Communication optical cable monitoring method and system for ocean multi-dimensional data

Abstract

The invention relates to the technical field of optical cable monitoring, in particular to a communication optical cable monitoring method and system for ocean multi-dimensional data, and the method comprises the steps: S1, obtaining a plurality of training data; s2, obtaining the number of core training data in all the obtained training data, and obtaining an analysis error rate; s3, checking whether a first pause condition is met or not, if yes, pausing the training processing for the analysis model, if not, continuing to check whether a second pause condition is met or not, if not, continuing to obtain a plurality of training data, skipping to S2, and if yes, pausing the training processing for the analysis model; s4, performing training processing on the analysis model by using all correct training data in all the obtained training data; and S5, generating feature data based on the monitoring data of the communication optical cable, and inputting the feature data into the trained analysis model. The analysis accuracy of the analysis model can be improved.

Description

Technical Field

[0001] The present application relates to the field of optical cable monitoring technology, and in particular to a method and system for monitoring communication optical cables for multi-dimensional marine data. Background Art

[0002] Currently, different sensors can be deployed to collect monitoring data of submarine communication optical cables. By analyzing the collected monitoring data of submarine communication optical cables, the status of the submarine communication optical cables can be known.

[0003] Similar prior art includes a Chinese patent application with publication number CN118826850A, which provides a method and system for monitoring the status of a power communication optical cable, and relates to the field of optical cable operation and maintenance technology. The specific technical scheme is as follows: step S1, processing known optical power data and known optical attenuation data to obtain original data, and building a feature library and a rule library based on the original data, wherein the rule library has multiple preset judgment rules; step S2, obtaining real-time alarm data of the power communication optical cable, and monitoring the real-time alarm data based on the preset judgment rules and the feature library to obtain real-time monitoring results; the real-time alarm data includes real-time optical power data and real-time optical attenuation data. In addition, similar prior art includes a Chinese patent application with publication number CN118890090A, which discloses an intelligent optical cable monitoring method and system based on digital twins, including: Step 1: Collecting real-time optical cable performance data of N optical cable segments, and screening out I optical cable segments with suspected transmission fault marks based on the real-time optical cable performance data; Step 2: Analyzing the optical cable segments corresponding to the suspected transmission fault marks to obtain the transmission fault levels of the corresponding optical cable segments; Step 3: Using a pre-built transmission fault analysis digital twin model, simulate and analyze the real-time optical cable performance data to generate simulation prediction data.

[0004] However, the reliability of the monitoring results obtained by the above two patent applications needs to be improved. Therefore, this application proposes a communication optical cable monitoring method and system for multi-dimensional marine data. Summary of the invention

[0005] The present application obtains a number of training data multiple times, uses all the obtained training data multiple times to perform training processing on the analysis model, determines whether it is time to suspend the training processing, removes a number of problematic training data from all the obtained training data, and uses all the correct training data for training processing. The present application aims to improve the analysis accuracy of the analysis model.

[0006] The present application provides a communication optical cable monitoring method for multi-dimensional ocean data, comprising the following steps: S1. The remote analysis module obtains a number of training data, where the training data includes historical monitoring data of the communication optical cable and historical analysis results corresponding to the historical monitoring data of the communication optical cable; S2, the remote analysis module uses all the acquired training data to perform training processing on the analysis model, obtains the number of core training data in all the acquired training data, and obtains the analysis error rate based on the trained analysis model; S3, the remote analysis module checks whether the first suspension condition is met according to the number of core training data obtained this time, and if so, the remote analysis module suspends the training process for the analysis model, and if not, the remote analysis module continues to check whether the second suspension condition is met according to the analysis error rate obtained this time, and if not, the remote analysis module continues to obtain a number of training data and jumps to S2, and if so, the remote analysis module suspends the training process for the analysis model; S4, the remote analysis module checks all the acquired training data to determine a number of problematic training data among all the acquired training data, and uses all the correct training data among all the acquired training data to perform training processing on the analysis model; S5. The acquisition module collects the monitoring data of the communication optical cable from different sensor modules, and the acquisition module transmits the monitoring data of the communication optical cable to the remote analysis module through the communication module. The remote analysis module generates feature data based on the monitoring data of the communication optical cable, and inputs the feature data into the trained analysis model. The analysis model outputs the analysis result on the status of the communication optical cable.

