An AI-based intelligent fault diagnosis system for optical fiber cables

Abstract

An intelligent fault diagnosis system for optical fiber cables based on artificial intelligence is disclosed, relating to the field of fault diagnosis technology. The system constructs a distribution map of the optical fiber cable, dividing it into several monitoring segments. Acquisition units are set up to obtain data, including historical fault records, historical acquisition data, and data characteristics. Based on the historical fault records and acquisition data of different monitoring segments, a fault judgment model is constructed. Real-time acquisition data is obtained, and the fault judgment model is used to determine whether a fault exists and generate a fault judgment record. Fault judgment records with misjudgments and their environmental parameters are acquired, and a misjudgment evaluation model is constructed to assess whether the fault judgment record contains a misjudgment. This system can form an effective fault judgment mechanism and optimize it, thereby improving the accuracy of fault judgment.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, specifically an intelligent fault diagnosis system for optical fiber cables based on artificial intelligence. Background Technology

[0002] As a transmission medium in modern communication networks, optical fiber cables play an important role in communication systems. However, in practical applications, optical fiber cables are easily affected by the external environment, leading to various types of faults. Traditional methods for diagnosing optical fiber cable faults suffer from problems such as low efficiency, high false positive rate, and poor positioning accuracy, making it difficult to meet the high requirements of modern communication networks for diagnosing optical fiber cable faults.

[0003] In the existing technology, there are technical solutions for fault diagnosis of optical fiber cables using artificial intelligence technology. However, these solutions lack effective means for integrating multi-source data for comprehensive judgment. Moreover, existing technologies often only provide fault diagnosis solutions but lack means for evaluating the judgment results. In order to address the shortcomings of existing technologies, this invention provides an intelligent fault diagnosis system for optical fiber cables based on artificial intelligence. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent fault diagnosis system for optical fiber cables based on artificial intelligence.

[0005] The objective of this invention can be achieved through the following technical solution: an intelligent fault diagnosis system for optical fiber cables based on artificial intelligence, comprising the following modules:

[0006] The map building module is used to obtain the distribution information of optical fiber cables and build the corresponding distribution map, which divides the optical fiber cables into several monitoring segments.

[0007] The data acquisition module is used to set up different acquisition units in each monitoring section and acquire the corresponding acquisition data, acquire the historical fault records of the optical fiber cable and its historical acquisition data, and obtain the data characteristics of different historical fault records based on the historical acquisition data.

[0008] The fault diagnosis module is used to construct corresponding fault diagnosis models based on historical fault records and historical data collected from different monitoring segments, obtain current real-time data, use the fault diagnosis model to determine whether a fault exists, and generate fault diagnosis records.

[0009] The misjudgment assessment module is used to obtain fault judgment records with misjudgment status and their environmental parameters, and to construct a corresponding misjudgment assessment model. The misjudgment assessment model is used in combination with environmental parameters to assess whether subsequent fault judgment records are misjudged.

[0010] Furthermore, the process of acquiring the distribution information of optical fibers and cables and constructing a corresponding distribution map, and dividing the optical fibers and cables into several monitoring segments in the distribution map, includes:

[0011] The distribution information refers to various data related to the distribution of optical fibers and cables, including geographical location information, cable specifications, and network topology information. The optical fibers and cables contain several key nodes.

[0012] The key nodes refer to the starting point, ending point, relay station, and junction box in the optical fiber cable. Using GIS technology, a distribution map of the optical fiber cable is constructed based on the distribution information. The optical fiber cable is divided into different monitoring segments according to the key nodes, and the optical fiber cable between adjacent key nodes is considered as a monitoring segment.

[0013] Furthermore, the process of setting up different acquisition units in each monitoring segment and acquiring the corresponding data includes:

[0014] A type I acquisition unit and a type II acquisition unit are set up on each monitoring segment. The type I acquisition unit collects the optical power difference of the corresponding monitoring segment in real time, and the type II acquisition unit collects the network performance parameters of the corresponding monitoring segment in real time, including bit error rate, transmission rate and bandwidth utilization.

