CN116614177B - Optical fiber state multidimensional parameter monitoring system - Google Patents
Optical fiber state multidimensional parameter monitoring system Download PDFInfo
- Publication number
- CN116614177B CN116614177B CN202310509077.0A CN202310509077A CN116614177B CN 116614177 B CN116614177 B CN 116614177B CN 202310509077 A CN202310509077 A CN 202310509077A CN 116614177 B CN116614177 B CN 116614177B
- Authority
- CN
- China
- Prior art keywords
- data
- model
- optical fiber
- time
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000013307 optical fiber Substances 0.000 title claims abstract description 106
- 238000012544 monitoring process Methods 0.000 title claims abstract description 31
- 238000004458 analytical method Methods 0.000 claims abstract description 25
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 20
- 230000003287 optical effect Effects 0.000 claims abstract description 11
- 238000000034 method Methods 0.000 claims description 46
- 230000015654 memory Effects 0.000 claims description 33
- 238000012549 training Methods 0.000 claims description 26
- 238000004891 communication Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 18
- 230000006399 behavior Effects 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 16
- 230000003993 interaction Effects 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 15
- 230000005540 biological transmission Effects 0.000 claims description 14
- 239000000835 fiber Substances 0.000 claims description 12
- 238000007621 cluster analysis Methods 0.000 claims description 10
- 238000013145 classification model Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000000007 visual effect Effects 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 230000007246 mechanism Effects 0.000 claims description 5
- 238000013139 quantization Methods 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 4
- 238000009826 distribution Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 claims description 4
- 238000007667 floating Methods 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 4
- 238000003064 k means clustering Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000002787 reinforcement Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 230000008859 change Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 125000004122 cyclic group Chemical group 0.000 claims 1
- 238000012423 maintenance Methods 0.000 description 4
- 238000000513 principal component analysis Methods 0.000 description 4
- 238000005452 bending Methods 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 241001391944 Commicarpus scandens Species 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008033 biological extinction Effects 0.000 description 1
- 239000011248 coating agent Substances 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012482 interaction analysis Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/07—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems
- H04B10/075—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal
- H04B10/079—Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems using an in-service signal using measurements of the data signal
- H04B10/0795—Performance monitoring; Measurement of transmission parameters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/34—Browsing; Visualisation therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/092—Reinforcement learning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1023—Server selection for load balancing based on a hash applied to IP addresses or costs
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Electromagnetism (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Animal Behavior & Ethology (AREA)
- Telephonic Communication Services (AREA)
Abstract
The application relates to an optical fiber state multidimensional parameter monitoring system, which relates to the technical field of optical fibers and comprises a data acquisition device; the data acquisition device is used for acquiring multi-source data information; the multi-source data information includes: electrical, optical, and thermal signals; and, an edge device; the edge equipment is used for receiving the multi-source data information, performing operation processing and information analysis through the convolutional neural network model, and sending the results of the operation processing and the information analysis to an external cloud server.
Description
Technical Field
The application relates to the technical field of optical fibers, in particular to an optical fiber state multidimensional parameter monitoring system.
Background
The electric energy becomes one of indispensable elements for people in the society today to produce life, and the stable operation of the electric power system becomes an important guarantee for social stability and national economy development. Compared with microwave communication and cable communication, the power optical fiber communication has the advantages of large information carrying capacity, rapid information transmission, low loss, high confidentiality, suitability for long-distance stable transmission, strong corrosion and interference resistance, convenient installation and the like, is widely applied to a power communication system, and carries more than 95% of power communication service. Therefore, the reliability and stability of the operation of the power optical fiber transmission network have become important posts for the safe operation of the power system, and once the optical fiber circuit fails, the communication interruption caused by the failure brings great economic loss to the national network company and great social influence.
At present, a part of power optical fibers are buried underground, and the power optical fibers are laid between transmission towers along with power cables, are greatly influenced by sudden natural disasters such as rain and snow, and are easy to break, so that normal optical communication is influenced. Meanwhile, some of the power optical fibers are erected in the air along with the power transmission line, so that the power optical fibers are more susceptible to natural factors (ice coating, high wind and the like) than other optical cables. The power optical fiber communication network is increasingly complex, and the maintenance difficulty and maintenance cost of the power optical fiber are gradually increased. Besides the interruption of power optical fiber communication caused by sudden rain and snow weather, the power optical fiber with long service time is easily subjected to the problems of slight cracks, bending and the like due to the influence of factors such as wind power, sunlight exposure, high temperature, humidity and the like, the cracks and bending can refract optical signals to form optical loss, the optical fiber circuit state aging is caused in the long time to influence the optical communication accuracy, and furthermore, the problem of the power optical cable is directly caused by the direct bending, cracking and breaking of the optical fiber.
In summary, the scale of the power optical fiber communication network is larger and larger, and the problems that the exposed optical fiber fault is not monitored timely, the maintenance difficulty of the optical fiber circuit is increased and the like are more and more serious. As a key part of power transmission scheduling communication of a power system, once a power optical fiber network fails, the communication interruption possibly causes great economic loss to the power system, the power optical fiber failure is an unavoidable practical problem, and the problems of low timeliness and high economic cost exist only by means of manual repair and manual maintenance, so that the on-line monitoring of the power optical fiber failure and the early warning of the power optical fiber line state are technical problems to be overcome in the field.
Disclosure of Invention
In order to at least partially solve the technical problems, the application provides an optical fiber state multidimensional parameter monitoring system.