[0007] As a preferred technical solution of the present application, the process of the remote analysis module obtaining the analysis error rate based on the trained analysis model includes: obtaining different test data, the test data including historical monitoring data of the communication optical cable, and the actual analysis results corresponding to the historical monitoring data of the communication optical cable, respectively inputting the historical monitoring data of the communication optical cable included in all the test data into the trained analysis model, respectively obtaining the output results of the analysis model, and counting the number of output results that are different from the actual analysis results included in the corresponding test data, and calculating the ratio to the number of all the test data.

[0008] As a preferred technical solution of the present application, the process of the remote analysis module checking whether the first pause condition is met includes: calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than a preset difference threshold, it is considered that the first pause condition is met, or calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than the preset difference threshold, continuing to check whether the number of consecutive occurrences of the situation in which the difference is less than the preset difference threshold in history exceeds the preset consecutive occurrence threshold, and if so, it is considered that the first pause condition is met.

[0009] As a preferred technical solution of the present application, the process of the remote analysis module checking whether the second pause condition is met includes: checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time, if so, it is considered that the second pause condition is met, or checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time, if so, continuing to check whether the number of consecutive occurrences of the situation in history where the analysis error rate obtained this time is greater than the analysis error rate obtained last time exceeds a preset consecutive occurrence threshold, if so, it is considered that the second pause condition is met.

[0010] As a preferred technical solution of the present application, the remote analysis module verifies and processes all the acquired training data, including the following steps: S411, the remote analysis module divides all the acquired training data into a preset number of first groups, so that the number of different training data included in each first group is the same; S412, the remote analysis module sets a preset number of second combinations, assigns values ​​from small to large to each second combination in sequence, and determines each second combination in sequence according to the corresponding values ​​from small to large, selects a first combination from all the first combinations for the first determined second combination, adds the selected first combination to the first determined second combination, selects another first combination from all the other first combinations that have not been selected for each second combination determined thereafter, and adds the selected other first combination together with the content of the previous second combination to the second combination; S413: For each second combination, the remote analysis module generates a verification model by training using different training data included in the second combination.

[0011] As a preferred technical solution of the present application, the remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S421, the remote analysis module obtains a verification data, the verification data includes historical monitoring data of the communication optical cable, and a real analysis result corresponding to the historical monitoring data of the communication optical cable, and for each verification model, the remote analysis module inputs the historical monitoring data of the communication optical cable included in the verification data into the verification model to obtain an output result of the verification model; S422, for each verification model, the remote analysis module checks whether the output result of the verification model is consistent with the real analysis result corresponding to the historical monitoring data of the communication optical cable included in the verification data, and the remote analysis module respectively classifies the second combination corresponding to the verification model with the corresponding inspection result being consistent into the first category, and respectively classifies the second combination corresponding to the verification model with the corresponding inspection result being inconsistent into the second category; S423. The remote analysis module determines the second combination in the first category that contains the largest number of all first combinations, determines the second combination in the second category that contains the smallest number of all first combinations, and obtains the difference first combination between the determined second combination in the first category and the determined second combination in the second category.

[0012] As a preferred technical solution of the present application, the remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S431, the remote analysis module determines the verification models corresponding to the respective second combinations in the first category, and for each determined verification model, inputs the historical monitoring data of the communication optical cable included in the different training data distinguishing the first combination into the verification model, and obtains the output results of the verification model respectively; S432. With respect to each training data in the distinguished first combination, the remote analysis module counts a first number of verification models whose corresponding output results are the same as the historical analysis results included in the training data, and counts a second number of verification models whose corresponding output results are different from the historical analysis results included in the training data. When the first number is greater than the second number, the training data is removed from the distinguished first combination. When the first number is less than or equal to the second number, the training data is retained, and the remote analysis module treats all the training data in the distinguished first combination as problematic training data.