[0015] The acquisition unit includes a first-class acquisition unit and a second-class acquisition unit, and the acquired data includes optical power difference, bit error rate, transmission rate, and bandwidth utilization.

[0016] Furthermore, the process of acquiring historical fault records and historical data of optical fiber cables, and obtaining data characteristics of different historical fault records based on the historical data, includes:

[0017] The historical fault records refer to data related to faults that have occurred in the optical fiber cable, including the time of the fault, the location of the fault, the type of fault, and the cause of the fault.

[0018] Based on the location of the fault, all historical fault records are uploaded to the distribution map for synchronization, and the total number of historical fault records for each monitoring segment is taken as the number of existing faults in the corresponding monitoring segment.

[0019] All data collected within a fixed time period before the fault time corresponding to the monitoring segment of a single historical fault record will be used as the historical data collected for that historical fault record.

[0020] The optical power difference, bit error rate, transmission rate, and bandwidth utilization of the historical data collected from individual historical fault records are used as their data features, including mean, variance, and trend.

[0021] Furthermore, the process of constructing corresponding fault judgment models based on historical fault records and historical data collected from different monitoring segments includes:

[0022] Based on the historical data and data characteristics corresponding to different historical fault records in different monitoring sections, a fault judgment set is generated, and the fault judgment set is divided into a first training set and a first test set.

[0023] Construct a first convolutional neural network by using different historical data and data features from the first training set as input data and whether a fault will occur as output data. Train the first convolutional neural network to obtain an initial first convolutional neural network.

[0024] The initial first convolutional neural network is validated using the first test set. The initial first convolutional neural network whose output is less than or equal to the preset first test error threshold is used as the fault judgment model.

[0025] Furthermore, the process of acquiring current real-time collected data, using a fault diagnosis model to determine whether a fault exists, and generating a fault diagnosis record includes:

[0026] All data collected within a fixed time period before the current moment in each monitoring segment are taken as real-time data, and the data characteristics of the real-time data are obtained, including mean, variance, and trend.

[0027] The single real-time collected data and its data characteristics are input into the fault judgment model, and the fault judgment model is used to determine whether there is a fault in the corresponding monitoring section;

[0028] If a fault judgment record exists, it will be generated for the corresponding monitoring segment and fed back to the relevant personnel; if it does not exist, no other operations will be performed.

[0029] Furthermore, the process of obtaining fault judgment records and their environmental parameters that exhibit misjudgment states, and constructing corresponding misjudgment assessment models, includes:

[0030] When a fault judgment record is generated, an optical time domain reflectometer is used to detect the monitoring segment where a fault is detected to obtain the fault location and fault type. If no fault is detected, the corresponding fault judgment record is marked as a misjudgment.

[0031] Obtain the environmental parameters and number of faults corresponding to the fault judgment record marked as a misjudgment state. The environmental parameters and number of faults refer to the temperature, humidity, pressure, pH, electromagnetic radiation and number of existing faults of the corresponding monitoring segment when the fault judgment record marked as a misjudgment state is generated.

[0032] Based on the historical data, data characteristics, environmental parameters, and number of faults corresponding to different fault judgment records marked as misjudged states, a misjudgment evaluation set is generated, and the misjudgment evaluation set is divided into a second training set and a second test set.

[0033] Construct a second convolutional neural network by using different historical data, data features, environmental parameters, and number of faults from the second training set as input data and whether misjudgment will occur as output data. Train the second convolutional neural network to obtain an initial second convolutional neural network.

[0034] The initial second convolutional neural network is validated using the second test set. The initial second convolutional neural network whose output is less than or equal to the preset second test error threshold is used as the misjudgment evaluation model.

