Intelligent analysis method and device for dumb resource terminal big data and medium
By introducing a federated learning network into the fiber optic communication network, the real-time data of the fiber optic link is collected and analyzed, the problem of the quality of the fiber optic link cannot be actively perceived, the intelligent analysis and prediction management of the fiber optic link is realized, and the network operation is improved Reliability and efficiency.
Patent Information
- Application Number
- CN202411049882.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-01
- Publication Date
- 2024-12-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
There are a large number of "dumb resources" in fiber optic communication networks, and their quality cannot be actively perceived and finely managed, resulting in bottlenecks in the digitalization and intelligence of the network, especially the quality of fiber optic links is difficult to monitor and judge, affecting network operation.
Introduce a federated learning network in a multi-node perception unit, generate target optical power data and optical signal identification data by collecting and processing real-time data, and use prediction models to perform intelligent analysis to identify abnormal nodes and fiber link status to achieve fiber link quality active perception and real-time assessment.
It improves the active perception of fiber link quality, improves network operation and maintenance efficiency and reliability, ensures the healthy operation of fiber networks, and solves the problem that fiber link quality cannot be actively perceived.
Smart Images

Figure CN119071672A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of optical fiber communication digital technology, and in particular to an intelligent analysis method, device and medium for dumb resource terminal big data. Background Art
[0002] At present, there are a large number of optical fiber resources in the communication network, and optical fiber is a "dumb resource". Its quality cannot be actively perceived and finely managed, which has become a bottleneck for network digitization and intelligence. The quality of optical fiber cannot be actively perceived, mainly based on the principle of optical fiber communication and the physical medium characteristics of optical fiber. In addition, the communication network involves multiple different nodes. In the construction and expansion of communication construction, as well as the business cutover and resource adjustment of operation, there is a problem that the resource data of each node is inconsistent with the actual on-site optical fiber terminals; and the quality performance of the optical fiber link is usually difficult to monitor and judge the quality performance and occupancy status of optical fibers without normal business signals, including idle and unused or unoccupied optical fibers, and these performance parameter data are difficult to perceive.
[0003] The existing technologies for big data of dumb resource terminals in optical fiber networks mainly develop intelligent ODN technology and optical fiber terminal information sampling. However, these technologies mainly focus on data collection and passive perception, and generally lack the complete functions of perception, learning, analysis, prediction and management. There is a lack of digital active perception, intelligent learning analysis, active prediction and processing; optical fiber problems with abnormal signal attenuation are difficult to predict and actively discover, the occupancy status of different nodes of dumb resource terminals is unclear, and the quality of optical fiber links cannot be actively perceived, resulting in the inability to guarantee the operation of optical fiber networks. Summary of the invention
[0004] The purpose of this application is to provide an intelligent analysis method, device and medium for dumb resource terminal big data, which can comprehensively and accurately grasp the operation status of the optical fiber links of different nodes of the dumb resource terminals, improve the active perception of the quality of the optical fiber links, and improve the operation and maintenance efficiency and reliability of the network through intelligent learning analysis and predictive management, thereby ensuring network operation.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides an intelligent analysis method for dumb resource terminal big data, comprising:
[0007] A federated learning network is introduced in a multi-node sensing unit; the multi-node sensing unit is a network composed of multiple nodes; the nodes are connected by optical fiber links; each node is used to independently collect, process and transmit real-time data; the multi-node sensing unit is used to collect dumb resource terminal big data; the dumb resource terminal big data includes the real-time data of each node;
[0008] Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to generate target real-time data for each node; the target real-time data includes: target optical power data and target optical signal identification data; the target optical power data represents the intensity of the optical signal received or sent by each node, which is used to evaluate the quality of the optical fiber link; the target optical signal identification data represents the characteristic information related to the optical signal itself in each node, and the characteristic information includes the transmission service identification of the optical signal in the optical fiber;
[0009] Input the target optical power data and the target optical signal identification data into a prediction model to generate a first prediction value and a first deviation corresponding to the target optical power data after attenuation, and a second prediction value and a target deviation corresponding to the target optical signal identification data; wherein the target deviation is determined based on the second prediction value corresponding to the target optical signal identification data, the preset service content signal identification data, and the preset service content signal identification data;
[0010] Based on the first deviation and the first preset threshold, determining the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation, and alarming the optical fiber link of the abnormal node;
[0011] Based on the target deviation, determining the use state of the optical fiber link of each node as an idle state or a service occupied state, so as to perform a real-time evaluation on the optical fiber health operation of each optical fiber link;
[0012] According to the optical fiber link of the abnormal node and the optical fiber link usage status of each node, the optical fiber operation status of the dumb resource terminal big data on the optical fiber links of different nodes is determined.