[0013] This application also provides a communication optical cable monitoring system for multi-dimensional marine data, including the following modules: A sensor module for monitoring the status of the communication optical cable; A collection module, used for collecting monitoring data of the communication optical cable from different sensor modules; A communication module, used to transmit the monitoring data of the communication optical cable collected by the acquisition module from different sensor modules to the remote analysis module; The remote analysis module is used to obtain a number of training data, the training data includes historical monitoring data of the communication optical cable, and historical analysis results corresponding to the historical monitoring data of the communication optical cable, and is also used to use all the obtained training data to train the analysis model, obtain the number of core training data in all the obtained training data, obtain the analysis error rate based on the trained analysis model, and check whether a first suspension condition is met according to the number of core training data obtained this time. If yes, suspend the training process for the analysis model. If no, continue to check whether a second suspension condition is met according to the analysis error rate obtained this time, if no, obtain a number of training data and continue the training process. If yes, suspend the training process for the analysis model, and also perform verification processing on all the obtained training data to determine a number of problematic training data in all the obtained training data, use all the correct training data in all the obtained training data to train the analysis model, and also generate feature data based on the monitoring data of the communication optical cable, input the feature data into the trained analysis model, and the analysis model outputs an analysis result on the state of the communication optical cable.

[0014] Compared with the prior art, the beneficial effects of the present application are at least as follows: In the technical solution provided in the present application, first, a number of training data are obtained. Then, all the obtained training data are used to train the analysis model, and the number of core training data in all the obtained training data is obtained, and the analysis error rate is obtained based on the trained analysis model. Secondly, check whether the first suspension condition is met according to the number of core training data obtained this time. If yes, suspend the training process for the analysis model. If not, continue to check whether the second suspension condition is met according to the analysis error rate obtained this time. If no, obtain a number of training data and continue the training process. If yes, suspend the training process for the analysis model. Thirdly, check all the obtained training data to determine a number of problem training data in all the obtained training data, and use all the correct training data in all the obtained training data to train the analysis model. Finally, collect the monitoring data of the communication optical cable, generate feature data based on the monitoring data of the communication optical cable, input the feature data into the trained analysis model, and the analysis model outputs the analysis result about the state of the communication optical cable. Through this application, not only can the appropriate time to suspend the training process be determined, but also the problematic training data can be determined from all the training data obtained, and all the correct training data can be used for training processing. Thus, this application can improve the analysis accuracy of the analysis model and ensure the reliability of the analysis results on the status of the communication optical cable. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0016] Figure 1 This is a flow chart of a method for monitoring a communication optical cable for multi-dimensional ocean data in an embodiment of the present application; Figure 2 Schematic diagram of each second combination generated in the embodiments of the present application; Figure 3 Part of the steps of the verification process performed in the embodiments of the present application; Figure 4 Schematic diagram of a communication optical cable monitoring system for multi-dimensional ocean data in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The present application embodiment provides a method and system for monitoring communication optical cables for marine multi-dimensional data. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0018] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 The communication optical cable monitoring method for ocean multi-dimensional data in the embodiment of the present application includes the following main steps: S1. The remote analysis module obtains a number of training data, where the training data includes historical monitoring data of the communication optical cable and historical analysis results corresponding to the historical monitoring data of the communication optical cable; S2. The remote analysis module uses all the acquired training data to perform training processing on the analysis model, obtains the number of core training data in all the acquired training data, and obtains the analysis error rate based on the trained analysis model; S3, the remote analysis module checks whether the first suspension condition is met according to the number of core training data obtained this time, and if so, the remote analysis module suspends the training process for the analysis model, and if not, the remote analysis module continues to check whether the second suspension condition is met according to the analysis error rate obtained this time, and if not, the remote analysis module continues to obtain a number of training data and jumps to S2, and if so, the remote analysis module suspends the training process for the analysis model; S4. The remote analysis module verifies all the acquired training data to determine a number of problematic training data among all the acquired training data, and uses all the correct training data among all the acquired training data to perform training processing on the analysis model; S5. The acquisition module collects the monitoring data of the communication optical cable from different sensor modules, and the acquisition module transmits the monitoring data of the communication optical cable to the remote analysis module through the communication module. The remote analysis module generates feature data based on the monitoring data of the communication optical cable, inputs the feature data into the trained analysis model, and the analysis model outputs the analysis result about the status of the communication optical cable.