The application provides an optical fiber state multidimensional parameter monitoring system which adopts the following technical scheme.
A fiber optic condition multi-dimensional parametric monitoring system, comprising:
a data acquisition device; the data acquisition device is used for acquiring multi-source data information; the multi-source data information includes: electrical, optical, and thermal signals; the method comprises the steps of,
an edge device; the edge equipment is used for receiving the multi-source data information, performing operation processing and information analysis through a convolutional neural network model, and sending the results of the operation processing and the information analysis to an external cloud server;
the front end of the edge equipment comprises an algorithm deployment part and an edge area total node;
the algorithm deployment part is used for deploying and optimizing the convolutional neural network model;
the edge region summary points are for:
data are collected in real time, and data information transmitted from each node is collected in real time;
transmitting data in real time, and transmitting data information transmitted from each node to a cloud server in real time;
and detecting data abnormality, and monitoring data quality in real time by using a convolutional neural network model.
Optionally, the deployment and optimization of the convolutional neural network model includes:
selecting a CRSnet model;
replacing the CSRnet network model from the vgnet model to a mobilent model;
deleting unimportant parameters and redundant channels in the model training process; the method comprises the steps of,
the original 32-bit floating point number weight and activation value are converted into an 8-bit fixed point value through parameter quantization pairs.
Optionally, the performing the operation processing through the convolutional neural network model includes:
step 301, constructing a fusion high-dimensional data matrix by using fiber real-time data and historical data;
step 302, PCA dimension reduction is carried out on the high-dimension data matrix;
step 303, performing domain-adaptive rough set cluster analysis on the data in step 302 to obtain a cluster center of an upper approximate set and a cluster center of a lower approximate set, and obtaining a cluster result after cluster analysis.
Optionally, the edge device is further configured with an optical fiber state classification model; the method for generating the optical fiber state classification model comprises the following steps:
a feature clustering step;
a DCNN training step;
the feature clustering step and the DCNN training step are circularly executed, so that the depth feature discriminant and the model recognition performance of all categories are enhanced;
wherein, the characteristic clustering step includes: extracting features to obtain depth features; clustering the depth features according to the categories by using a k-means clustering method to obtain a distribution structure of the data in a DCNN feature space;
the DCNN training step comprises the following steps: calculating cross entropy loss and clustering metric loss according to the DCNN output, the sample label and the clustering result; the model output and the feature representation of the DCNN are jointly optimized using cross entropy loss and cluster metric loss.
Optionally, the edge device is further configured with an optical fiber state prediction model; the method for generating the optical fiber state prediction model comprises the following steps:
step 501, sampling historical data, real-time data and accumulated fault data to establish a time sequence fault mode model based on a long-time and short-time memory network; the long-short-term memory network comprises: cell status and hidden status; the long-short-time memory network comprises the following three stages: forgetting the stage; the forgetting stage is used for selectively forgetting the input transmitted by the last node, controlling the part needing to be included in the information of the last state and the part needing to be forgotten, and finally obtaining an intermediate vector; a selection memory stage; the selection memory stage is used for: selectively memorizing the input; important part is recorded more, unimportant few records; an output stage; the output stage is used for deciding which will be used as the output of the current state through the output control function and performing scaling processing on the signal of the previous stage selection memory stage;
step 502, fusing optical fiber related priori mechanism models in each long-short-term memory network;
step 503, obtaining a plurality of time sequence fault mode models of the optical fiber under different service conditions after training;
step 504, obtaining fault mode curves of a plurality of optical fiber service stages through a plurality of time sequence fault mode models; each curve is fitted by a different time-series fiber performance decay or failure mode model;
step 505, performing approximate integration on each curve to obtain a damage degree prediction result of a specific optical fiber stage; the different prediction results form a result interval;
step 506, calculating the variance between different prediction curves to obtain the confidence coefficient of the prediction result at different time nodes;
step 507, feeding back new data to the model trained by the historical data, and detecting the prediction accuracy and precision of the model, and iterating continuously.
Optionally, the information analysis by the convolutional neural network model includes:
symbolizing knowledge in a plurality of fields and constructing a unified knowledge graph;
mapping the domain knowledge graph and massive optical fiber data to the same vector space by using a knowledge representation learning method and a multi-mode feature extraction method respectively, and maintaining the complex structure of knowledge and the inherent semantics of the data;
the analysis process and the analysis result are presented in the form of images, texts and maps for visual presentation for specific data analysis tasks and specific user groups.
Optionally, the visual presentation further includes:
the behavior sequence data of the user and the corresponding system internal data are obtained through the past interaction records,
performing down-line training on the designed convolutional neural network model;
dynamically predicting interaction behavior of a user at the next moment on line, and possible influence on a system, communication among nodes and load conditions;
and allocating resources and coordinating user interaction behaviors through a deep reinforcement learning algorithm based on the predicted user behaviors and system states.
Optionally, the edge area total node is further configured to: and (5) local data backup, namely storing the original data and the processed data information in a local storage.
Optionally, the edge area total node is further configured to: uploading the data log to a cloud function, and periodically sending the log to a cloud server through a network.
Drawings
FIG. 1 is a system block diagram of a fiber state multidimensional parameter monitoring system in accordance with an embodiment of the present application;
FIG. 2 is a schematic diagram of a system for monitoring a multi-dimensional parameter of an optical fiber state according to an embodiment of the present application.