[0013] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the intelligent analysis method of dumb resource terminal big data.
[0014] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the intelligent analysis method of dumb resource terminal big data.
[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0016] The present invention introduces a federated learning network in a multi-node perception unit to promote nodes with scarce data or limited resources to participate in a large-scale, efficient and secure joint learning process, solve the problem of insufficient node data, solve the problem of data not leaving the node, and improve the security of the data. Through the prediction model, the target optical power data and the target optical signal identification data are intelligently analyzed to obtain the first predicted value and the first deviation corresponding to the attenuation of the target optical power data, as well as the second predicted value and the target deviation corresponding to the target optical signal identification data, so as to realize the monitoring and tracking of the optical power data of different nodes and the optical signal identification data of different nodes. Specifically, the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation is determined according to the first deviation, and the optical fiber link of the abnormal node is alarmed, so as to realize the prediction of the optical fiber link of the node with potential faults, improve the guarantee of the operation quality of the optical fiber network, and at the same time, based on the second deviation, determine that the optical fiber links of different nodes of the dumb resource terminal are in an idle state or a business occupied state, so as to realize the optical fiber health operation of the optical fiber links of each node, and perform real-time evaluation. Finally, the real-time operation of the optical fiber links of the dumb resource terminal big data at different nodes is fully and accurately grasped, the active perception of the quality of the optical fiber link is improved, and the network operation is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0018] Figure 1 A schematic diagram of a flow chart of an intelligent analysis method for dumb resource terminal big data provided in an embodiment of the present application;
[0019] Figure 2 This is a structural block diagram of an intelligent analysis method for dumb resource terminal big data provided in an embodiment of the present application. DETAILED DESCRIPTION
[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0021] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.
[0022] like Figure 1 As shown, the present application provides an intelligent analysis method for dumb resource terminal big data, including:
[0023] Step 101: Introduce a federated learning network in a multi-node perception unit 1; the multi-node perception unit 1 is a network composed of multiple nodes; the nodes are connected by optical fiber links; each node is used to independently collect, process and transmit real-time data; the multi-node perception unit 1 is used to collect dumb resource terminal big data; the dumb resource terminal big data includes the real-time data of each node.
[0024] Among them, in the multi-node sensing unit 1, a node usually refers to an optical fiber transmission node in a communication network, which is commonly found in various communication rooms. These nodes are distributed in the network and jointly complete the tasks of data collection, processing and transmission.
[0025] In practical applications, a collection system is constructed using multi-node sensing units 1, and dumb resource terminal big data is collected in time sequence and written into a database, which covers real-time data of fiber optic links of n different nodes. Considering the different processing capabilities of different node terminals, in order to expand the universality of multiple nodes, the collection adopts a method of selecting matrix vector data of different sizes of 7*24, 12*24 and 365*24 formed according to different time lengths.
[0026] Step 102: Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to generate target real-time data for each node; the target real-time data includes: target optical power data and target optical signal identification data; the target optical power data represents the intensity of the optical signal received or sent by each node, which is used to evaluate the quality of the optical fiber link; the target optical signal identification data represents the characteristic information related to the optical signal itself in each node, and the characteristic information includes the service identification of the optical signal in the optical fiber.
[0027] Among them, the federated learning network can effectively solve the problem of insufficient node data and promote nodes with scarce data or limited resources to participate in large-scale, efficient and secure joint learning processes.
[0028] In some embodiments, based on the federated learning network and preset rules, the dumb resource terminal big data is processed to generate target real-time data for each node, specifically including: based on the federated learning network and preset rules, the dumb resource terminal big data is processed to perform online learning or offline learning; in the case of online learning, the dumb resource terminal big data is centrally processed using a learning model to generate a first learning result; in the case of offline learning, the learning model is dispersedly sunk to the multi-node perception unit 1 to process the dumb resource terminal big data to generate a second learning result; based on the first learning result and the second learning result, a general knowledge base is generated; based on the general knowledge base, the target real-time data of each node is obtained.