[0019] Specifically, in S1, the remote analysis module obtains a number of training data, wherein the training data includes the historical monitoring data of the communication optical cable, and the historical analysis results corresponding to the historical monitoring data of the communication optical cable. It should be noted that the historical analysis results indicate the state of the communication optical cable, which can be generated manually and may be wrong. In S2, the remote analysis module uses all the obtained training data to train the analysis model, which can be a support vector machine. The number of core training data in all the obtained training data is counted. The core training data refers to the training data of the separation surface involved in generating the analysis model. The core training data can be determined by the existing technology, so it will not be repeated. The analysis error rate is also calculated based on the trained analysis model, and the calculation method will be described below. In S3, the remote analysis module checks whether the first pause condition is met based on the number of core training data obtained this time. If the first pause condition is met, the training process for the analysis model is paused. The first pause condition will be described below. If the first pause condition is not met, the remote analysis module continues to check whether the second pause condition is met based on the analysis error rate obtained this time. The second pause condition will be described below. If the second pause condition is not met, continue to obtain several training data and jump to S2 to continue execution. If the second pause condition is met, the training process for the analysis model is paused. In S4, the remote analysis module verifies all the acquired training data. The verification process will be described below. The purpose is to determine several problematic training data among all the acquired training data. Problematic training data refers to training data including historical analysis results that are erroneous. If problematic training data is used for training, it will have an adverse effect on the analysis accuracy of the analysis model. The analysis accuracy corresponds to the performance of the analysis model when processing different feature data in the future. All correct training data except several problematic training data among all the acquired training data are used to train the analysis model. It should be noted that the analysis model here refers to the analysis model when the second pause condition is met, so as to obtain a trained analysis model, and the analysis accuracy of the trained analysis model is improved. In S5, the acquisition module collects monitoring data of the communication optical cable from different sensor modules. For example, the monitoring data of the communication optical cable may include multi-dimensional data such as the ambient temperature around the submarine optical cable and the vibration around the submarine optical cable. The monitoring data of the communication optical cable is transmitted to the remote analysis module through the communication module. The remote analysis module generates feature data based on the monitoring data of the communication optical cable, inputs the feature data into a trained analysis model, and the analysis model outputs analysis results on the status of the communication optical cable.

[0020] Furthermore, the process of the remote analysis module obtaining the analysis error rate based on the trained analysis model includes: obtaining different test data, the test data including historical monitoring data of the communication optical cable, and actual analysis results corresponding to the historical monitoring data of the communication optical cable, respectively inputting the historical monitoring data of the communication optical cable included in all the test data into the trained analysis model, respectively obtaining the output results of the analysis model, and counting the number of output results that are different from the actual analysis results included in the corresponding test data, and calculating the ratio to the number of all the test data.

[0021] Specifically, the process of obtaining the analysis error rate based on the trained analysis model by the remote analysis module is introduced. The first step is to obtain different test data, wherein the test data includes the historical monitoring data of the communication optical cable and the real analysis results corresponding to the historical monitoring data of the communication optical cable. The historical monitoring data of the communication optical cable included in all the test data are respectively input into the trained analysis model to obtain the output results of the trained analysis model. The second step is to check whether each output result is the same as the real analysis result included in the corresponding test data, and then count the number of output results that are different from the real analysis result included in the corresponding test data, and calculate the ratio of the number to the number of all the test data. The above method can verify the training effect of the analysis model.

[0022] Furthermore, the process of the remote analysis module checking whether the first pause condition is met includes: calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than a preset difference threshold, it is considered that the first pause condition is met, or calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than the preset difference threshold, continuing to check whether the number of consecutive occurrences of the situation in which the difference is less than the preset difference threshold in history exceeds a preset consecutive occurrence threshold, and if so, it is considered that the first pause condition is met.