Detailed Description
The application is further illustrated by the following description of specific embodiments in conjunction with the accompanying drawings 1-2:
first, what needs to be described here is: in the description of the present application, terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," "outer," and the like are used for convenience of description only as regards orientation or positional relationship as shown in the accompanying drawings, and do not denote or imply that the apparatus or element in question must have a particular orientation, be constructed and operated in a particular orientation, and thus should not be construed as limiting the present application; moreover, the numerical terms such as the terms "first," "second," "third," etc. are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In addition, unless explicitly stated or limited otherwise, the terms "mounted," "connected," and "connected" should be construed broadly, and may be, for example, a fixed connection, a releasable connection, an interference fit, a transition fit, or an integral connection; can be directly connected or indirectly connected through an intermediate medium; the specific meaning of the above terms in the present application will be understood by those skilled in the art according to the specific circumstances.
The embodiment of the application discloses an optical fiber state multidimensional parameter monitoring system. As an embodiment of a fiber optic condition multi-dimensional parameter monitoring system, the system comprises:
a data acquisition device; the data acquisition device is used for acquiring multi-source data information; the multi-source data information includes: electrical, optical, and thermal signals; the method comprises the steps of,
an edge device; the edge equipment is used for receiving the multi-source data information, performing operation processing and information analysis through the convolutional neural network model, and sending the results of the operation processing and the information analysis to an external cloud server;
the front end of the edge equipment comprises an algorithm deployment part and an edge area total node;
the algorithm deployment part is used for deploying and optimizing the convolutional neural network model;
the edge region summary points are used to:
data are collected in real time, and data information transmitted from each node is collected in real time;
transmitting data in real time, and transmitting data information transmitted from each node to a cloud server in real time;
data anomaly detection, and real-time monitoring of data quality by using convolutional neural network model
Specifically, edge computing is a technical scheme for providing end services nearby at a data source side or a device physical environment layer by adopting an open platform integrating storage, computing, network and application core capabilities. The edge computing forms an edge computing node (edge computing node) by issuing computing tasks and functions of the core node to edge side equipment with computing capability, fully utilizes the processing computing capability of the edge side to perform preliminary processing on information, and even completely provides service which is originally operated in a cloud computing server for users. The state monitoring of the power optical fiber transmission system needs to collect various parameters, and mainly comprises three parameters, namely an optical fiber body state, an ambient environment state and a data transmission state. These parameters relate to different types of signals including: electrical signals, optical signals, thermal signals, etc. need to be collected by the data collection device. In order to ensure that the network is not broken in the optical fiber state acquisition process, the data acquisition device also needs to integrate an optical fiber device.
The application, which is used for carrying out the research and application of the optical fiber state multidimensional parameter monitoring, diagnosis and prediction system by combining with the advanced information technologies such as artificial intelligence and big data, and the like, is used for forming the real-time monitoring capability of the optical fiber multidimensional parameter, the classifying, diagnosing and predicting capability of the optical fiber performance in different stages and the multimode faults of the optical fiber performance evolution by carrying out the optical fiber electric-optical-thermal multiparameter real-time data acquisition technology and device, the optical fiber performance evolution and fault evaluation index system, the optical fiber state classifying model and the optical fiber state prediction method.
Referring to fig. 1, the system of the present application mainly comprises an "electro-optical-thermal" multi-type data acquisition device and an edge device deployed with a neural network algorithm. In the system, the neural network algorithm is deployed into the edge equipment, so that real-time data processing and analysis can be performed on the edge equipment, and the problems of data transmission delay and bandwidth bottleneck are avoided. Meanwhile, the edge equipment collects multi-source data information in real time through the data collection device. After the acquired data is subjected to operation processing and information analysis of a neural network algorithm, the result is sent to a cloud server so as to further process and store the data.
Referring to fig. 2, in the whole edge computing front end, mainly focusing on an edge system implementation part, including deployment of a neural network algorithm, acquisition and processing of edge end multi-source multi-parameter data, simple analysis and prediction of edge nodes, and communication with a cloud server. In addition, in order to cope with the challenges of real-time mass data acquisition, abnormal points of the data are judged in real time by using a deep learning model, and real-time preprocessing of the data is performed, so that the efficiency and accuracy of data processing are improved.
The data acquisition device is developed corresponding to three types of signals of 'electricity-light-heat'. The data acquisition devices are transmitted to the regional total node through a bus protocol, and an ARCioddk development board can be adopted as the edge terminal total node.
As a specific implementation mode of the optical fiber state multidimensional parameter monitoring system, the deployment and optimization of the convolutional neural network model comprises the following steps:
selecting a CRSnet model;
replacing the CSRnet network model from the vgnet model to a mobilent model;
deleting unimportant parameters and redundant channels in the model training process; the method comprises the steps of,
the original 32-bit floating point number weight and activation value are converted into an 8-bit fixed point value through parameter quantization pairs.
Specifically, the application refers to the structural design of the CRSnet model and carries out targeted improvement according to the characteristic of poor computing capability of an edge processor. The main optimization comprises the following steps:
the CSRnet network model is replaced by a lightweight mobilet model from a vgnet model, and the mobilet is a lightweight convolutional neural network and is suitable for being deployed at the edge end with relatively low performance, and the calculated amount is greatly reduced through model replacement.
Parameter pruning is performed on the model. By means of the evaluation algorithm, unimportant parameters and redundant channels are deleted in the training process, and the operation amount of the network is reduced under the condition that accuracy is guaranteed.
