Multimodal data-based anomaly detection method and system for in-service OPGW optical cable
By integrating multimodal data and an SVM classifier, the inaccuracy of traditional OPGW optical cable detection methods has been addressed, enabling more accurate fault identification and real-time monitoring, and improving the reliability and stability of the optical cable.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- HUBEI CENT CHINA TECH DEV OF ELECTRIC POWER
- Filing Date
- 2025-10-27
- Publication Date
- 2026-07-30
AI Technical Summary
Traditional OPGW optical cable anomaly detection methods rely on a single data source, resulting in insufficient detection accuracy and real-time performance, and failing to fully reflect the operating status of the optical cable.
By integrating multimodal data from different sensors, including Brillouin frequency shift parameters, wind speed, and tower distance and height difference, and combining them with an SVM classifier for feature extraction and decision-level fusion, fault identification can be achieved.
It improves the accuracy and reliability of OPGW optical cable anomaly detection, enables real-time monitoring and fault diagnosis, reduces maintenance time and costs, and extends the life of optical cables.
Smart Images
Figure CN2025130030_30072026_PF_FP_ABST
Abstract
Description
An abnormality detection method and system for in-service OPGW optical cable based on multi-modal data
[0001] Cross-reference to related applications
[0002] The present application claims priority to the Chinese patent application No. 2025101036598, filed on January 22, 2025, and entitled "An abnormality detection method and system for in-service OPGW optical cable based on multi-modal data", the entire contents of which are incorporated herein by reference. TECHNICAL FIELD
[0003] The present application relates to the field of optical fiber cable fault diagnosis, and specifically relates to an abnormality detection method and system for in-service OPGW optical cable based on multi-modal data. BACKGROUND
[0004] In modern power and communication systems, optical cables are widely used in the construction of overhead power lines and communication networks as an important transmission medium. Among them, the optical fiber composite overhead ground wire (OPGW) optical cable not only has good optical signal transmission characteristics, but also can provide additional lightning protection and power transmission capabilities. However, with the influence of environmental factors, natural disasters, and human activities, OPGW optical cables may experience various abnormalities during operation, such as optical fiber breakage, external damage, temperature changes, etc., which can seriously affect the stability and reliability of power and communication. Therefore, it is particularly important to detect abnormalities in OPGW optical cables in a timely and effective manner.
[0005] Traditional abnormality detection methods rely on a single data source, such as optical fiber transmission loss, temperature monitoring, or electromechanical sensors, which often cannot fully reflect the operating state of the optical cable, resulting in insufficient accuracy and real-time performance of the detection. SUMMARY
[0006] In view of the above, the present application provides an abnormality detection method for in-service OPGW optical cable based on multi-modal data, which can more comprehensively evaluate the health status of the optical cable by integrating data from different sensors, thereby improving the accuracy and reliability of abnormality detection.
[0007] To achieve the above-mentioned purpose, according to one aspect of the present application, an abnormality detection method for in-service OPGW optical cable based on multi-modal data is provided, comprising:
[0008] S1, collecting the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, and simultaneously collecting the state information of the corresponding OPGW optical cables, wherein the state information of the OPGW optical cables includes the wind speed at the time of collecting the Brillouin frequency shift parameters of the OPGW optical cables, and the tower distance and height difference of the corresponding OPGW optical cables;
[0009] S2, respectively, the collected OPGW optical cable Brillouin frequency shift parameters and its corresponding OPGW optical cable state information is extracted, and the Brillouin frequency shift parameter characteristics of the OPGW optical cable and the state information characteristics of the OPGW optical cable are obtained;
[0010] S3, the Brillouin frequency shift parameter characteristics of the OPGW optical cable and the state information characteristics of the OPGW optical cable extracted in step S2 are multi-modal fusion, and the OPGW optical cable fault is classified by combining the SVM classifier, so as to complete the fault recognition.
[0011] Further, the multi-modal fusion in step S3 includes feature layer fusion and decision layer fusion, wherein the feature layer fusion includes fusion analysis of the extracted Brillouin frequency shift parameter characteristics of the OPGW optical cable and the state information characteristics of the OPGW optical cable, and a new feature vector is obtained; the decision layer fusion includes pre-classification of the extracted Brillouin frequency shift parameter characteristics of the OPGW optical cable and the state information characteristics of the OPGW optical cable, and two binary SVM classifiers are established, two kinds of to-be-identified fault states are combined to construct an identification framework Θ={CL1, CL2}, the two kinds of to-be-identified fault states include fault state and normal state, and a D-S reasoning distribution function is established through the output result of the SVM classifier, and the probability value (BPA) of the two kinds of fault states is calculated, and the probability value is fused by orthogonal rule to complete the decision layer fusion:
[0012] In the formula, the feature vector after multi-modal fusion is marked as m(CL1) and m(CL2), wherein m(CL1) represents the trust degree of considering the fault state, and m(CL2) represents the trust degree of considering the normal state; the trust function of each kind of fault state is marked as Bel(CL), the identification framework is marked as Θ, and the measurement of all B subsets in set A is marked as Bel(A).
[0013] Further, in step S3, the OPGW optical cable fault is classified by combining the SVM classifier, so as to complete the fault recognition, which specifically includes: setting a fixed threshold a, if Bel(A) is greater than the set fixed threshold a, it is considered that the OPGW optical cable has a fault, that is, the fault diagnosis is completed.
[0014] Further, an OPGW optical cable abnormality detection system based on multi-modal data, characterized in that it comprises:
[0015] The data acquisition module is used to collect the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, and at the same time collect the status information of the corresponding OPGW optical cables. The status information of the OPGW optical cables includes: the wind speed when the Brillouin frequency shift parameters of the OPGW optical cables are collected, and the tower distance and height difference of the corresponding OPGW optical cables.
[0016] The feature extraction module is used to extract features from the Brillouin frequency shift parameters of the collected OPGW optical cable and the corresponding state information of the OPGW optical cable, respectively, to obtain the Brillouin frequency shift parameter features and the state information features of the OPGW optical cable.
[0017] The multimodal fusion module is used to perform multimodal fusion of the extracted Brillouin frequency shift parameter features and the state information features of the OPGW optical cable, and combine them with an SVM classifier to classify OPGW optical cable faults, thereby completing fault identification.
[0018] Furthermore, the multimodal fusion module performs multimodal fusion on the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable. Specifically, this includes feature-level fusion and decision-level fusion. Feature-level fusion involves fusing and analyzing the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable to obtain new feature vectors. Decision-level fusion involves pre-classifying the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable, establishing two binary SVM classifiers, and combining two types of fault states to be identified to construct a recognition framework Θ = {CL1, CL2}. These two fault states include fault states and normal states. A DS inference assignment function is established using the output of the SVM classifiers to calculate the probability values (BPA) of the two types of fault states. And feature fusion of probability values is performed through orthogonal rules to complete the decision-level fusion:
[0019] In the formula, the feature vectors after multimodal fusion are labeled in the form of m(CL1) and m(CL2), where m(CL1) represents the confidence level of the fault state and m(CL2) represents the confidence level of the normal state; the confidence function of each type of fault state is labeled in the form of Bel(CL), the identification frame is labeled Θ, and the metric and label of all subsets B in set A is Bel(A).
[0020] Furthermore, the multimodal fusion module combines an SVM classifier to classify OPGW optical cable faults, thereby completing fault identification. Specifically, it includes setting a fixed threshold a. If Bel(A) is greater than the set fixed threshold a, then the OPGW optical cable is considered to have a fault, thus completing fault diagnosis.
[0021] This invention enables real-time monitoring and fault diagnosis by utilizing the Brillouin frequency shift parameters and status information parameters of OPGW optical cables. The innovation of this invention lies in the selected parameter combination. Compared with other existing multimode data fusion schemes, it adds status information parameters of the OPGW optical cable, specifically the wind speed and the corresponding tower distance and height difference when collecting the Brillouin frequency shift parameters. This real-time performance and flexibility allow the system to respond quickly to faults and take appropriate measures to ensure the reliability and stability of the OPGW optical cable. Simultaneously, it enables predictive maintenance for potential faults. By identifying problems in advance and taking measures, the system can reduce maintenance time and costs, and extend the life of the OPGW optical cable. By comprehensively utilizing this data, faults in the OPGW optical cable can be diagnosed more accurately. Attached Figure Description
[0022] Figure 1 is a flowchart of an anomaly detection method for in-operation OPGW optical cables based on multimodal data provided in an embodiment of the present invention;
[0023] Figure 2 is a schematic diagram of an in-operation OPGW optical cable anomaly detection system based on multimodal data provided in an embodiment of the present invention. Detailed Implementation
[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0025] As shown in Figure 1, this embodiment of the invention provides a method for detecting anomalies in operational OPGW optical cables based on multimodal data, comprising the following steps:
[0026] S1. Collect the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, as well as the corresponding status information of OPGW optical cables. The status information of OPGW optical cables includes: the wind speed when collecting the Brillouin frequency shift parameters of OPGW optical cables and the tower distance and height difference of the corresponding OPGW optical cables.
[0027] S2. Perform feature extraction on the two types of data collected above, that is, extract features from the Brillouin frequency shift parameters of the collected OPGW optical cable and the state information of the corresponding OPGW optical cable to obtain the features of the Brillouin frequency shift parameters of the OPGW optical cable and the features of the state information of the OPGW optical cable.
[0028] Step S2 includes the following sub-steps:
[0029] S21. After denoising and normalizing the Brillouin frequency shift parameters of the collected OPGW optical cable, its characteristic parameter x1 is obtained.
[0030] S22. After denoising and normalizing the collected OPGW optical cable status information, its characteristic parameter x2 is obtained.
[0031] S3. A multimodal fusion method combined with an SVM classifier is used to classify OPGW optical cable faults, thereby completing fault identification. First, a multi-layer classification model is established based on the SVM classifier. Then, the acquired features are fused using a multimodal fusion method. Finally, the established classification model is used to diagnose OPGW optical cable faults.
[0032] Specifically, S3 can be implemented in the following ways:
[0033] S31. Feature layer fusion is shown below:
[0034] The Brillouin frequency shift parameters of the OPGW optical cable obtained above are fused and analyzed with the state information of the corresponding OPGW optical cable to obtain a new feature vector, as shown in equation (1); y=x1⊙x2, (1)
[0035] In the formula, the new feature vector is labeled y, and ⊙ is the Hadamard product.
[0036] S32. The decision-making level integration is shown below:
[0037] First, the Brillouin frequency shift parameters of the acquired OPGW optical cable and the corresponding state information features of the OPGW optical cable are pre-classified. Two binary SVM classifiers are established, each classifying the input feature y. The output results are used for fusion in the decision layer. Two types of fault states to be identified (fault state and normal state) are combined to construct an identification framework Θ = {CL1, CL2} (fault state and normal state). The DS inference assignment function is established using the output results of the SVM classifiers to calculate the probability values (BPA) of the two types of fault states. Let S1, S2, and S3 represent the probability values that classifier S1 considers the state to be faulty, normal, and normal, respectively. The probability values are then fused using orthogonal rules, and the result is shown in equation (2).
[0038] In the formula, the feature vectors after multimodal fusion are labeled in the form of m(CL1) and m(CL2), where m(CL1) represents the confidence level of the fault state and m(CL2) represents the confidence level of the normal state; the confidence function of each type of fault state is labeled in the form of Bel(CL), the identification frame is labeled Θ, and the metric and label of all subsets B in set A is Bel(A). In this invention, Bel(A) represents the confidence function value of the fault state and represents the sum of the basic probability allocation functions of all subsets of the fault state. Finally, a fixed threshold a is set. If Bel(A) is greater than the set fixed threshold a, the OPGW optical cable is considered to have a fault, and the fault diagnosis is completed.
[0039] According to another aspect of the present invention, as shown in FIG2, an anomaly detection system for an operational OPGW optical cable based on multimodal data is provided, comprising:
[0040] The data acquisition module is used to collect the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, and at the same time collect the status information of the corresponding OPGW optical cables. The status information of the OPGW optical cables includes: the wind speed when the Brillouin frequency shift parameters of the OPGW optical cables are collected, and the distance and height difference between the corresponding OPGW optical cables and the towers.
[0041] The feature extraction module is used to extract features from the Brillouin frequency shift parameters of the collected OPGW optical cable and the corresponding state information of the OPGW optical cable, respectively, to obtain the Brillouin frequency shift parameter features and the state information features of the OPGW optical cable.
[0042] The multimodal fusion module is used to perform multimodal fusion of the extracted Brillouin frequency shift parameter features and the state information features of the OPGW optical cable, and combine them with an SVM classifier to classify OPGW optical cable faults, thereby completing fault identification.
[0043] This invention's method for detecting anomalies in operational OPGW optical cables based on multimodal data combines multiple data sources, including the Brillouin frequency shift parameters and state information parameters of the OPGW optical cable. While wind speed affects the wind pressure ratio of the optical cable, more significantly, it easily causes the cable to gallop and vibrate at high altitudes. This vibration and galloping not only affects the operating status of the optical cable but also seriously threatens the normal operation of the transmission line. When consecutive spans of OPGW optical cables are subjected to wind force, the stress components in the vertical and horizontal directions of the cable differ due to variations in the span spacing and elevation difference angle between spans. Consequently, the displacements generated at each suspension point in different line directions will also vary. Therefore, compared with other existing multimodal data fusion schemes, this method adds state information parameters of the OPGW optical cable, namely, the wind speed at which the Brillouin frequency shift parameters of the OPGW optical cable are collected, and the corresponding tower distance and height difference of the OPGW optical cable. By incorporating environmental factors such as wind speed into the analysis, the performance of the optical cable under actual operating conditions can be more accurately evaluated. This multi-dimensional data fusion helps identify potential anomalies, thereby reducing false alarms and missed alarms. Wind speed and tower information reflect the dynamic response of optical cables under different environmental conditions, enabling the system to monitor changes in the cable's condition in a timely manner and issue early warnings to prevent damage or accidents caused by wind and other factors. Furthermore, different wind speeds, tower distances, and height differences cause variations in the stress components of the optical cable in different directions. By comprehensively considering these parameters, a more complete analysis of the cable's stress state can be performed, thereby assessing its safety and reliability.
[0044] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for detecting anomalies in operational OPGW optical cables based on multimodal data, characterized in that, include: S1. Collect the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, and at the same time collect the status information of the corresponding OPGW optical cables. The status information of OPGW optical cables includes: the wind speed when collecting the Brillouin frequency shift parameters of OPGW optical cables and the tower distance and height difference of the corresponding OPGW optical cables. S2. Extract features from the Brillouin frequency shift parameters of the collected OPGW optical cable and the corresponding state information of the OPGW optical cable to obtain the features of the Brillouin frequency shift parameters and the state information features of the OPGW optical cable. S3. The Brillouin frequency shift parameter features and state information features of the OPGW optical cable extracted in step S2 are fused in a multimodal manner, and the OPGW optical cable faults are classified by combining the SVM classifier, thereby completing the fault identification.
2. The method for detecting anomalies in operational OPGW optical cables based on multimodal data as described in claim 1, characterized in that: The multimodal fusion described in step S3 includes feature-level fusion and decision-level fusion. Feature-level fusion involves fusing and analyzing the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable to obtain a new feature vector. Decision-level fusion involves pre-classifying the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable, establishing two binary SVM classifiers, and combining two types of fault states to be identified to construct a recognition framework Θ = {CL1, CL2}. These two fault states include fault states and normal states. A DS inference assignment function is established using the output of the SVM classifiers to calculate the probability values (BPA) of the two types of fault states. And feature fusion of probability values is performed through orthogonal rules to complete the decision-level fusion: In the formula, the feature vectors after multimodal fusion are labeled in the form of m(CL1) and m(CL2), where m(CL1) represents the confidence level of the fault state and m(CL2) represents the confidence level of the normal state; the confidence function of each type of fault state is labeled in the form of Bel(CL), the identification frame is labeled Θ, and the metric and label of all subsets B in set A is Bel(A).
3. The method for detecting anomalies in operational OPGW optical cables based on multimodal data as described in claim 2, characterized in that: In step S3, the OPGW optical cable faults are classified using an SVM classifier to complete fault identification. Specifically, a fixed threshold a is set. If Bel(A) is greater than the set fixed threshold a, the OPGW optical cable is considered to have a fault, thus completing the fault diagnosis.
4. An anomaly detection system for in-service OPGW optical cables based on multimodal data, characterized in that, include: The data acquisition module is used to collect the Brillouin frequency shift parameters of OPGW optical cables under different operating conditions in actual production, and at the same time collect the status information of the corresponding OPGW optical cables. The status information of the OPGW optical cables includes: the wind speed when the Brillouin frequency shift parameters of the OPGW optical cables are collected, and the tower distance and height difference of the corresponding OPGW optical cables. The feature extraction module is used to extract features from the Brillouin frequency shift parameters of the collected OPGW optical cable and the corresponding state information of the OPGW optical cable, respectively, to obtain the Brillouin frequency shift parameter features and the state information features of the OPGW optical cable. The multimodal fusion module is used to perform multimodal fusion of the extracted Brillouin frequency shift parameter features and the state information features of the OPGW optical cable, and combine them with an SVM classifier to classify OPGW optical cable faults, thereby completing fault identification.
5. The in-service OPGW optical cable anomaly detection system based on multimodal data as described in claim 4, characterized in that: The multimodal fusion module performs multimodal fusion on the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable. Specifically, this includes feature-level fusion and decision-level fusion. Feature-level fusion involves fusing and analyzing the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable to obtain new feature vectors. Decision-level fusion involves pre-classifying the extracted Brillouin frequency shift parameter features and state information features of the OPGW optical cable, establishing two binary SVM classifiers, and combining two types of fault states to be identified to construct a recognition framework Θ = {CL1, CL2}. These two fault states include fault states and normal states. A DS inference assignment function is established based on the output of the SVM classifiers to calculate the probability values (BPA) of the two types of fault states. And feature fusion of probability values is performed through orthogonal rules to complete the decision-level fusion: In the formula, the feature vectors after multimodal fusion are labeled in the form of m(CL1) and m(CL2), where m(CL1) represents the confidence level of the fault state and m(CL2) represents the confidence level of the normal state; the confidence function of each type of fault state is labeled in the form of Bel(CL), the identification frame is labeled Θ, and the metric and label of all subsets B in set A is Bel(A).
6. The in-service OPGW optical cable anomaly detection system based on multimodal data as described in claim 5, characterized in that: The multimodal fusion module combines an SVM classifier to classify OPGW optical cable faults, thereby completing fault identification. Specifically, it includes setting a fixed threshold a. If Bel(A) is greater than the set fixed threshold a, then the OPGW optical cable is considered to have a fault, thus completing fault diagnosis.