[0029] In some embodiments, based on the federated learning network and preset rules, the dumb resource terminal big data is processed for online learning or offline learning, specifically including: the preset rules include node performance thresholds and data security requirement thresholds; for the real-time data of any node in the dumb resource terminal big data, when the performance corresponding to the node is not less than the node performance threshold, the real-time data is subjected to online learning; otherwise, the real-time data is subjected to offline learning; the performance includes the storage capacity of the memory and the real-time computing processing capability of the processor; for the real-time data of any node in the dumb resource terminal big data, when the data security requirement corresponding to the real-time data is not less than the data security requirement threshold, the real-time data is subjected to offline learning; otherwise, the real-time data is subjected to online learning.
[0030] The processor includes a CPU and a GPU.
[0031] In practical applications, each node can flexibly choose offline learning mode or online learning mode according to the different configurations of the node's computing processing and storage capabilities for local data learning and the different security requirements of whether the data is allowed to leave the local node.
[0032] Among them, the online learning mode uses the real-time data collected by each node, combined with the online knowledge base, and uses the learning model for learning. In this mode, the real-time data of each node is gathered online to a centralized computer database to achieve learning and data exchange.
[0033] Among them, the offline learning mode uses the real-time data collected by a single node, combined with the offline knowledge base, and uses the learning model for learning. In this mode, the own data of each node does not leave the local. Under the condition of ensuring data security and privacy, data aggregation is achieved to improve the performance of each node. Among them, the offline knowledge base uses the big data of historical dumb resource terminals of similar network nodes as a training set, combined with the comprehensive parameter fine-tuning learning of the single node.
[0034] Step 103: Input the target optical power data and the target optical signal identification data into a prediction model to generate a first prediction value and a first deviation corresponding to the attenuation of the target optical power data, as well as a second prediction value and a target deviation corresponding to the target optical signal identification data; wherein the target deviation is determined based on the second prediction value corresponding to the target optical signal identification data, preset business content signal identification data, and preset business content signal identification data.
[0035] Among them, the prediction model is generally deployed on the central cloud server. After the multi-node perception unit 1 collects the data of the dumb resource terminal, the data is uploaded to the prediction model for analysis and judgment. The initial input data of the prediction model is the real-time data of each node historically collected by the dumb resource terminal in the same network, and fine-tuned in combination with the life span, environmental conditions and comprehensive parameters of the optical fiber link weight of the actual network. During offline learning, the trained and dispersed part of the prediction model can be deployed on the node devices that need offline learning.
[0036] Among them, the first prediction value corresponding to the attenuation of the target optical power data includes backward prediction values, that is, backward prediction values of different lengths of future days / weeks / months.
[0037] In some embodiments, the construction process of the prediction model specifically includes: the prediction model includes CNN, multi-layer LSTM and attention mechanism module; the target optical power data and the target optical signal identification data are input into CNN to obtain local features; wherein the target optical power data and the target optical signal identification data are time series data of different lengths; the local features are input into the multi-layer LSTM to generate long-term dependencies in the time series data; the long-term dependencies are input into the attention mechanism module to generate the first target feature and the second target feature of the time series data of different lengths, and the first weight corresponding to the first target feature and the second weight corresponding to the second target feature; based on the first target feature and the second target feature, and the first weight corresponding to the first target feature and the second weight corresponding to the second target feature, determine the target first prediction value and the target second prediction value; the target first prediction value is the first prediction value corresponding to the attenuation of the target optical power data; the target second prediction value is the second prediction value corresponding to the target optical signal identification data.
[0038] Among them, the prediction model can be embedded in front-end fiber monitoring, back-end platform and management control equipment, with strong flexibility and support for endogenous intelligence generation. Combined with the network characteristics of fiber dumb resources, multiple attention modules such as life span, environmental conditions and fiber link weight comprehensive parameters are adopted to monitor the fiber network in accordance with the actual situation on site, which is more targeted and accurate. By monitoring and tracking data such as optical power and optical signals, the healthy operation of the optical fiber is evaluated in real time, and hidden faults such as aging and attenuation are predicted to improve the guarantee of the operation quality of the optical fiber network.
[0039] Step 104: Based on the first deviation and the first preset threshold, determine the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation, and issue an alarm for the optical fiber link of the abnormal node.
[0040] Among them, by identifying the optical fiber link early warning of abnormal nodes corresponding to abnormal optical fiber attenuation, it is possible to track and analyze the attenuation trend in real time over a long period of time, alleviate the problem of difficulty in finding abnormal optical fibers, and eliminate the hidden dangers of optical fiber aging.