The parameters are quantized. The original 32-bit floating point number weight and activation value are converted into an 8-bit fixed point value through parameter quantization pairs. The quantized 8-bit fixed-point number parameter not only can reduce the storage space of the parameter, but also requires fewer resources in hardware for fixed-point number operation and has higher running speed compared with floating-point number operation.
The model compression is carried out in a mode of model channel cutting and parameter quantization, and the parameter quantity is reduced to 6.67% of the original model. The compressed model is output in the tensorf lowlite format.
As a specific implementation mode of the optical fiber state multidimensional parameter monitoring system, the operation processing is carried out through a convolutional neural network model, and the method comprises the following steps:
step 301, constructing a fusion high-dimensional data matrix by using fiber real-time data and historical data;
step 302, performing PCA dimension reduction on the high-dimensional data matrix;
step 303, performing domain-adaptive rough set cluster analysis on the data in step 302 to obtain a cluster center of an upper approximate set and a cluster center of a lower approximate set, and obtaining a cluster result after cluster analysis.
Specifically, the application provides a clustering method for reducing dimension by using the principle of principal component analysis and calculating an upper approximation set and a lower approximation set by using the principle of a field self-adaptive rough set, thereby solving the problem of unclear boundaries in the traditional clustering method. The model uses the ideas of principal component analysis and rough set for cluster analysis, namely, three methods of principal component analysis, rough set and cluster analysis are combined together, real-time data and historical data of optical fiber states are researched, and clustering is carried out. The specific method comprises the following steps: the data is subjected to principal component analysis, and then clustering analysis is performed by using a clustering algorithm based on a rough set.
The method comprises the following steps:
constructing a fusion high-dimensional data matrix by using the real-time data and the historical data of the optical fibers;
PCA dimension reduction is carried out on the high-dimension data matrix;
and performing cluster analysis based on the field self-adaptive rough set on the previous data to obtain a cluster center of an upper approximate set and a cluster center of a lower approximate set, and obtaining a cluster result after the cluster analysis, wherein the cluster result comprises the upper approximate set and the lower approximate set.
As a specific implementation mode of the optical fiber state multidimensional parameter monitoring system, the edge equipment is further provided with an optical fiber state classification model; the method for generating the optical fiber state classification model comprises the following steps:
a feature clustering step;
a DCNN training step;
the feature clustering step and the DCNN training step are circularly executed, so that the depth feature discriminant and the model recognition performance of all categories are enhanced;
wherein, the step of feature clustering comprises: extracting features to obtain depth features; clustering the depth features according to the categories by using a k-means clustering method to obtain a distribution structure of the data in a DCNN feature space;
the DCNN training step comprises the following steps: calculating cross entropy loss and clustering metric loss according to the DCNN output, the sample label and the clustering result; the model output and the feature representation of the DCNN are jointly optimized using cross entropy loss and cluster metric loss.
Specifically, in the actual operation process of the power optical fiber transmission network system, in the process of optical fiber performance evolution and different faults, the data volume is huge, but obvious unbalance of data types can occur, so that the accuracy and the reliability are required to be improved when the optical fiber state classification is carried out. Meanwhile, repeated utilization and modeling of mass data bring about overwhelming of the machine, so that real-time performance is greatly reduced, and prediction accuracy is further reduced. Aiming at the performance reduction of few classes caused by data unbalance and complexity, the application designs an optical fiber state classification model. Specifically, the designed method utilizes a deep convolutional neural network to map original data to a feature space, and adopts a clustering metric loss function to correct the depth feature so as to obtain the discriminative feature representation. The loss function is used for restraining and optimizing the relation among the clustering distances of all the categories, the intra-category distances and the inter-category distances, so that the clusters of the same category become compact, the clusters of different categories become sparse, and further a boundary with obvious margin is formed, and the distinguishing performance and the generalization performance of the depth features are effectively improved.
The method for generating the optical fiber state classification model comprises the following steps: feature clustering step and DCNN training step. In the clustering process, firstly, feature extraction is carried out to obtain depth features, and then the k-means clustering method is utilized to cluster the depth features according to categories to obtain a distribution structure of data in a DCNN feature space. In the training process, cross entropy loss and clustering metric loss are calculated according to DCNN output, sample labels and clustering results, and the model output and feature representation of the DCNN are subjected to joint optimization by using the supervision signals of the two different layers. And the feature clustering step and the DCNN training step are circularly carried out, so that the depth feature discriminant and the model recognition performance of all categories are enhanced. The key to depth metric learning is the correction of the features. The application realizes the feature correction of all categories through the clustering measurement loss function, so that the features of different categories can have obvious interfaces, and the influence of data unbalance and complexity on feature learning is effectively reduced.