[0023] Specifically, the process of the remote analysis module checking whether the first suspension condition is met is introduced. The first process is to calculate the difference between the number of core training data obtained this time and the number of core training data obtained last time. If the difference is less than the preset difference threshold, the difference threshold is set according to the actual application scenario, and it is considered that the first suspension condition is met. The second process is to calculate the difference between the number of core training data obtained this time and the number of core training data obtained last time. If the difference is less than the preset difference threshold, continue to check whether the number of consecutive occurrences of this situation in history exceeds the preset consecutive occurrence threshold. The occurrence threshold is set according to the actual application scenario. If it exceeds, it is considered that the first suspension condition is met. In this embodiment, it can be understood that when the training process starts, the number of core training data is insufficient. As the training process is continuously performed, the growth rate of the number of core training data is relatively fast until the correct above-mentioned separation surface can be used to generate the analysis model. After the training process is performed again, the growth rate of the number of core training data begins to decline, or even stops increasing. At this time, even if the training data continues to be added, it will not have much beneficial effect on the analysis accuracy of the analysis model, and the training process should be suspended.

[0024] Furthermore, the process of the remote analysis module checking whether the second pause condition is met includes: checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, it is considered that the second pause condition is met, or checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, continuing to check whether the number of consecutive occurrences of the situation in history where the analysis error rate obtained this time is greater than the analysis error rate obtained last time exceeds a preset consecutive occurrence threshold. If so, it is considered that the second pause condition is met.

[0025] Specifically, the process of the remote analysis module checking whether the second suspension condition is met is introduced. The first process is to check whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, it is considered that the second suspension condition is met. The second process is to check whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, continue to check whether the number of consecutive occurrences of this situation in history exceeds a preset consecutive occurrence threshold. If so, it is considered that the second suspension condition is met. In this embodiment, it can be understood that since the historical analysis results of all the acquired training data can be generated manually, when the training process is started, the number of problematic training data in all the acquired training data is still small, so that the analysis error rate is relatively low. As multiple training data are acquired for multiple training processes, because the number of problematic training data in all the acquired training data continues to increase, the analysis error rate also continues to increase. At this time, even if the training data continues to be increased, it will not have much beneficial effect on the analysis accuracy of the analysis model, and the training process should be suspended.

[0026] Furthermore, the remote analysis module verifies and processes all the acquired training data, including the following steps: S411, the remote analysis module divides all the acquired training data into a preset number of first groups, so that the number of different training data included in each first group is the same; S412, the remote analysis module sets a preset number of second combinations, assigns values ​​from small to large to each second combination in sequence, and determines each second combination in sequence according to the corresponding values ​​from small to large, selects a first combination from all the first combinations for the first determined second combination, adds the selected first combination to the first determined second combination, selects another first combination from all the other first combinations that have not been selected for each second combination determined thereafter, and adds the selected other first combination together with the content of the previous second combination to the second combination; S413: For each second combination, the remote analysis module generates a verification model by training using different training data included in the second combination.

[0027] Specifically, the process of the remote analysis module verifying and processing all the acquired training data is introduced. In S411, the remote analysis module divides all the acquired training data into a preset number of first combinations, and the number of different training data included in each first combination is the same, and the training data included in each first combination are all different. In S412, the remote analysis module sets a preset number of second combinations. Initially, the content of each second combination is empty. A value is assigned to each second combination in sequence from small to large. For example, 5 second combinations are assigned values ​​from 1 to 5 in sequence. The remote analysis module determines each second combination in sequence according to the corresponding values ​​from small to large. For example, the second combination with a corresponding value of 1 is determined, the second combination with a corresponding value of 2 is determined, and so on, until the second combination with a corresponding value of 5 is determined. For the first determined second combination, the remote analysis module selects a first combination from all the first combinations, which can be selected arbitrarily, and adds the selected first combination to the first determined second combination. For each second combination determined thereafter, the remote analysis module selects another first combination from all the other first combinations that have not been selected, and adds the selected other first combinations together with the content of the previous second combination to the second combination. For ease of understanding, for example, after the execution of S412 is completed, the remote analysis module obtains each second combination such as Figure 2 As shown, B1 to B5 are respectively the second combination with the corresponding value of 1 to the second combination with the corresponding value of 5, and A1 to A5 are 5 different first combinations. In S413, for each second combination, the remote analysis module uses the different training data included in the second combination to train and generate a verification model. Each verification model and the analysis model have the same problem to be solved, that is, both can generate output results about the state of the communication optical cable according to the input historical monitoring data of the communication optical cable, but the two are not the same model.