[0035] Furthermore, the process of using a misjudgment assessment model in conjunction with environmental parameters to evaluate whether subsequent fault judgment records contain misjudgments includes:

[0036] When a fault judgment record is generated subsequently, the environmental parameters and number of faults corresponding to the fault judgment record are obtained. These are then combined with the real-time data and data characteristics of the monitoring segment corresponding to the fault judgment record and input into the misjudgment assessment model. The misjudgment assessment model is then used to assess whether the fault judgment record is misjudged.

[0037] If it exists, a misjudgment signal is generated for the corresponding fault judgment record, and the misjudgment signal and the fault judgment record are fed back to the relevant personnel. If it does not exist, no other operation is performed.

[0038] Compared with the prior art, the beneficial effects of the present invention are:

[0039] This invention divides optical fiber cables into different monitoring segments based on their key nodes, and extracts data features from historical fault records and historical data collected from different monitoring segments. This allows for the acquisition of changes in the acquired data of the optical fiber cable before each fault occurs, thereby constructing a fault judgment model. This facilitates the determination of whether a fault will occur in the corresponding monitoring segment based on changes in real-time acquired data, thus forming an effective fault judgment mechanism.

[0040] By evaluating the accuracy of each fault diagnosis record and combining the environmental parameters and number of faults in fault diagnosis records with misjudgments, we can identify potential influencing factors that lead to misjudgments and construct corresponding misjudgment assessment models. This is beneficial for evaluating whether subsequent fault diagnosis records have misjudgments, further optimizing the formed fault diagnosis mechanism, improving the accuracy of fault diagnosis, and enabling timely detection and feedback of fault situations. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the present invention. Detailed Implementation

[0042] like Figure 1 As shown, an intelligent fault diagnosis system for optical fiber cables based on artificial intelligence includes the following modules:

[0043] The map building module is used to obtain the distribution information of optical fiber cables and build the corresponding distribution map, which divides the optical fiber cables into several monitoring segments.

[0044] The data acquisition module is used to set up different acquisition units in each monitoring section and acquire the corresponding acquisition data, acquire the historical fault records of the optical fiber cable and its historical acquisition data, and obtain the data characteristics of different historical fault records based on the historical acquisition data.

[0045] The fault diagnosis module is used to construct corresponding fault diagnosis models based on historical fault records and historical data collected from different monitoring segments, obtain current real-time data, use the fault diagnosis model to determine whether a fault exists, and generate fault diagnosis records.

[0046] The misjudgment assessment module is used to obtain fault judgment records with misjudgment status and their environmental parameters, and to construct a corresponding misjudgment assessment model. The misjudgment assessment model is used in combination with environmental parameters to assess whether subsequent fault judgment records are misjudged.

[0047] It should be further explained that, in the specific implementation process, the process of obtaining the distribution information of optical fibers and cables and constructing the corresponding distribution map, and dividing the optical fibers and cables into several monitoring segments in the distribution map, includes:

[0048] The distribution information refers to various data related to the distribution of optical fibers and cables, including geographical location information, cable specifications, network topology information, etc.

[0049] The geographical location information refers to the latitude and longitude coordinates and altitude of each key node of the optical fiber cable. The key node refers to the starting point, ending point, repeater station, and junction box in the optical fiber cable. The optical cable specification information refers to the optical fiber cable type, core count, and outer sheath material between adjacent key nodes. The network topology information refers to the connection method and sequence and topology structure between each key node.

[0050] Using GIS technology, a distribution map of optical fiber cables is constructed based on the acquired geographic location information, optical cable specification information, and network topology information. The optical fiber cables are divided into different monitoring segments according to each key node, and the optical fiber cables between adjacent key nodes are considered as one monitoring segment.

[0051] It should be further explained that, in the specific implementation process, the process of setting up different acquisition units in each monitoring segment and acquiring the corresponding data includes:

[0052] Taking a single monitoring segment as an example, a first-class acquisition unit and a second-class acquisition unit are set up on the single monitoring segment. The optical power difference of the single monitoring segment is collected in real time through the first-class acquisition unit. The optical power difference refers to the difference between the optical signal power output from the transmitting end and the optical signal power received by the receiving end.

