparation strategy.
4. The optical port degradation analysis method of claim 3, wherein, The action comprises the following steps: Risk analysis is carried out, and migration is automatically adopted for risk application, middleware and databases.
5. The optical port degradation analysis method of claim 3, wherein, The notification comprises the following steps: A notification email is sent, or is recorded in a database and a report is generated.
6. The optical port degradation analysis method of claim 1, wherein, The preprocessing comprises the following steps: First, a smoothing function is used to smooth the optical port data, then a sliding average filtering method is used for further filtering, and finally a maximum-minimum normalization method is used to normalize the filtered data.