[0041] In some embodiments, based on the first deviation and the first preset threshold, the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation is determined, and an alarm is issued for the optical fiber link of the abnormal node, specifically including: subtracting the measured value corresponding to the target optical power data from the first predicted value corresponding to the target optical power data after attenuation to obtain the first deviation; comparing whether the first deviation is not less than the first preset threshold to obtain a first result; if the first result is yes, it is determined that the attenuation of the target optical power data is abnormal, and the node corresponding to the target optical power data is determined as the optical fiber link of the abnormal node, and an alarm is issued for the optical fiber link of the abnormal node; if the first result is no, the optical fiber link of the node corresponding to the target optical power data is determined as the optical fiber link of a normal node.
[0042] Step 105: Based on the target deviation, determine whether the use state of the optical fiber link of each node is an idle state or a service occupied state, so as to perform a real-time evaluation on the optical fiber health operation status of the optical fiber link of each node.
[0043] Among them, the target deviation can quickly confirm whether there are nodes with idle optical fiber resources, and assist in judging whether the optical fiber is working normally, solving the problem of unclear optical fiber resource terminal occupancy status.
[0044] In some embodiments, the calculation process of the target deviation specifically includes: subtracting the preset signal identification data containing business content from the second predicted value corresponding to the target optical signal identification data to obtain a second deviation; subtracting the preset signal identification data not containing business content from the second predicted value corresponding to the target optical signal identification data to obtain a third deviation; the target deviation includes the second deviation and the third deviation.
[0045] In some embodiments, based on the target deviation, the fiber optic link status of each node is determined to be an idle state or an occupied state, specifically including: comparing whether the second deviation is less than a second preset threshold value to obtain a second result; if the second result is yes, determining that the fiber optic link of the node is in a business occupied state; if the second result is no, the fiber optic link of the node is in an idle state; comparing whether the third deviation is greater than a third preset threshold value to obtain a third result; if the third result is yes, determining that the fiber optic link of the node is in an idle state; if the third result is no, the fiber optic link of the node is in a business occupied state.
[0046] In practical applications, refer to Figure 2, the multi-node perception unit 1 collects dumb resource terminal big data, and the intelligent learning analysis unit 2 (i.e., the prediction model) uses the optical power data and optical signal identification data obtained after processing the collected historical dumb resource terminal big data as the training set data. After the training is completed, it is stored in the general knowledge base, and the target optical power data and the target optical signal identification data obtained after processing the collected current dumb resource terminal big data are used as the test set to predict and obtain the first prediction value, and the prediction model is predicted backward as needed to obtain the backward prediction value of different lengths of future days / weeks / months. The target optical power data and the target optical signal identification data obtained after processing the dumb resource terminal big data of a certain period (such as a quarter or a year) are converted into a training set for iterative training and updating of the prediction model. Further, the dumb resource terminal big data is used as the measured data and the predicted value for comprehensive analysis. When the first deviation exceeds the deviation limit, the early warning link of the optical fiber abnormality of the decision and optimization unit 3 will be entered to process and control the optical fiber link of the abnormal node. When the target optical signal identification data has long-term regularity data and there is a large deviation from the signal identification data containing business content, the decision and optimization unit 3 will be entered to identify and judge the regular target optical signal identification data, that is, compared with the signal identification data containing business content, if the second deviation is large enough, it is determined that the optical fiber link of the node does not contain the signal identification data of business content, and the optical fiber link of the node is determined to be idle, and compared with the signal identification data without business content, if the third deviation is small enough, it is determined that the optical fiber link of the node is idle, and the target real-time data is processed and controlled according to the business occupancy status of the optical fiber link of the current node. In this way, the incorrect occupancy and occupancy optimization of optical fiber resources are corrected, and available optical fiber resources are mined.
[0047] Step 106: According to the optical fiber link of the abnormal node and the usage status of the optical fiber link of each node, the optical fiber operation status of the dumb resource terminal big data on the optical fiber links of different nodes is determined.
[0048] Among them, through the alarm of abnormal node fiber optic links and the analysis of the usage status of node fiber optic links, the fiber utilization rate can be improved and the investment in optical cables can be saved. Through 24-hour online monitoring and analysis, the optical path operation status of passive optical fiber dumb resources can be fully and accurately grasped, the active perception of fiber optic link quality can be improved, and network operation can be guaranteed.