The application also provides a system for monitoring the multidimensional parameter of the optical fiber state, which comprises: the edge equipment is also provided with an optical fiber state prediction model; the method for generating the optical fiber state prediction model comprises the following steps:
step 501, sampling historical data, real-time data and accumulated fault data to establish a time sequence fault mode model based on a long-time and short-time memory network; the long-short-term memory network comprises: cell status and hidden status; the long-short-time memory network comprises the following three stages: forgetting the stage; the forgetting stage is used for selectively forgetting the input transmitted by the last node, controlling the part needing to be included in the information of the last state and the part needing to be forgotten, and finally obtaining an intermediate vector; a selection memory stage; the selection memory stage is used for: selectively memorizing the input; important part is recorded more, unimportant few records; an output stage; the output stage is used for deciding which will be used as the output of the current state through the output control function and performing scaling processing on the signal of the previous stage selection memory stage;
step 502, fusing optical fiber related priori mechanism models in each long-short-term memory network;
step 503, obtaining a plurality of time sequence fault mode models of the optical fiber under different service conditions after training;
step 504, obtaining fault mode curves of a plurality of optical fiber service stages through a plurality of time sequence fault mode models; each curve is fitted by a different time-series fiber performance decay or failure mode model;
step 505, performing approximate integration on each curve to obtain a damage degree prediction result of a specific optical fiber stage; the different prediction results form a result interval;
step 506, calculating the variance between different prediction curves to obtain the confidence coefficient of the prediction result at different time nodes;
step 507, feeding back new data to the model trained by the historical data, and detecting the prediction accuracy and precision of the model, and iterating continuously.
Specifically, historical data, real-time data and accumulated fault data with time sequence information of a power optical fiber transmission network are mined, hidden information existing in the data is mined through a deep learning technology, a long and short time memory model structure is adopted, a mechanism model is fused, a plurality of time sequence fault mode models are built, prediction functions of a plurality of time sequence fault mode prediction models can be obtained, dynamic fault rate and damage degree prediction intervals of the optical fiber are obtained through approximate integral transformation, confidence intervals of fault mode and damage degree prediction of a specific time node are given through calculating variance of prediction results, and a plurality of models are integrated to obtain a final result. Meanwhile, considering that the data is updated and iterated continuously, the accuracy of the model is reduced along with the extension of the prediction time, the model is iterated and optimized in the full life cycle of the evolution of the optical fiber performance by adopting a data iteration verification mode and using new data to verify the accuracy and the generalization performance of the existing model.
Firstly, historical data, real-time data and accumulated fault data are sampled, and then a time sequence fault mode model based on a long-short-time memory network is established. Long and short term memory networks (LSTM) are a deep learning architecture based on Recurrent Neural Networks (RNNs), which have proven to have powerful modeling capabilities on time series data in many applications, mainly to solve the problems of gradient extinction and gradient explosion during long sequence training, and to perform better in longer sequences than common RNNs. Compared with the single transfer state of the RNN, the long-short-term memory network has two transmission states, namely a cell state and a hidden state. Inside the LSTM there are mainly three phases: the first phase is called forget phase, forget gate, which is mainly to selectively forget the input coming in from the last node. Simply forget that it is not important, remembers that it is important. Specifically, the gating function obtained through calculation is used as forgetting gating, which information of the last state needs to be left and which needs to be forgotten is controlled, and finally an intermediate vector is obtained; the second phase is a selection memory phase, input gates, which selectively memorize the inputs of this phase. Mainly, the input state is selected and memorized. The important part is recorded more, and the unimportant part is recorded less. The current input is represented by the previously calculated intermediate vector. The selected gating signal is controlled by a memory gating function, and the vector value transmitted to the next state can be obtained by adding the results obtained in the two steps; the third phase is called output phase, output gate, this phase will decide which will be the output of the current state, mainly by the output control function, and also the memory gate signal of the previous phase is scaled.
After training, a plurality of time sequence fault mode models of the optical fiber under different service conditions can be obtained, the models can predict the fault mode of the optical fiber in a certain time period, the fault mode changes along with the change of time, the final model can obtain fault mode curves of a plurality of optical fiber service stages, each curve is fitted by different time sequence optical fiber performance decay or fault mode models, an approximate integral is carried out on each curve, a damage degree prediction result of a specific optical fiber stage can be obtained, different prediction results form a result interval, and the confidence of the prediction result of nodes in different time can be obtained by calculating the variance among different prediction curves. And finally, considering that the data is continuously updated along with the progress of time, adopting an iterative verification method to continuously optimize the model, specifically feeding new data back to the model trained by the historical data, and continuously iterating to obtain the dynamic fault mode prediction and damage degree prediction model with more and more excellent generalization performance and prediction results, wherein the prediction accuracy and precision of the detection model are continuously improved.
As one embodiment of an optical fiber state multidimensional parameter monitoring system, the information analysis by a convolutional neural network model comprises:
symbolizing knowledge in a plurality of fields and constructing a unified knowledge graph;
mapping the domain knowledge graph and massive optical fiber data to the same vector space by using a knowledge representation learning method and a multi-mode feature extraction method respectively, and maintaining the complex structure of knowledge and the inherent semantics of the data;
the analysis process and the analysis result are presented in the form of images, texts and maps for visual presentation for specific data analysis tasks and specific user groups.
In particular, there is a great deal of field knowledge about the performance evolution and multimode faults of the optical fibers, and the combination of massive optical fiber data can be used for carrying out solvable deep reasoning and analysis. Firstly, symbolizing knowledge in a plurality of fields, constructing a unified knowledge graph, mapping the knowledge graph in the fields and massive optical fiber data into the same vector space by using a knowledge representation learning method and a multi-mode feature extraction method, and maintaining the complex structure of knowledge and the internal semantics of the data. Then, constructing a micro inference rule based on fuzzy logic in a vector space, fusing knowledge and data to perform interpretable deduction inference, induction inference and the like, and discovering potential association between the data. Finally, the analysis process and analysis results are presented in the form of images, texts, patterns and the like for specific data analysis tasks and specific user groups.