[0028] Furthermore, the remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S421, the remote analysis module obtains a verification data, the verification data includes historical monitoring data of the communication optical cable, and the real analysis result corresponding to the historical monitoring data of the communication optical cable, and for each verification model, the remote analysis module inputs the historical monitoring data of the communication optical cable included in the verification data into the verification model to obtain the output result of the verification model; S422. For each verification model, the remote analysis module checks whether the output result of the verification model is consistent with the real analysis result corresponding to the historical monitoring data of the communication optical cable included in the verification data, and the remote analysis module classifies the second combination corresponding to the verification model with the corresponding inspection result being consistent into the first category, and classifies the second combination corresponding to the verification model with the corresponding inspection result being inconsistent into the second category; S423. The remote analysis module determines the second combination in the first category that contains the largest number of all first combinations, determines the second combination in the second category that contains the smallest number of all first combinations, and obtains the difference first combination between the determined second combination in the first category and the determined second combination in the second category.

[0029] Specifically, refer to Figure 3 As shown, the process of the remote analysis module verifying all the acquired training data is further introduced. In S421, the remote analysis module obtains a verification data, wherein the verification data includes the historical monitoring data of the communication optical cable and the actual analysis results corresponding to the historical monitoring data of the communication optical cable. For each verification model, the remote analysis module inputs the historical monitoring data of the communication optical cable included in the verification data into the verification model to obtain the output result of the verification model. In S422, for each verification model, the remote analysis module checks whether the output result of the verification model is the same as the actual analysis result corresponding to the historical monitoring data of the communication optical cable included in the verification data. The remote analysis module classifies the second combination corresponding to the verification model with consistent inspection results into the first category, and classifies the second combination corresponding to the verification model with inconsistent inspection results into the second category, respectively. Figure 2 In the generation method of each second combination shown, there is only one different first combination between two adjacent second combinations. For example, the second combination B3 has one more first combination A3 than the second combination B2. Then, when there is no problem training data in the second combination B2, but there is problem training data in the first combination A3, the output result of the corresponding verification model is likely to be different from the real analysis result corresponding to the historical monitoring data of the communication optical cable included in the verification data. For example, the second combination B1, the second combination B2, and the second combination B3 belong to the first category, and the second combination B4 and the second combination B5 belong to the second category. In S423, the remote analysis module determines the second combination with the largest number of all the first combinations in the first category, and determines the second combination with the smallest number of all the first combinations in the second category, and obtains the difference first combination between the second combination in the determined first category and the second combination in the determined second category. Following the above example, the difference first combination between the second combination B3 and the second combination B4 is A4, and the training data in the difference first combination is preliminarily identified as problem training data.

[0030] Furthermore, the remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S431, the remote analysis module determines the verification models corresponding to the respective second combinations in the first category, and for each determined verification model, inputs the historical monitoring data of the communication optical cable included in the different training data distinguishing the first combination into the verification model, and obtains the output results of the verification model respectively; S432. With respect to each training data in the distinguished first combination, the remote analysis module counts a first number of verification models whose corresponding output results are the same as the historical analysis results included in the training data, and counts a second number of verification models whose corresponding output results are different from the historical analysis results included in the training data. When the first number is greater than the second number, the training data is removed from the distinguished first combination. When the first number is less than or equal to the second number, the training data is retained, and the remote analysis module treats all the training data in the distinguished first combination as problematic training data.

[0031] Specifically, the process of the remote analysis module performing verification processing on all the acquired training data is further introduced. In S431, the remote analysis module determines the verification models corresponding to each second combination in the first category respectively. For each determined verification model, the historical monitoring data of the communication optical cable included in the different training data in the first combination is respectively input into the verification model, and the output results of the verification model are respectively obtained. Since each second combination in the first category does not contain the problem training data, it is considered that the output results of the verification models corresponding to each second combination in the first category are correct. In S432, for each training data in the first combination, the remote analysis module counts the first number of verification models whose corresponding output results are the same as the historical analysis results included in the training data, and counts the second number of verification models whose corresponding output results are different from the historical analysis results included in the training data. If the first number is greater than the second number, the training data is removed from the first combination, that is, the training data is not the problem training data. If the first number is less than or equal to the second number, the training data is retained, that is, the training data is the problem training data. Then, the remote analysis module treats all the training data in the first combination as the problem training data.