[0049] In some embodiments, the target optical power data and the target optical signal identification data are input into the prediction model to generate a first prediction value and a first deviation corresponding to the attenuation of the target optical power data, which then includes: determining a high-quality target node based on the first prediction value and a preset attenuation threshold corresponding to the attenuation of the target optical power data; and determining the optimal path for multi-node transmission in the network using multi-criteria decision analysis based on the high-quality target nodes, optical fiber distance, and line capacity.
[0050] Among them, Multi-Criteria Decision Analysis (MCDA) is a systematic method for dealing with decision problems involving multiple conflicting or incommensurable (i.e., not directly comparable) criteria. In many real-world scenarios, decision makers are faced with not only a single goal, but also need to find the best balance between multiple goals. MCDA provides a series of tools and techniques to help decision makers systematically evaluate and compare different options to make more informed decisions.
[0051] Specifically, multi-criteria decision analysis is used to optimize and judge the node transmission route to determine the optimal path. Using the first predicted value corresponding to the attenuation of the target optical power data, the high-quality target node with better data can be selected. Combined with other factors such as fiber distance and line capacity, multi-criteria decision analysis is used to select the optimal transmission path for multiple nodes in the network. During the optimization process, the above corresponding factors are comprehensively considered according to multiple criteria, a weight is assigned to each criterion, and the total score of each path is calculated, and finally the optimal path is selected. In this way, not only the problem of the non-active perception of the quality of the optical fiber link is solved, but also the optimization and improvement of the quality of the transmission network is promoted. To a certain extent, the optimization of network transmission quality and the reduction of network latency are achieved.
[0052] Dumb resource terminal big data introduces different matrix vectors of 7*24, 12*24 and 365*24. Different applicable matrices can be adopted from low to high according to the computing processing capabilities of the network hardware terminal, which improves universality and promotion.
[0053] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0054] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0055] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0056] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRdM), magnetic memory (MRdM), ferroelectric memory (FRdM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RdM) or external cache memory, etc. By way of illustration and not limitation, RdM may be in various forms, such as static random access memory (SRdM) or dynamic random access memory (DRdM).
[0057] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0058] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0059] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. An intelligent analysis method for dumb resource terminal big data, It is characterized in that include: Introducing federated learning networks in multi-node sensing units; The multi-node sensing unit is a network composed of multiple nodes; The nodes are connected by optical fiber links; Each node is used to independently collect, process and transmit real-time data; the multi-node sensing unit is used to collect dumb resource terminal big data; dumb resource terminal big data includes the real-time data of each node; Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to generate target real-time data for each node; The target real-time data includes: target optical power data and target optical signal identification data; the target optical power data represents the intensity of the optical signal received or sent by each node, and is used to evaluate the quality of the optical fiber link; the target optical signal identification data represents the characteristic information related to the optical signal itself in each node, and the characteristic information includes the transmission service identification of the optical signal in the optical fiber; Input the target optical power data and the target optical signal identification data into a prediction model to generate a first prediction value and a first deviation corresponding to the target optical power data after attenuation, and a second prediction value and a target deviation corresponding to the target optical signal identification data; wherein the target deviation is determined based on the second prediction value corresponding to the target optical signal identification data, the preset service content signal identification data, and the preset service content signal identification data; Based on the first deviation and the first preset threshold, determining the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation, and alarming the optical fiber link of the abnormal node; Based on the target deviation, determining the use state of the optical fiber link of each node as an idle state or a service occupied state, so as to perform a real-time evaluation on the optical fiber health operation of each optical fiber link; According to the optical fiber link of the abnormal node and the optical fiber link usage status of each node, the optical fiber operation status of the dumb resource terminal big data on the optical fiber links of different nodes is determined.
2. According to claim 1, the intelligent analysis method of dumb resource terminal big data, It is characterized in that Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to generate target real-time data for each node, specifically including: Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to perform online learning or offline learning; In the case of the online learning, the dumb resource terminal big data is centrally processed using a learning model to generate a first learning result; In the case of offline learning, the learning model is used to be decentralized and sunk to the multi-node sensing unit, the dumb resource terminal big data is processed, and a second learning result is generated; Generate a general knowledge base based on the first learning result and the second learning result; Based on the general knowledge base, target real-time data of each node is obtained.