As one embodiment of the optical fiber state multidimensional parameter monitoring system, the system performs visual presentation and further comprises:
the behavior sequence data of the user and the corresponding system internal data are obtained through the past interaction records,
performing down-line training on the designed convolutional neural network model;
dynamically predicting interaction behavior of a user at the next moment on line, and possible influence on a system, communication among nodes and load conditions;
and allocating resources and coordinating user interaction behaviors through a deep reinforcement learning algorithm based on the predicted user behaviors and system states.
In particular, the data scale of the optical fiber has mass property, and current researchers often observe and analyze the data in a real-time interaction mode, which brings great challenges to the visual analysis of the data, because the real-time property requirement of the interaction analysis has a necessary contradiction with the mass property of the data. Aiming at the contradiction between the data mass property and the interaction real-time property; firstly, obtaining behavior sequence data of a user and corresponding system internal data through previous interaction records, and performing downlink training on a designed machine learning model; then dynamically predicting the interaction behavior of the user at the next moment and the influence of the interaction behavior on the system, the communication and the load condition among nodes; finally, based on the predicted user behavior and system state, through a deep reinforcement learning algorithm, how to more reasonably allocate resource allocation and coordinate user interaction behavior is learned.
As one embodiment of the optical fiber state multidimensional parameter monitoring system, the edge area total node is further configured to: and (5) local data backup, namely storing the original data and the processed data information in a local storage.
As one embodiment of the optical fiber state multidimensional parameter monitoring system, the edge area total node is further configured to: uploading the data log to a cloud function, and periodically sending the log to a cloud server through a network.
It should be noted that: the above embodiments are only for illustrating the present application and not for limiting the technical solutions described in the present application, and although the present application has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that the present application may be modified or substituted by the same, and all the technical solutions and modifications thereof without departing from the spirit and scope of the present application are intended to be included in the scope of the claims of the present application.
Claims (5)
1. An optical fiber state multidimensional parameter monitoring system, comprising:
a data acquisition device; the data acquisition device is used for acquiring multi-source data information; the multi-source data information includes: electrical, optical, and thermal signals; the method comprises the steps of,
an edge device; the edge equipment is used for receiving the multi-source data information, performing operation processing and information analysis through a convolutional neural network model, and sending the results of the operation processing and the information analysis to an external cloud server;
the front end of the edge equipment comprises an algorithm deployment part and an edge area total node;
the algorithm deployment part is used for deploying and optimizing the convolutional neural network model;
the edge region summary points are for:
data are collected in real time, and data information transmitted from each node is collected in real time;
transmitting data in real time, and transmitting data information transmitted from each node to a cloud server in real time;
detecting abnormal data, and monitoring the quality of the data in real time by using a convolutional neural network model;
the deployment and optimization of the convolutional neural network model comprises the following steps:
selecting a CRSnet model;
replacing the CSRnet network model from the vgnet model to a mobilent model;
deleting unimportant parameters and redundant channels in the model training process; the method comprises the steps of,
converting the weight and the activation value of the original 32-bit floating point number into an 8-bit fixed point value through a parameter quantization pair;
the operation processing through the convolutional neural network model comprises the following steps:
step 301, constructing a fusion high-dimensional data matrix by using fiber real-time data and historical data;
step 302, PCA dimension reduction is carried out on the high-dimension data matrix;
step 303, performing domain-adaptive rough set cluster analysis on the data in step 302 to obtain a cluster center of an upper approximate set and a cluster center of a lower approximate set, and obtaining a cluster result after cluster analysis;
the edge equipment is further provided with an optical fiber state classification model; the method for generating the optical fiber state classification model comprises the following steps:
a feature clustering step;
a DCNN training step;
the feature clustering step and the DCNN training step are circularly executed, so that the depth feature discriminant and the model recognition performance of all categories are enhanced;
wherein, the characteristic clustering step includes: extracting features to obtain depth features; clustering the depth features according to the categories by using a k-means clustering method to obtain a distribution structure of the data in a DCNN feature space;
the DCNN training step comprises the following steps: calculating cross entropy loss and clustering metric loss according to the DCNN output, the sample label and the clustering result; performing joint optimization on the model output and the feature representation of the DCNN by using the cross entropy loss and the clustering metric loss;
the edge equipment is further provided with an optical fiber state prediction model; the method for generating the optical fiber state prediction model comprises the following steps:
step 501, sampling historical data, real-time data and accumulated fault data to establish a time sequence fault mode model based on a long-time and short-time memory network; the long-short-term memory network comprises: cell status and hidden status; the long-short-time memory network comprises the following three stages: forgetting the stage; the forgetting stage is used for selectively forgetting the input transmitted by the last node, controlling the part needing to be included in the information of the last state and the part needing to be forgotten, and finally obtaining an intermediate vector; a selection memory stage; the selection memory stage is used for: selectively memorizing the input; important part is recorded more, unimportant few records; an output stage; the output stage is used for deciding which will be used as the output of the current state through the output control function and performing scaling processing on the signal of the previous stage selection memory stage;