[0053] The transmitting end refers to the starting point of the single monitoring segment relative to the direction of optical signal transmission, and the receiving end refers to the ending point of the single monitoring segment relative to the direction of optical signal transmission.

[0054] The network performance parameters of the single monitoring segment are collected in real time by the two types of acquisition units, including bit error rate, transmission rate, bandwidth utilization, etc. The network performance parameters are used to reflect the operating status of the optical fiber cable of each monitoring segment when transmitting data.

[0055] The acquisition unit includes a first-class acquisition unit and a second-class acquisition unit. The acquired data includes optical power difference, bit error rate, transmission rate, bandwidth utilization, etc. The same method is used to set up first-class acquisition units and second-class acquisition units in each monitoring segment, and acquire the corresponding acquisition data respectively.

[0056] It should be further explained that, in the specific implementation process, the process of obtaining historical fault records and historical data of optical fiber cables, and obtaining the data characteristics of different historical fault records based on the historical data, includes:

[0057] The historical fault records refer to data related to faults that have occurred in the optical fiber cable, including the time of the fault, the location of the fault, the type of fault, and the cause of the fault.

[0058] Based on the location of the historical fault records, each historical fault record is uploaded to the corresponding monitoring segment in the distribution map for synchronization. The total number of historical fault records that have occurred in a single monitoring segment is obtained in the distribution map and used as the number of existing faults in that single monitoring segment.

[0059] All data collected within a fixed time period before the fault time corresponding to the monitoring segment of the historical fault record are used as the historical data of that historical fault record, including optical power difference, bit error rate, transmission rate, bandwidth utilization, etc.

[0060] Since all historical data are quantifiable, they can be analyzed to obtain their characteristic information, including mean, variance, trend, etc. The data characteristics include the characteristic information of each historical data, and the data characteristics corresponding to each historical fault record are obtained and bound to the corresponding monitoring segment.

[0061] The mean is used to reflect the average level of the data, the variance is used to reflect the dispersion of the data relative to the mean, and the trend is used to reflect the direction of the data change over time, including upward trend, downward trend, and horizontal trend.

[0062] It should be further explained that, in the specific implementation process, the process of constructing corresponding fault judgment models based on historical fault records and historical data collected from different monitoring sections includes:

[0063] A fault judgment set is generated based on the historical data and data characteristics corresponding to different historical fault records in different monitoring segments, and the fault judgment set is divided into a first training set and a first test set.

[0064] Construct a first convolutional neural network by using different historical data and data features from the first training set as input data and whether a fault will occur as output data. Train the first convolutional neural network to obtain an initial first convolutional neural network.

[0065] The initial first convolutional neural network is validated using the first test set. The initial first convolutional neural network whose output is less than or equal to the preset first test error threshold is used as the corresponding fault judgment model.

[0066] It should be further explained that, in the specific implementation process, the process of acquiring the current real-time collected data, using the fault diagnosis model to determine whether a fault exists, and generating a fault diagnosis record includes:

[0067] Taking a single monitoring segment as an example, all the data collected within a fixed time period before the current moment of the single monitoring segment is used as its current real-time data, including optical power difference, bit error rate, transmission rate, bandwidth utilization, etc.

[0068] The same method is used to obtain the data characteristics of each real-time data collection, including mean, variance, trend, etc. The real-time data collection of the single monitoring segment and its data characteristics are input into the constructed fault judgment model, and the fault judgment model is used to determine whether there is a fault in the single monitoring segment.

[0069] If a fault exists, a corresponding fault judgment record is generated for that single monitoring segment and fed back to the relevant personnel. If a fault does not exist, no other operations are performed. The same method is used to determine whether each monitoring segment has a fault and generate a corresponding fault judgment record for feedback.

[0070] It should be further explained that, in the specific implementation process, the process of obtaining fault judgment records and their environmental parameters where misjudgment exists, and constructing the corresponding misjudgment assessment model, includes:

[0071] When a fault judgment record is generated, relevant personnel use an optical time domain reflectometer to detect the monitoring segment where a fault is suspected to exist to obtain the fault location and fault type. If no fault is detected, the corresponding fault judgment record is marked as a misjudgment.

[0072] Obtain the environmental parameters and number of faults corresponding to the fault judgment record marked as a misjudgment state. The environmental parameters refer to the environmental parameters of the corresponding monitoring segment when the fault judgment record marked as a misjudgment state is generated, including temperature, humidity, pressure, acidity, alkalinity, electromagnetic radiation, etc. The number of faults refers to the number of existing faults in the corresponding monitoring segment when the fault judgment record marked as a misjudgment state is generated.

[0073] A misjudgment assessment set is generated based on the historical data, data characteristics, environmental parameters, and number of faults corresponding to different fault judgment records marked as misjudged states. The misjudgment assessment set is then divided into a second training set and a second test set.

[0074] Construct a second convolutional neural network by using different historical data, data features, environmental parameters, and number of faults from the second training set as input data and whether misjudgment will occur as output data. Train the second convolutional neural network to obtain an initial second convolutional neural network.

[0075] The initial second convolutional neural network is validated using the second test set. The initial second convolutional neural network whose output is less than or equal to the preset second test error threshold is used as the corresponding misjudgment evaluation model.

[0076] It should be further explained that, in the specific implementation process, the process of using the misjudgment assessment model in conjunction with environmental parameters to assess whether there are misjudgments in the subsequent fault judgment records includes:

[0077] When a fault judgment record is generated subsequently, the environmental parameters and number of faults corresponding to the fault judgment record are obtained, and the real-time collected data and data characteristics of the monitoring segment corresponding to the fault judgment record are input into the constructed misjudgment assessment model.

[0078] The misjudgment assessment model is used to evaluate whether there is a misjudgment in the fault judgment record. If there is, a corresponding misjudgment signal is generated for the fault judgment record, and the generated misjudgment signal and the fault judgment record are fed back to the relevant personnel. If there is no misjudgment, no other operation is performed.

[0079] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An intelligent fault diagnosis system for optical fiber cables based on artificial intelligence, characterized in that, Includes the following modules: The map building module is used to obtain the distribution information of optical fiber cables and build the corresponding distribution map, which divides the optical fiber cables into several monitoring segments. The data acquisition module is used to set up different acquisition units in each monitoring section and acquire the corresponding acquisition data, acquire the historical fault records of the optical fiber cable and its historical acquisition data, and obtain the data characteristics of different historical fault records based on the historical acquisition data. The fault diagnosis module is used to construct corresponding fault diagnosis models based on historical fault records and historical data collected from different monitoring segments, obtain current real-time data, use the fault diagnosis model to determine whether a fault exists, and generate fault diagnosis records. The misjudgment assessment module is used to obtain fault judgment records with misjudgment status and their environmental parameters, and to build a corresponding misjudgment assessment model. The misjudgment assessment model is used in combination with environmental parameters to assess whether subsequent fault judgment records are misjudged. The process of obtaining fault judgment records and their environmental parameters that contain misjudgments, and constructing a misjudgment assessment model, includes: When a fault judgment record is generated, an optical time domain reflectometer is used to detect the monitoring segment where a fault is detected to obtain the fault location and fault type. If no fault is detected, the corresponding fault judgment record is marked as a misjudgment. Obtain the environmental parameters and number of faults corresponding to the fault judgment record marked as a misjudgment state. The environmental parameters and number of faults refer to the temperature, humidity, pressure, pH, electromagnetic radiation and number of existing faults of the corresponding monitoring segment when the fault judgment record marked as a misjudgment state is generated. Based on the historical data, data characteristics, environmental parameters, and number of faults corresponding to different fault judgment records marked as misjudged states, a misjudgment evaluation set is generated, and the misjudgment evaluation set is divided into a second training set and a second test set. Construct a second convolutional neural network by using different historical data, data features, environmental parameters, and number of faults from the second training set as input data and whether misjudgment will occur as output data. Train the second convolutional neural network to obtain an initial second convolutional neural network. The initial second convolutional neural network is validated using the second test set. The initial second convolutional neural network whose output is less than or equal to the preset second test error threshold is used as the misjudgment evaluation model.

2. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 1, characterized in that, The process of constructing a distribution map and dividing the fiber optic cable into several monitoring segments includes: The distribution information refers to various data related to the distribution of optical fibers and cables, including geographical location information, cable specifications, and network topology information. The optical fibers and cables contain several key nodes. The key nodes refer to the starting point, ending point, relay station, and junction box in the optical fiber cable. Using GIS technology, a distribution map of the optical fiber cable is constructed based on the distribution information. The optical fiber cable is divided into different monitoring segments according to the key nodes, and the optical fiber cable between adjacent key nodes is considered as a monitoring segment.

3. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 2, characterized in that, The process of setting up the acquisition unit and acquiring the data includes: A type I acquisition unit and a type II acquisition unit are set up on each monitoring segment. The type I acquisition unit collects the optical power difference of the corresponding monitoring segment in real time, and the type II acquisition unit collects the network performance parameters of the corresponding monitoring segment in real time, including bit error rate, transmission rate and bandwidth utilization. The acquisition unit includes a first-class acquisition unit and a second-class acquisition unit, and the acquired data includes optical power difference, bit error rate, transmission rate, and bandwidth utilization.

4. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 3, characterized in that, The process of obtaining historical fault records, their historical data, and data characteristics includes: The historical fault records refer to data related to faults that have occurred in the optical fiber cable, including the time of the fault, the location of the fault, the type of fault, and the cause of the fault. Based on the location of the fault, all historical fault records are uploaded to the distribution map for synchronization, and the total number of historical fault records for each monitoring segment is taken as the number of existing faults in the corresponding monitoring segment. All data collected within a fixed time period before the fault time corresponding to the monitoring segment of a single historical fault record will be used as the historical data collected for that historical fault record. The optical power difference, bit error rate, transmission rate, and bandwidth utilization of the historical data collected from individual historical fault records are used as their data features, including mean, variance, and trend.

5. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 4, characterized in that, The process of building a fault diagnosis model includes: Based on the historical data and data characteristics corresponding to different historical fault records in different monitoring sections, a fault judgment set is generated, and the fault judgment set is divided into a first training set and a first test set. Construct a first convolutional neural network by using different historical data and data features from the first training set as input data and whether a fault will occur as output data. Train the first convolutional neural network to obtain an initial first convolutional neural network. The initial first convolutional neural network is validated using the first test set. The initial first convolutional neural network whose output is less than or equal to the preset first test error threshold is used as the fault judgment model.

6. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 5, characterized in that, The process of acquiring real-time collected data, determining whether a fault exists, and generating a fault determination record includes: All data collected within a fixed time period before the current moment in each monitoring segment are taken as real-time data, and the data characteristics of the real-time data are obtained, including mean, variance, and trend. The single real-time collected data and its data characteristics are input into the fault judgment model, and the fault judgment model is used to determine whether there is a fault in the corresponding monitoring section; If a fault judgment record exists, it will be generated for the corresponding monitoring segment and fed back to the relevant personnel; if it does not exist, no other operations will be performed.

7. The intelligent fault diagnosis system for optical fiber cables based on artificial intelligence according to claim 6, characterized in that, The process of assessing whether subsequent fault diagnosis records contain any misjudgments includes: When a fault judgment record is generated subsequently, the environmental parameters and number of faults corresponding to the fault judgment record are obtained. These are then combined with the real-time data and data characteristics of the monitoring segment corresponding to the fault judgment record and input into the misjudgment assessment model. The misjudgment assessment model is then used to assess whether the fault judgment record is misjudged. If it exists, a misjudgment signal is generated for the corresponding fault judgment record, and the misjudgment signal and the fault judgment record are fed back to the relevant personnel. If it does not exist, no other operation is performed.

Images