3. According to claim 2, the intelligent analysis method of dumb resource terminal big data, It is characterized in that Based on the federated learning network and preset rules, the dumb resource terminal big data is processed to perform online learning or offline learning, specifically including: The preset rules include a node data processing performance threshold and a data security requirement threshold; For the real-time data of any node in the dumb resource terminal big data, when the data processing performance corresponding to the node is not less than the node data processing performance threshold, the real-time data is learned online; otherwise, the real-time data is learned offline; the performance includes the storage capacity of the memory and the real-time computing processing capacity of the processor; For the real-time data of any node in the dumb resource terminal big data, when the data security requirement corresponding to the real-time data is not less than the data security requirement threshold, offline learning is performed on the real-time data; otherwise, online learning is performed on the real-time data.
4. The intelligent analysis method of dumb resource terminal big data according to claim 1, It is characterized in that Based on the first deviation and the first preset threshold, determining the optical fiber link of the abnormal node corresponding to the optical fiber with abnormal attenuation, and alarming the optical fiber link of the abnormal node, specifically includes: Subtracting a measured value corresponding to the target optical power data from a first predicted value corresponding to the attenuated target optical power data to obtain the first deviation; Comparing whether the first deviation is not less than a first preset threshold value, and obtaining a first result; If the first result is yes, the target optical power data attenuation is determined to be abnormal, and the optical fiber link of the node corresponding to the target optical power data is determined to be an abnormal node optical fiber link, and an alarm is issued for the abnormal node optical fiber link; if the first result is no, the optical fiber link of the node corresponding to the target optical power data is determined to be a normal node optical fiber link.
5. The intelligent analysis method of dumb resource terminal big data according to claim 1, It is characterized in that The calculation process of the target deviation specifically includes: Subtracting the preset signal identification data containing business content from the second prediction value corresponding to the target optical signal identification data to obtain a second deviation; Subtract the preset data of the signal identification without service content from the second predicted value corresponding to the target optical signal identification data to obtain a third deviation; The target deviation includes the second deviation and the third deviation.
6. The intelligent analysis method of dumb resource terminal big data according to claim 5, It is characterized in that Based on the target deviation, determining the use state of the optical fiber link of each node as an idle state or a business occupied state specifically includes: Comparing whether the second deviation is less than a second preset threshold value to obtain a second result; if the second result is yes, determining that the optical fiber link of the node is in a service occupied state; If the second result is no, determining that the optical fiber link of the node is in an idle state; Comparing whether the third deviation is greater than a third preset threshold value to obtain a third result; if the third result is yes, determining that the optical fiber link of the node is in an idle state; If the third result is no, it is determined that the optical fiber link of the node is in a service occupied state.
7. The intelligent analysis method of dumb resource terminal big data according to claim 1, It is characterized in that The prediction model construction process specifically includes: the prediction model includes CNN, multi-layer LSTM and attention mechanism modules; Input the target optical power data and the target optical signal identification data into CNN to obtain local features; wherein the target optical power data and the target optical signal identification data are both time series data of different lengths; Inputting the local features into a multi-layer LSTM to generate long-term dependencies in the time series data; Inputting the long-term dependency into an attention mechanism module to generate a first target feature and a second target feature of time series data of different lengths, and a first weight corresponding to the first target feature and a second weight corresponding to the second target feature; Based on the first target feature and the second target feature, and the first weight corresponding to the first target feature and the second weight corresponding to the second target feature, a target first prediction value and a target second prediction value are determined; the target first prediction value is the first prediction value corresponding to the target optical power data attenuation; the target second prediction value is the second prediction value corresponding to the target optical signal identification data.
8. The intelligent analysis method of dumb resource terminal big data according to claim 1, It is characterized in that The target optical power data and the target optical signal identification data are input into a prediction model to generate a first prediction value and a first deviation corresponding to the attenuated target optical power data, and then the method further includes: Determine a high-quality target node based on a first predicted value corresponding to the attenuation of the target optical power data and a preset attenuation threshold; According to the high-quality target nodes, optical fiber distance and line capacity, multi-criteria decision analysis is used to determine the optimal path for multi-node transmission in the network.
9. A computer device, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent analysis method for dumb resource terminal big data described in any one of claims 1-8.
10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the intelligent analysis method for dumb resource terminal big data described in any one of claims 1 to 8 is implemented.
Citation Information
Cited By
Micro-service-oriented inter-container abnormal information tracing method and system
CN120915700A
A microservice-oriented inter-container exception information tracing method and system
CN120915700B