step 502, fusing optical fiber related priori mechanism models in each long-short-term memory network;
step 503, obtaining a plurality of time sequence fault mode models of the optical fiber under different service conditions after training;
step 504, obtaining fault mode curves of a plurality of optical fiber service stages through a plurality of time sequence fault mode models; each curve is fitted by a different time-series fiber performance decay or failure mode model;
step 505, performing approximate integration on each curve to obtain a damage degree prediction result of a specific optical fiber stage; the different prediction results form a result interval;
step 506, calculating the variance between different prediction curves to obtain the confidence coefficient of the prediction result at different time nodes;
step 507, feeding back new data to the model trained by the historical data, and detecting the prediction accuracy and precision of the model, and iterating continuously;
historical data, real-time data and accumulated fault data with time sequence information of the power optical fiber transmission network are mined, hidden information existing in the data is mined through a deep learning technology, a long and short time memory model structure is adopted, a mechanism model is fused, a plurality of time sequence fault mode models are established, prediction functions of a plurality of time sequence fault mode prediction models can be obtained, dynamic fault rate and damage degree prediction intervals of the optical fiber are obtained through approximate integral transformation, and confidence intervals of fault mode and damage degree prediction at specific time nodes are given through calculation of variance of prediction results; adopting a data iteration verification mode, verifying the accuracy and generalization performance of the existing model by using new data, and iterating and optimizing the model in the full life cycle of the optical fiber performance evolution;
firstly, sampling historical data, real-time data and accumulated fault data, and then establishing a time sequence fault mode model based on a long-short time memory network; the long-short time memory network LSTM is a deep learning architecture based on a cyclic neural network RNN, and the long-short time memory network has two transmission states, one is a cell state and the other is a hidden state; there are three phases inside the LSTM: the first phase is called forget phase, forget gate, and this phase is to forget the input from the last node selectively; forgetting to be unimportant, remembering to be important; the gating function obtained through calculation is used as forgetting gating, which information of the last state needs to be left and which needs to be forgotten is controlled, and finally an intermediate vector is obtained; the second stage is a selection memory stage, input gates, which selectively memorizes the inputs of this stage; the input state is selected and memorized; important part is recorded more, unimportant few records; the current input is represented by the previously calculated intermediate vector; the selected gating signal is controlled by a memory gating function, and the vector value transmitted to the next state can be obtained by adding the results obtained in the two steps; the third stage is called output stage, output gate, this stage will decide which will be the output of the current state, control through the output control function, also carry on the scaling process to the input gate signal of the previous stage;
after training, a plurality of time sequence fault mode models of the optical fiber under different service conditions can be obtained, the models can predict the fault mode of the optical fiber in a certain time period, the fault mode changes along with the change of time, the final model can obtain fault mode curves of a plurality of optical fiber service stages, each curve is fitted by different time sequence optical fiber performance decay or fault mode models, each curve is approximately integrated, damage degree prediction results of specific optical fiber stages can be obtained, different prediction results form a result interval, and the confidence of prediction results of nodes in different time periods can be obtained by calculating variances among different prediction curves; and continuously optimizing the model by adopting an iterative verification method, feeding new data back to the model trained by the historical data, and continuously iterating the model to obtain a dynamic fault mode prediction and damage degree prediction model with more and more excellent generalization performance and prediction results by detecting the prediction accuracy and precision of the model.
2. The optical fiber state multidimensional parameter monitoring system according to claim 1, wherein the information analysis by the convolutional neural network model comprises:
symbolizing knowledge in a plurality of fields and constructing a unified knowledge graph;
mapping the domain knowledge graph and massive optical fiber data to the same vector space by using a knowledge representation learning method and a multi-mode feature extraction method respectively, and maintaining the complex structure of knowledge and the inherent semantics of the data;
the analysis process and the analysis result are presented in the form of images, texts and maps for visual presentation for specific data analysis tasks and specific user groups.
3. The fiber optic state multi-dimensional parametric monitoring system of claim 2, wherein the visual presentation further comprises:
the behavior sequence data of the user and the corresponding system internal data are obtained through the past interaction records,
performing down-line training on the designed convolutional neural network model;
dynamically predicting interaction behavior of a user at the next moment on line, and possible influence on a system, communication among nodes and load conditions;
and allocating resources and coordinating user interaction behaviors through a deep reinforcement learning algorithm based on the predicted user behaviors and system states.
4. A fiber optic state multidimensional parameter monitoring system in accordance with claim 3, wherein said edge zone total node is further configured to: and (5) local data backup, namely storing the original data and the processed data information in a local storage.