[0032] According to another aspect of the embodiment of the present application, refer to Figure 4 As shown, the present application also provides a communication optical cable monitoring system for ocean multi-dimensional data, including a sensor module, a collection module, a communication module, and a remote analysis module to implement the communication optical cable monitoring method for ocean multi-dimensional data described above.

[0033] The functions of each module are as follows: A sensor module for monitoring the status of the communication optical cable; A collection module, used for collecting monitoring data of the communication optical cable from different sensor modules; A communication module, used to transmit the monitoring data of the communication optical cable collected by the acquisition module from different sensor modules to the remote analysis module; The remote analysis module is used to obtain a number of training data, the training data includes historical monitoring data of the communication optical cable, and historical analysis results corresponding to the historical monitoring data of the communication optical cable, and is also used to use all the obtained training data to train the analysis model, obtain the number of core training data in all the obtained training data, obtain the analysis error rate based on the trained analysis model, and check whether a first suspension condition is met according to the number of core training data obtained this time. If yes, suspend the training process for the analysis model. If no, continue to check whether a second suspension condition is met according to the analysis error rate obtained this time, if no, obtain a number of training data and continue the training process. If yes, suspend the training process for the analysis model, and also perform verification processing on all the obtained training data to determine a number of problematic training data in all the obtained training data, use all the correct training data in all the obtained training data to train the analysis model, and also generate feature data based on the monitoring data of the communication optical cable, input the feature data into the trained analysis model, and the analysis model outputs an analysis result on the state of the communication optical cable.

[0034] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0035] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

[0036] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A communication optical cable monitoring method for multi-dimensional marine data, characterized in that: The method comprises the following steps: S1. The remote analysis module obtains a number of training data, where the training data includes historical monitoring data of the communication optical cable and historical analysis results corresponding to the historical monitoring data of the communication optical cable; S2, the remote analysis module uses all the acquired training data to perform training processing on the analysis model, obtains the number of core training data in all the acquired training data, and obtains the analysis error rate based on the trained analysis model; S3, the remote analysis module checks whether the first suspension condition is met according to the number of core training data obtained this time, and if so, the remote analysis module suspends the training process for the analysis model, and if not, the remote analysis module continues to check whether the second suspension condition is met according to the analysis error rate obtained this time, and if not, the remote analysis module continues to obtain a number of training data and jumps to S2, and if so, the remote analysis module suspends the training process for the analysis model; S4, the remote analysis module checks all the acquired training data to determine a number of problematic training data among all the acquired training data, and uses all the correct training data among all the acquired training data to perform training processing on the analysis model; S5. The acquisition module collects the monitoring data of the communication optical cable from different sensor modules, and the acquisition module transmits the monitoring data of the communication optical cable to the remote analysis module through the communication module. The remote analysis module generates feature data based on the monitoring data of the communication optical cable, and inputs the feature data into the trained analysis model. The analysis model outputs the analysis result on the status of the communication optical cable.

2. The method according to claim 1, characterized in that The process of the remote analysis module obtaining the analysis error rate based on the trained analysis model includes: obtaining different test data, the test data including historical monitoring data of the communication optical cable, and the actual analysis results corresponding to the historical monitoring data of the communication optical cable, respectively inputting the historical monitoring data of the communication optical cable included in all the test data into the trained analysis model, respectively obtaining the output results of the analysis model, and counting the number of output results that are different from the actual analysis results included in the corresponding test data, and calculating the ratio to the number of all the test data.

3. The method according to claim 1, characterized in that The process of the remote analysis module checking whether the first suspension condition is met includes: calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than a preset difference threshold, it is considered that the first suspension condition is met, or calculating the difference between the number of core training data obtained this time and the number of core training data obtained last time, and if the difference is less than the preset difference threshold, continuing to check whether the number of consecutive occurrences of the situation in which the difference is less than the preset difference threshold in history exceeds the preset consecutive occurrence threshold, and if so, it is considered that the first suspension condition is met.

4. The method according to claim 1, characterized in that: The process of the remote analysis module checking whether the second suspension condition is met includes: checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, it is considered that the second suspension condition is met, or checking whether the analysis error rate obtained this time is greater than the analysis error rate obtained last time. If so, continue to check whether the number of consecutive occurrences of the situation in history where the analysis error rate obtained this time is greater than the analysis error rate obtained last time exceeds a preset consecutive occurrence threshold. If so, it is considered that the second suspension condition is met.

5. The method according to claim 1, characterized in that The remote analysis module verifies and processes all the acquired training data, including the following steps: S411, the remote analysis module divides all the acquired training data into a preset number of first groups, so that the number of different training data included in each first group is the same; S412, the remote analysis module sets a preset number of second combinations, assigns values ​​from small to large to each second combination in sequence, and determines each second combination in sequence according to the corresponding values ​​from small to large, selects a first combination from all the first combinations for the first determined second combination, adds the selected first combination to the first determined second combination, selects another first combination from all the other first combinations that have not been selected for each second combination determined thereafter, and adds the selected other first combination together with the content of the previous second combination to the second combination; S413: For each second combination, the remote analysis module generates a verification model by training using different training data included in the second combination.

6. The method according to claim 5, characterized in that The remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S421, the remote analysis module obtains a verification data, the verification data includes historical monitoring data of the communication optical cable, and a real analysis result corresponding to the historical monitoring data of the communication optical cable, and for each verification model, the remote analysis module inputs the historical monitoring data of the communication optical cable included in the verification data into the verification model to obtain an output result of the verification model; S422, for each verification model, the remote analysis module checks whether the output result of the verification model is consistent with the real analysis result corresponding to the historical monitoring data of the communication optical cable included in the verification data, and the remote analysis module respectively classifies the second combination corresponding to the verification model with the corresponding inspection result being consistent into the first category, and respectively classifies the second combination corresponding to the verification model with the corresponding inspection result being inconsistent into the second category; S423. The remote analysis module determines the second combination in the first category that contains the largest number of all first combinations, determines the second combination in the second category that contains the smallest number of all first combinations, and obtains the difference first combination between the determined second combination in the first category and the determined second combination in the second category.

7. The method according to claim 6, characterized in that The remote analysis module verifies and processes all the acquired training data, and further includes the following steps: S431, the remote analysis module determines the verification models corresponding to the respective second combinations in the first category, and for each determined verification model, inputs the historical monitoring data of the communication optical cable included in the different training data distinguishing the first combination into the verification model, and obtains the output results of the verification model respectively; S432. With respect to each training data in the distinguished first combination, the remote analysis module counts a first number of verification models whose corresponding output results are the same as the historical analysis results included in the training data, and counts a second number of verification models whose corresponding output results are different from the historical analysis results included in the training data. When the first number is greater than the second number, the training data is removed from the distinguished first combination. When the first number is less than or equal to the second number, the training data is retained, and the remote analysis module treats all the training data in the distinguished first combination as problematic training data.

8. A communication optical cable monitoring system for marine multi-dimensional data, used to implement the method according to any one of claims 1 to 7, characterized in that: Includes the following modules: A sensor module for monitoring the status of the communication optical cable; A collection module, used for collecting monitoring data of the communication optical cable from different sensor modules; A communication module, used to transmit the monitoring data of the communication optical cable collected by the acquisition module from different sensor modules to the remote analysis module; The remote analysis module is used to obtain a number of training data, the training data includes historical monitoring data of the communication optical cable, and historical analysis results corresponding to the historical monitoring data of the communication optical cable, and is also used to use all the obtained training data to train the analysis model, obtain the number of core training data in all the obtained training data, obtain the analysis error rate based on the trained analysis model, and check whether a first suspension condition is met according to the number of core training data obtained this time. If yes, suspend the training process for the analysis model. If no, continue to check whether a second suspension condition is met according to the analysis error rate obtained this time, if no, obtain a number of training data and continue the training process. If yes, suspend the training process for the analysis model, and also perform verification processing on all the obtained training data to determine a number of problematic training data in all the obtained training data, use all the correct training data in all the obtained training data to train the analysis model, and also generate feature data based on the monitoring data of the communication optical cable, input the feature data into the trained analysis model, and the analysis model outputs an analysis result on the state of the communication optical cable.

Images