5. A fiber optic state multidimensional parameter monitoring system in accordance with claim 3, wherein said edge zone total node is further configured to: uploading the data log to a cloud function, and periodically sending the log to a cloud server through a network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310509077.0A CN116614177B (en) | 2023-05-08 | 2023-05-08 | Optical fiber state multidimensional parameter monitoring system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310509077.0A CN116614177B (en) | 2023-05-08 | 2023-05-08 | Optical fiber state multidimensional parameter monitoring system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116614177A CN116614177A (en) | 2023-08-18 |
| CN116614177B true CN116614177B (en) | 2023-11-17 |
Family
ID=87677426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310509077.0A Active CN116614177B (en) | 2023-05-08 | 2023-05-08 | Optical fiber state multidimensional parameter monitoring system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116614177B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776901B (en) * | 2023-08-25 | 2024-04-30 | 深圳市爱德泰科技有限公司 | Optical fiber distribution frame label management system applied to electric power communication machine room |
| CN117674993B (en) * | 2023-11-09 | 2024-04-26 | 中交广州航道局有限公司 | Optical fiber network running state detection system and method |
| CN117250870B (en) * | 2023-11-16 | 2024-03-26 | 清控环境(北京)有限公司 | Reclaimed water recycling control system based on data information processing |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107769972A (en) * | 2017-10-25 | 2018-03-06 | 武汉大学 | A kind of power telecom network equipment fault Forecasting Methodology based on improved LSTM |
| CN110008898A (en) * | 2019-04-02 | 2019-07-12 | 中国计量大学 | Industrial equipment data edges processing method based on symbol and convolutional neural networks |
| CN110455397A (en) * | 2018-05-07 | 2019-11-15 | 光子瑞利科技（北京）有限公司 | A kind of shop equipment failure optical fiber sensing method and system based on LSTM |
| CN112286751A (en) * | 2020-11-24 | 2021-01-29 | 华中科技大学 | Intelligent diagnosis system and method for high-end equipment fault based on edge cloud cooperation |
| CN112394702A (en) * | 2020-12-10 | 2021-02-23 | 安徽理工大学 | Optical cable manufacturing equipment fault remote prediction system based on LSTM |
| CN114063601A (en) * | 2021-11-12 | 2022-02-18 | 江苏核电有限公司 | Equipment state diagnosis system and method based on artificial intelligence |
| CN114362367A (en) * | 2021-12-30 | 2022-04-15 | 中国电力科学研究院有限公司 | Cloud edge cooperation-oriented power transmission line monitoring system and method, and cloud edge cooperation-oriented power transmission line identification system and method |
| CN115798131A (en) * | 2023-02-13 | 2023-03-14 | 成都陆迪盛华科技有限公司 | Multi-dimensional characteristic intrusion detection method based on distributed optical fiber |
-
2023
- 2023-05-08 CN CN202310509077.0A patent/CN116614177B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107769972A (en) * | 2017-10-25 | 2018-03-06 | 武汉大学 | A kind of power telecom network equipment fault Forecasting Methodology based on improved LSTM |
| CN110455397A (en) * | 2018-05-07 | 2019-11-15 | 光子瑞利科技（北京）有限公司 | A kind of shop equipment failure optical fiber sensing method and system based on LSTM |
| CN110008898A (en) * | 2019-04-02 | 2019-07-12 | 中国计量大学 | Industrial equipment data edges processing method based on symbol and convolutional neural networks |
| CN112286751A (en) * | 2020-11-24 | 2021-01-29 | 华中科技大学 | Intelligent diagnosis system and method for high-end equipment fault based on edge cloud cooperation |
| CN112394702A (en) * | 2020-12-10 | 2021-02-23 | 安徽理工大学 | Optical cable manufacturing equipment fault remote prediction system based on LSTM |
| CN114063601A (en) * | 2021-11-12 | 2022-02-18 | 江苏核电有限公司 | Equipment state diagnosis system and method based on artificial intelligence |
| CN114362367A (en) * | 2021-12-30 | 2022-04-15 | 中国电力科学研究院有限公司 | Cloud edge cooperation-oriented power transmission line monitoring system and method, and cloud edge cooperation-oriented power transmission line identification system and method |
| CN115798131A (en) * | 2023-02-13 | 2023-03-14 | 成都陆迪盛华科技有限公司 | Multi-dimensional characteristic intrusion detection method based on distributed optical fiber |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116614177A (en) | 2023-08-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116614177B (en) | Optical fiber state multidimensional parameter monitoring system | |
| CN108564254B (en) | Power distribution equipment state visualization platform based on big data | |
| CN113156917B (en) | Power grid equipment fault diagnosis method and system based on artificial intelligence | |
| CN109146093A (en) | A kind of electric power equipment on-site exploration method based on study | |
| CN109872003A (en) | Obj State prediction technique, system, computer equipment and storage medium | |
| CN114723285B (en) | Power grid equipment safety evaluation prediction method | |
| CN116308304B (en) | New energy intelligent operation and maintenance method and system based on meta learning concept drift detection | |
| CN115423009A (en) | Cloud edge coordination-oriented power equipment fault identification method and system | |
| CN117391674A (en) | Reliability-based preventive maintenance optimization method and device for electrical equipment | |
| CN116796168A (en) | CNN-BiLSTM high-altitude multi-factor power transmission line audible noise prediction method based on multi-head attention mechanism | |
| CN117033923A (en) | Method and system for predicting crime quantity based on interpretable machine learning | |
| CN113961708B (en) | Power equipment fault tracing method based on multi-level graph convolutional network | |
| Hou et al. | A federated learning-based fault detection algorithm for power terminals | |
| CN114265837A (en) | Station-side interactive data feature extraction method | |
| CN115598459A (en) | Power failure prediction method for 10kV feeder line fault of power distribution network | |
| CN114487643A (en) | On-spot handing-over of extra-high voltage GIL equipment is accepted and is synthesized test platform | |
| CN110837932A (en) | Thermal power prediction method of solar heat collection system based on DBN-GA model | |
| CN112801815B (en) | Power communication network fault early warning method based on federal learning | |
| CN117833243B (en) | Method, system, equipment and medium for predicting short-term demand of electric power | |
| Bai et al. | Abnormal Detection Scheme of Substation Equipment based on Intelligent Fusion Terminal | |
| CN117148048B (en) | Power distribution network fault prediction method and system based on digital twin technology | |
| Xue et al. | State evaluation method of distribution equipment based on health index in big data environment | |
| Huo et al. | Fault Prediction of IoT Terminals based on Improved ResNet and BiLSTM Models | |
| Dong et al. | Log fusion technology of power information system based on fuzzy reasoning | |
| CN116131444A (en) | Intelligent analysis system for photovoltaic power station |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |