A vibration event recognition method, device and computer storage medium

Abstract

The application relates to the technical field of information recognition, in particular to a vibration event recognition method and device and a computer storage medium. The recognition method comprises the following steps: acquiring a vibration detection matrix corresponding to a current time window; determining channel features corresponding to a plurality of recognition channels in the current time window; determining a candidate channel set corresponding to the current time window; taking a next time window of the current time window as the current time window; repeatedly executing the steps of acquiring the vibration detection matrix corresponding to the current time window until taking the next time window of the current time window as the current time window; until the current time window is the last time window in a target time period; determining an effective channel set corresponding to the target time period; determining the type of a vibration event; and performing screening processing on the data input for recognition before the vibration event is recognized, so that the data processing amount for recognition is reduced, the vibration event recognition efficiency is improved, and the recognition cost is reduced.

Description

Technical Field

[0001] This application relates to the field of information recognition technology, and in particular to a vibration event recognition method, device, and computer storage medium. Background Technology

[0002] Fiber optic sensors are sensors that use light as the carrier of sensitive information and optical fiber as the transmission medium for sensitive information. Compared with other types of sensors, fiber optic sensors have advantages such as good electrical insulation, strong resistance to electromagnetic interference, corrosion resistance, explosion-proof, and easy long-distance detection. Fiber optic sensors are commonly used in security monitoring, submarine optical cables, pipeline leak monitoring, and intrusion event monitoring.

[0003] In the existing technology, when using long-distance fiber optic sensors, it is often necessary to process a large amount of data in order to achieve accurate positioning and analysis of vibration events. This results in greater complexity, longer recognition time, and higher recognition costs for vibration event identification. Summary of the Invention

[0004] In view of the above-mentioned problems in the prior art, the purpose of this application is to filter and process the input data for vibration event identification before identification, thereby reducing the amount of identification data to be processed, improving the efficiency of vibration event processing, and reducing identification costs.

[0005] To address the aforementioned problems, this application provides a vibration event identification method, comprising:

[0006] Obtain the vibration detection matrix corresponding to the current time window; the vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber;

[0007] Based on the vibration detection matrix corresponding to the current time window, channel features are extracted to obtain the channel features corresponding to each of the multiple identification channels under the current time window.

[0008] Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the candidate channel set of the previous time window, the candidate channel set corresponding to the current time window is determined.

[0009] The next time window of the current time window is taken as the current time window;

[0010] Repeat the following steps: obtain the vibration detection matrix corresponding to the current time window, and then use the next time window of the current time window as the current time window; until the current time window is the last time window in the target time period;

[0011] The candidate channel set corresponding to the current time window is determined as the effective channel set corresponding to the target time period;

[0012] Based on the vibration detection matrix corresponding to the target time period of the effective channel set, vibration events are identified and the type of vibration event is determined.

[0013] In this embodiment of the application, determining the candidate channel set corresponding to the current time window based on the channel features corresponding to each of the plurality of identification channels under the current time window and the candidate channel set of the previous time window includes:

[0014] Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the first threshold, channel filtering is performed to obtain a first candidate set;

[0015] Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the second threshold, channel filtering is performed to obtain a second candidate set; the second threshold is greater than the first threshold.

[0016] The intersection of the first candidate set and the candidate channel set of the previous time window is performed to determine the set of interconnected channels;

[0017] Based on the set of traffic channels in the phase and the second candidate set, the set of candidate channels corresponding to the current time window is determined; the identified channels in the set of candidate channels corresponding to the current window exist in the set of traffic channels in the phase or the second candidate set.

[0018] In this embodiment of the application, the vibration event identification method further includes:

[0019] Obtain sample channel features corresponding to multiple sample recognition channels;

[0020] Based on the sample channel features corresponding to the multiple sample recognition channels, feature value statistics are performed to obtain the maximum and minimum feature values ​​corresponding to the sample channel features;

[0021] Numerical range expansion processing is performed based on the minimum and maximum feature values ​​to obtain a first threshold corresponding to the minimum feature value and a second threshold corresponding to the maximum feature value; the first threshold is less than the minimum feature value and the second threshold is greater than the maximum feature value.

[0022] In this embodiment of the application, the channel features include multiple feature values, the first threshold includes first threshold values ​​corresponding to each of the multiple feature values, and the second threshold includes second threshold values ​​corresponding to each of the multiple feature values; the step of channel filtering based on the channel features corresponding to each of the multiple identification channels under the current time window and the first threshold to determine the first candidate set includes:

[0023] If all the feature values ​​corresponding to any identification channel are greater than or equal to the first threshold value corresponding to each of the feature values, then the identification channel is determined as the first candidate channel.

[0024] The first candidate set is generated based on the first candidate channel;

[0025] The step of determining the second candidate set by filtering channels based on the channel features corresponding to each of the multiple identification channels under the current time window and the second threshold includes:

[0026] If all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the second threshold value corresponding to each of the multiple feature values, then the identification channel is determined as the second candidate channel.

[0027] The second candidate set is generated based on the second candidate channel.

[0028] In another embodiment of this application, the channel features include feature values ​​corresponding to multiple feature labels, the first threshold includes first threshold values ​​corresponding to each of the multiple feature values, and the second threshold includes second threshold values ​​corresponding to each of the multiple feature values; the step of channel filtering based on the channel features corresponding to each of the multiple recognition channels under the current time window and the first threshold to determine the first candidate set includes:

[0029] Traverse the various feature labels;

[0030] Determine the feature values ​​corresponding to each of the multiple recognition channels under the first current feature label;

[0031] Determine the first label candidate channel set corresponding to the first current feature label; the feature value of the recognition channel in the first label candidate channel set under the first current feature label is greater than or equal to the first threshold value corresponding to the first current feature label.

[0032] The first candidate set is determined based on the first candidate channel set corresponding to each of the multiple feature labels.

[0033] In another embodiment of this application, the step of determining the second candidate set by filtering channels based on the channel features corresponding to each of the plurality of identification channels under the current time window and the second threshold includes:

[0034] Traverse the various feature labels;

[0035] Determine the feature values ​​corresponding to each of the multiple recognition channels under the second current feature label;

[0036] Determine the second label candidate channel set corresponding to the second current feature label; the feature value of the recognition channel in the second label candidate channel set under the second current feature label is greater than or equal to the second threshold value corresponding to the second current feature label.

[0037] The second candidate set is determined based on the second candidate channel set corresponding to each of the multiple feature labels.

[0038] In another embodiment of this application, the first candidate set includes a first label candidate channel set corresponding to each of the multiple feature labels; the second candidate set includes a second label candidate channel set corresponding to each of the multiple feature labels; the candidate channel set of the previous time window includes a third label candidate channel set corresponding to each of the multiple feature labels; the step of finding the intersection of the first candidate set and the candidate channel set of the previous time window to determine the corresponding channel set includes:

[0039] The intersection of the first tag candidate channel set under the same feature label and the third tag candidate channel set under the same feature label is performed to obtain the fourth tag candidate channel set corresponding to the same feature label.

[0040] Based on the fourth label candidate channel set corresponding to each of the multiple feature labels, the corresponding channel set is determined;

[0041] The step of determining the candidate channel set corresponding to the current time window based on the set of corresponding traffic channels and the second candidate set includes:

[0042] The fourth label candidate channel set and the second label candidate channel set under the same feature label are respectively subjected to union processing to obtain the fifth label candidate channel set corresponding to the same label;

[0043] The number of each candidate channel in the fifth label candidate channel set corresponding to each of the various feature labels is counted to obtain the statistical number of each candidate channel;

[0044] Candidate channels whose statistical count is greater than or equal to a preset number are identified as target channels;

[0045] A set of candidate channels corresponding to the current time window is generated based on the target channel.

[0046] In this embodiment of the application, the vibration event identification method further includes:

[0047] If the previous time window does not exist in the current time window, the candidate channel set of the previous time window is determined to be an empty set.

[0048] In this embodiment of the application, the step of identifying vibration events and determining the type of vibration event based on the vibration detection matrix corresponding to the target time period using the effective channel set includes:

[0049] Amplitude filtering is performed on the amplitude waves corresponding to each of the multiple valid identification channels in the set of valid channels to obtain the filtered signals corresponding to each of the multiple valid identification channels.

[0050] Feature extraction is performed on the filtered signals corresponding to each of the multiple effective recognition channels to obtain multiple feature data corresponding to each of the multiple effective recognition channels;

[0051] Based on the multiple feature data corresponding to each of the multiple effective recognition channels, normalized data combination is performed to obtain the input vector corresponding to each of the multiple effective recognition channels;

[0052] The input vectors corresponding to each of the multiple effective identification channels are input into the vibration event identification model to identify the event type, thereby obtaining the vibration event type corresponding to each of the multiple effective identification channels.

[0053] On the other hand, this application also provides a vibration event identification device, the device comprising:

[0054] The matrix acquisition module is used to acquire the vibration detection matrix corresponding to the current time window; the vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber.

[0055] The channel feature extraction module is used to extract channel features based on the vibration detection matrix corresponding to the current time window, so as to obtain the channel features corresponding to each of the multiple identification channels under the current time window.

[0056] The candidate channel set determination module is used to determine the candidate channel set corresponding to the current time window based on the channel features corresponding to each of the multiple identification channels under the current time window and the candidate channel set of the previous time window.

[0057] The current time window determination module is used to take the next time window of the current time window as the current time window;

[0058] The loop module is used to repeatedly execute the following steps: obtain the vibration detection matrix corresponding to the current time window, and then use the next time window of the current time window as the current time window; until the current time window is the last time window in the target time period.

[0059] The effective channel set determination module is used to determine the candidate channel set corresponding to the current time window as the effective channel set corresponding to the target time period;

[0060] The event identification module is used to identify vibration events and determine the type of vibration event based on the vibration detection matrix corresponding to the target time period of the effective channel set.

[0061] On the other hand, this application also provides an electronic device, the device including a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the vibration event recognition method described above.

[0062] On the other hand, this application also provides a computer storage medium storing at least one instruction or at least one program, wherein the at least one instruction or the at least one program is loaded and executed by a processor to implement the vibration event recognition method described above.

[0063] Due to the above technical solution, the vibration event identification method described in this application has the following beneficial effects:

[0064] The vibration event recognition method in this application extracts the effective channels within the target time period based on the vibration detection matrix corresponding to the time window, thereby reducing the amount of input data for event recognition, which in turn reduces the amount of recognition data processing, improves the efficiency of vibration event recognition, and reduces the cost of event recognition. Furthermore, by using a time window, the vibration amplitude information and time correlation information within the target time period are segmented for processing, thereby reducing the complexity of data processing and improving the speed of data processing. Attached Figure Description

[0065] To more clearly illustrate the technical solutions of this application, the accompanying drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0066] Figure 1 This is a schematic flowchart of a vibration event identification method provided in an embodiment of this application;

[0067] Figure 2 This is a schematic diagram of the process for obtaining the candidate channel set in a vibration event identification method provided in this application embodiment;

[0068] Figure 3 This is a schematic diagram of the threshold acquisition process in a vibration event identification method provided in an embodiment of this application;

[0069] Figure 4 This is a schematic diagram illustrating the generation process of the first candidate set and the second candidate set in a vibration event identification method provided in this application embodiment;

[0070] Figure 5 This is the complete process framework corresponding to the vibration event identification method provided in the embodiments of this application. Figure 1 ;

[0071] Figure 6 This is a schematic diagram of the process for generating the first candidate set in a vibration event identification method provided in this application embodiment;

[0072] Figure 7 This is a schematic diagram of the second candidate set generation process in a vibration event identification method provided in this application embodiment;

[0073] Figure 8 This is a schematic diagram of the process for obtaining the candidate channel set in a vibration event identification method provided in this application embodiment;

[0074] Figure 9 This is the complete process framework corresponding to the vibration event identification method provided in the embodiments of this application. Figure 2 ;

[0075] Figure 10 This is a schematic flowchart of the event identification process in a vibration event identification method provided in an embodiment of this application;

[0076] Figure 11 This is a schematic diagram of the structure of a vibration event identification device provided in an embodiment of this application;

[0077] Figure 12 This is a hardware structure block diagram of a vibration event recognition method provided in an embodiment of this application. Detailed Implementation

[0078] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0079] The term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of this application. In the description of this application, it should be understood that the terms "upper," "lower," "left," "right," "top," "bottom," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. Moreover, the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein.

[0080] Combination Figure 1 This application introduces a vibration event identification method provided by an embodiment of the present application. The method includes:

[0081] S1001. Obtain the vibration detection matrix corresponding to the current time window; the vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber; the current time window is the time window being processed, which can be a historical time window or a real-time time window; the identification channel refers to the smallest optical fiber segment that can define the location of the event; specifically, the multiple identification channels of the target optical fiber are determined based on the length of the target optical fiber and the spatial resolution of the target optical fiber. For example, if the length of the target optical fiber is 40km and the spatial resolution of the target optical fiber is 10m, then there are 4000 multiple identification channels.

[0082] In the embodiments of this application, the target optical fiber can be a distributed optical fiber sensor or other optical fiber sensors based on Rayleigh scattering; specifically, it can be a distributed optical fiber sensor based on Φ-OTDR (phase-sensitive optical time domain reflectometer) or a distributed optical fiber sensor based on C-OTDR (coherent optical time domain reflectometer).

[0083] S1002. Based on the vibration detection matrix corresponding to the current time window, channel features are extracted to obtain the channel features corresponding to each of the multiple identification channels under the current time window. Channel feature extraction refers to extracting features from the element columns in the vibration detection matrix. Specifically, channel features include one or more features in the time domain or frequency domain.

[0084] S1003. Based on the channel features corresponding to each of the multiple recognition channels under the current time window and the candidate channel set of the previous time window, channel filtering is performed to determine the candidate channel set corresponding to the current time window; channel filtering refers to filtering channels that meet preset conditions among multiple recognition channels; for example, the preset conditions may be that the channel features are greater than a preset threshold and / or are in the candidate channel set of the previous time window, or they may be less than a preset threshold and / or are in the candidate channel set of the previous time window; the candidate channel set may include at least one recognition channel, or it may be an empty set that does not include any recognition channels.

[0085] S1004. Take the next time window of the current time window as the current time window; specifically, the time window slides as time changes, and the time length of the time window is greater than or equal to the sliding step size.

[0086] In this embodiment, the duration of the time window is less than or equal to twice the sliding step size. For example, if the duration of the time window is 600ms, then the sliding step size needs to be greater than or equal to 300ms. By making the duration of the time window less than or equal to twice the sliding step size, the overlapping area of ​​adjacent time windows is ensured to be greater than or equal to half the duration of the time window, thereby preserving the stable part of the signal to the greatest extent and enabling the time window to cover the target time period, thus ensuring the comprehensiveness of feature extraction and improving the channel filtering accuracy.

[0087] S1005. Repeat the following steps: Obtain the vibration detection matrix corresponding to the current time window, and use the next time window as the current time window; until the current time window is the last time window within the target time period; the target time period is longer than the length of each time window, and the sliding step size can be adaptively changed according to the target time period so that the target time period can be completely covered by multiple time windows; for example, if the target time period is 3s, then the length of the time window can be 600ms, and the sliding step size can be 300ms, so that the time window can completely cover the target time period; for another example, if the target time period is 1000ms, then if the length of the time window is 600ms, the sliding step size is 400ms, so that two time windows can completely cover the target time period; the target time period can be adaptively modified based on the actual application scenario.

[0088] In another embodiment of this application, the total length of time corresponding to multiple time windows can be slightly greater than the length of the target time. For example, if the length of a time window is 600ms, the sliding step is 300ms, and the target time period is 1000ms, then the target time period can be divided into three time windows: 0-600ms, 300-900ms, and 900ms-1200ms. The last time window within the target time period is the time window corresponding to 900-1200ms. That is, if an integer number of time windows cannot be obtained for the target time period, the length of the target time period is adaptively increased to update the target time period.

[0089] S1006. Determine the set of candidate channels corresponding to the current time window as the set of valid channels corresponding to the target time period; the set of valid channels includes valid channels, which refer to the vibrations that occur at the beginning of the event within the target time period and continue to vibrate.

[0090] S1007. Based on the vibration detection matrix corresponding to the effective channel set in the target time period, vibration event identification is performed to determine the type of vibration event. Vibration event identification refers to analyzing the input vibration detection matrix to obtain the vibration event type corresponding to the input data. Vibration event types include natural vibration events and man-made vibration events. For example, natural vibration events refer to vibration events caused by wind and rain, while man-made vibration events refer to vibration events caused by actions such as scratching, kicking, leaning, patting, knocking, and climbing.

[0091] In this embodiment of the application, the vibration event identification method has the following beneficial effects:

[0092] The vibration event recognition method in this application extracts the effective channels within the target time period based on the vibration detection matrix corresponding to the time window, thereby reducing the amount of input data for event recognition, which in turn reduces the amount of recognition data processing, improves the efficiency of vibration event recognition, and reduces the cost of event recognition. Furthermore, by using a time window, the vibration amplitude information and time correlation information within the target time period are segmented for processing, thereby reducing the complexity of data processing and improving the speed of data processing.

[0093] In this embodiment of the application, before S1001, the vibration event identification method further includes:

[0094] Acquire fiber vibration data from the target fiber; fiber vibration data refers to the light intensity signal of the target fiber after interference based on Rayleigh scattering light.

[0095] Based on preset decomposition conditions, the fiber vibration data is decomposed to obtain the vibration amplitude information corresponding to each of the multiple identification channels. The preset decomposition conditions refer to the conditions for decoupling the fiber vibration data. Specifically, the preset decomposition conditions can be to output a light intensity signal with high signal-to-noise ratio and low interference fading.

[0096] Based on a preset storage time, the vibration amplitude information corresponding to each of the multiple recognition channels is stored in the data storage module, which can be a data storage disk. The preset storage time refers to the time for storing the vibration amplitude information.

[0097] In this embodiment of the application, the preset storage time is much longer than the data acquisition time of the target optical fiber. For example, if the data acquisition time of the target optical fiber is 1ms, then the preset storage time can be 100ms.

[0098] In this embodiment, the size of the vibration detection matrix is ​​related to the length of the time window, the number of multiple identification channels, and the data acquisition time of the target optical fiber. For example, if the length of the time window is 600ms, the number of multiple identification channels is 4000, and the data acquisition time of the target optical fiber is 1ms, then the size of the vibration detection matrix is ​​600×4000. The columns of the vibration detection matrix represent the relationship between the vibration amplitude information of any identification channel and time, and the rows of the vibration detection matrix represent the change of vibration amplitude information with distance at any acquisition time node. The distance is expressed by the channel number.

[0099] refer to Figure 2 In this embodiment of the application, S1003 includes:

[0100] S2001. Based on the channel features corresponding to each of the multiple recognition channels under the current time window and the first threshold, channel filtering is performed to obtain a first candidate set; preferably, the channel features corresponding to the recognition channels in the first candidate set are greater than the first threshold.

[0101] S2002. Based on the channel features corresponding to each of the multiple recognition channels under the current time window and the second threshold, channel filtering is performed to obtain a second candidate set; the second threshold is greater than the first threshold; preferably, the channel features corresponding to the recognition channels in the second candidate set are greater than the second threshold.

[0102] S2003. Perform intersection processing on the first candidate set and the candidate channel set of the previous time window to determine the set of interconnected channels.

[0103] S2004. Based on the set of phased traffic channels and the second candidate set, determine the candidate channel set corresponding to the current time window; the identified channels in the candidate channel set corresponding to the current window exist in the set of phased traffic channels or the second candidate set; specifically, the set of phased traffic channels and the second candidate set can be combined to obtain the candidate channel set corresponding to the current time window; or the combined channel set can be filtered to obtain the candidate channel set corresponding to the current time window.

[0104] In the embodiments of this application, if any set in the first candidate set or the candidate channel set of the previous time window is empty, the corresponding channel set is empty.

[0105] In this embodiment, multiple identification channels are filtered using a first threshold and a second threshold to obtain a first candidate set and a second candidate set. The candidate channel set corresponding to the current time window is determined based on the first candidate set, the second candidate set, and the candidate channel set of the previous time window. The candidate channel set corresponding to the current time window is filtered using thresholds, thereby improving the completeness of signal extraction. For example, after an event occurs, the identification channel will continue to vibrate for a certain period of time after acquiring fiber optic vibration data.

[0106] In this embodiment, when the channel feature corresponding to the identification channel passes the first threshold, the time window in which the identification channel is located is considered to be the time window where the event is occurring; when the channel feature corresponding to the identification channel passes the second threshold, the event window in which the identification channel is located is considered to be the initial stage of the event. By performing intersection processing on the first candidate set and the candidate channel set of the previous time window, the continuity of the event is ensured. By determining the candidate channel set corresponding to the current window based on the corresponding channel set and the second candidate set, it is ensured that the identification channels entering the initial stage of the event can be screened out, that is, the screening accuracy and screening completeness of the candidate channel set corresponding to the current window are improved.

[0107] refer to Figure 3 In this embodiment of the application, the vibration event identification method further includes:

[0108] S3001. Obtain sample channel features corresponding to multiple sample recognition channels; a sample recognition channel refers to a channel in the event testing process, for example, tapping the recognition channel to obtain the sample channel features corresponding to the tapping; multiple sample recognition channels can be 1000 sample recognition channels or 4000 sample recognition channels; sample channel features include the start features of the event and the end features of the event; preferably, the sample channel features can be preliminarily screened, for example, invalid signals, interfered signals, and signals with poor quality can be excluded.

[0109] S3002. Based on the sample channel features corresponding to multiple sample recognition channels, perform feature value statistics to obtain the maximum and minimum feature values ​​corresponding to the sample channel features; preferably, feature statistics can be performed on the sample features after preliminary screening; feature statistics can refer to sorting the sample channel features to obtain the maximum and minimum feature values ​​corresponding to the sample channel features.

[0110] S3003. Based on the minimum and maximum feature values, a numerical range expansion process is performed to obtain a first threshold corresponding to the minimum feature value and a second threshold corresponding to the maximum feature value. The first threshold is less than the minimum feature value, and the second threshold is greater than the maximum feature value. Specifically, the minimum feature value is multiplied by a first preset adjustment coefficient to obtain the first threshold, which is greater than zero and less than one. The maximum feature value is multiplied by a second preset adjustment coefficient to obtain the second threshold, which is greater than one and less than the ratio of the maximum feature value to the minimum feature value. Both the first and second preset adjustment coefficients can be adjusted based on the actual scenario and events.

[0111] In this embodiment of the application, the first threshold and the second threshold can be calculated in advance.

[0112] In a specific embodiment of this application, for example, the sample channel feature set corresponding to multiple sample recognition channels is δ = {S1, S2, ... S...} m Then, the first threshold can be calculated using the following formula:

[0113] X1 = K × min(S1, S2, ..., S) m (1)

[0114] Where X1 is the first threshold, K is the preset first adjustment coefficient, and 0 <K<1,min(S1,S2,……S m ) represents the eigenvalue minimum, i.e., δ = {S1, S2, ..., S} m The minimum value in}.

[0115] The second threshold can be calculated using the following formula:

[0116] X2 = Q × max(S1, S2, ..., S) m (2)

[0117] Where X2 is the second threshold, Q is the preset second adjustment coefficient, and max(S1, S2, ..., S) m ) represents the eigenvalue, i.e., δ = {S1, S2, ..., S} m The maximum value in}.

[0118] In this embodiment, by expanding the range of the maximum and minimum values ​​of features in the sample recognition channel, lowering the first threshold and raising the second threshold, the screening requirements can be reduced, thereby covering as many channels as possible that meet the requirements and improving the accuracy of vibration event recognition.

[0119] In this embodiment, the channel features include multiple feature values. The first threshold includes the first threshold value corresponding to each of the multiple feature values, and the second threshold includes the second threshold value corresponding to each of the multiple feature values. Specifically, the channel features include time-domain features and frequency-domain features. Time-domain features include short-time energy, absolute mean, root square amplitude, root mean square amplitude, standard deviation, average amplitude difference, zero-crossing rate, threshold crossing rate, kurtosis, skewness, peak factor, etc. Frequency-domain features include frequency variance, frequency standard deviation, mean square frequency, root mean square frequency, etc. The multiple feature values ​​can be any number of the aforementioned feature values. Specifically, the types of multiple feature values ​​can be determined based on the type of event.

[0120] refer to Figure 4 In this embodiment of the application, S2001 includes:

[0121] S4001. If all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the first threshold value corresponding to each of the multiple feature values, then any identification channel is determined as the first candidate channel.

[0122] S4002. Generate a first candidate set based on the first candidate channel;

[0123] S2002 includes:

[0124] S4003. If all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the second threshold values ​​corresponding to each of the multiple feature values, then any identification channel is determined as the second candidate channel.

[0125] S4004. Generate a second candidate set based on the second candidate channel.

[0126] In this embodiment, if multiple feature values ​​corresponding to the identification channel all pass the first threshold and the second threshold, the identification channel is determined as a candidate channel, thereby improving the screening accuracy of the candidate channel, reducing the data processing complexity in the vibration event identification process, and improving the data processing speed.

[0127] In a specific embodiment of this application, the extraction formulas for some channel feature values ​​are as follows:

[0128] Short-time energy feature extraction formula:

[0129]

[0130] Where E refers to the short-time energy corresponding to the m-th identification channel within any time window; n refers to the ratio of the time window length to the data acquisition time, which is the number of rows in the vibration detection matrix; s i It refers to the vibration amplitude information corresponding to the m-th identification channel and the i-th row, where 0≤i≤n.

[0131] Peak factor extraction formula:

[0132]

[0133] Among them, PEAK m This refers to the peak factor of the m-th recognition channel, s i It refers to the vibration amplitude information corresponding to the m-th identification channel and the i-th row, where 0≤i≤n.

[0134] Pulse factor extraction formula:

[0135]

[0136] Among them, IMPULSE m This refers to the pulse factor of the m-th recognition channel, si It refers to the vibration amplitude information corresponding to the m-th identification channel and the i-th row, where 0≤i≤n.

[0137] Margin factor extraction formula:

[0138]

[0139] Among them, MAGIN m This refers to the margin factor of the m-th recognition channel, s i It refers to the vibration amplitude information corresponding to the m-th identification channel and the i-th row, where 0≤i≤n.

[0140] In this embodiment, the extraction formula for channel features is stored in a preset configuration file. When a certain feature value needs to be calculated, the extraction formula corresponding to a certain feature value is extracted from the preset configuration file.

[0141] In this embodiment of the application, before extracting frequency domain features, it is necessary to perform a fast Fourier transform on the data under the same identification channel. Before performing the fast Fourier transform, a window function can be used to process the data, so that the data better meets the requirements of the fast Fourier transform.

[0142] In a specific embodiment of this application, multi-threading can be used to extract features from multiple feature values ​​corresponding to each of the multiple recognition channels, thereby improving the feature extraction rate and thus improving the vibration time recognition rate.

[0143] For details, please refer to the appendix. Figure 5 :

[0144] Obtain multiple feature values ​​corresponding to each of the multiple recognition channels, as well as the candidate channel set corresponding to the previous time window; select multiple recognition channels corresponding to any type of feature value; determine the first candidate set and the second candidate set corresponding to any feature value based on the first threshold value and the second threshold value corresponding to any type of feature value; determine whether the first candidate set and the second candidate set are empty sets. Specifically, if the first candidate set and the second candidate set are empty sets, output that the candidate channel set of the current time window is an empty set.

[0145] If the first candidate set and the second candidate set are not empty sets, determine whether it is the first type of feature value, that is, whether any feature value is processed for the first time. Specifically: if any feature value is not processed for the first time, the first candidate set and the second candidate set are respectively subjected to corresponding intersection processing to update the first candidate set and the second candidate set. That is, the first candidate set is subjected to intersection processing with the first candidate set after the previous feature value screening to update the first candidate set, and the second candidate set is subjected to intersection processing with the second candidate set after the previous feature value screening to update the second candidate set.

[0146] If any feature value has not been processed for the first time, or if all feature values ​​have been traversed, cache and output the first candidate set and the second candidate set; determine whether the candidate channel set corresponding to the previous time window is empty; if the candidate channel set corresponding to the previous time window is empty, determine the second candidate set as the candidate channel set corresponding to the current time window, and output and cache it.

[0147] If the candidate channel set corresponding to the previous time window is not empty, the first candidate set is intersected with the candidate channel set corresponding to the previous time window to determine the corresponding channel set; the corresponding channel set is unioned with the second candidate set to obtain the candidate channel set corresponding to the current time window, which is then output and cached.

[0148] In this embodiment of the application, by adopting Figure 5 The vibration event recognition method described above can improve the feature extraction rate and thus the vibration time recognition rate by making multiple judgments.

[0149] In a specific embodiment of this application, it is assumed that there are three time windows and three recognition channels within the target time period: channel 1, channel 2, and channel 3; the first threshold includes a first threshold value a corresponding to the first feature and a second threshold b corresponding to the first feature, wherein a is less than b; the second threshold includes a first threshold value c corresponding to the first feature and a second threshold value d corresponding to the second feature, wherein c is less than d.

[0150] In the first time window, the feature values ​​of the first feature corresponding to each of the three recognition channels are all greater than 'a', and the feature values ​​of the second feature corresponding to each of the three recognition channels are all greater than 'c', so the first candidate set is (Channel 1, Channel 2, Channel 3); the feature values ​​of the first feature corresponding to each of the three recognition channels are all greater than 'b', and the feature values ​​of the second feature corresponding to each of Channel 1 and Channel 3 are all greater than 'd', so the second candidate set is (Channel 1, Channel 3); since there are no other time windows before the first time window, the output second candidate set is the candidate channel set corresponding to the first time window, that is, the candidate channel set corresponding to the first time window is (Channel 1, Channel 3).

[0151] In the second time window, the judgment rules are the same as in the first time window: in the first candidate set, the multiple feature values ​​corresponding to each recognition channel are all greater than the first threshold value corresponding to each of the multiple feature values; in the second candidate set, the multiple feature values ​​corresponding to each recognition channel are all greater than the second threshold value corresponding to each of the multiple feature values; specifically, the feature values ​​of the first feature corresponding to each channel 1 and channel 2 are all greater than 'a', and the feature values ​​of the second feature corresponding to each channel 1 and channel 2 are all greater than 'c'; the feature value of the first feature corresponding to channel 2 is greater than 'b', and the feature value of the second feature corresponding to channel 2 is greater than 'd', thus obtaining the first candidate set as (channel 1, channel 2); the second candidate set as (channel 2); then the candidate channel set corresponding to the second time window is: the intersection of the first candidate set and the candidate channel set corresponding to the first time window, and then the union of the first candidate set and the second candidate set, that is, the candidate channel set corresponding to the second time window is (channel 1, channel 2).

[0152] In the third time window, the judgment rules are the same as those in the first time window. Specifically, the feature values ​​of the first feature corresponding to each of channels 2 and 3 are all greater than a and less than b, and the feature values ​​of the second feature corresponding to each of channels 2 and 3 are all greater than c and less than d, resulting in the first candidate set being (channel 2, channel 3); the second candidate set is an empty set; then the candidate channel set corresponding to the third time window is the intersection of the first candidate set and the candidate channel set corresponding to the second time window, and then the union of the first candidate set and the second candidate set, that is, the candidate channel set corresponding to the second time window is (channel 2).

[0153] Therefore, the set of valid channels within the target time period is (Channel 2).

[0154] Reference Appendix Figure 6 In another embodiment of this application, the channel features include feature values ​​corresponding to multiple feature labels, the first threshold includes first threshold values ​​corresponding to each of the multiple feature values, and the second threshold includes second threshold values ​​corresponding to each of the multiple feature values.

[0155] S2001 includes:

[0156] S6001, Traverse multiple feature labels; feature labels can be numbers or special symbols, and feature labels correspond one-to-one with feature values.

[0157] S6002. Determine the feature values ​​corresponding to each of the multiple recognition channels under the first current feature label;

[0158] S6003. Determine the first label candidate channel set corresponding to the first current feature label; the feature value of the recognition channel in the first label candidate channel set under the first current feature label is greater than or equal to the first threshold value corresponding to the first current feature label.

[0159] S6004. Determine the first candidate set based on the first candidate channel set corresponding to each of the multiple feature labels.

[0160] refer to Figure 7 S2002 includes:

[0161] S7001, Traverse multiple feature labels;

[0162] S7002. Determine the feature values ​​corresponding to each of the multiple recognition channels under the second current feature label;

[0163] S7003. Determine the set of candidate channels for the second label corresponding to the second current feature label; the feature value of the recognition channel in the set of candidate channels for the second label under the second current feature label is greater than or equal to the second threshold value corresponding to the second current feature label.

[0164] S7004. Determine the second candidate set based on the second label candidate channel set corresponding to each of the multiple feature labels.

[0165] In another embodiment of this application, by performing channel filtering on multiple recognition channels corresponding to the current feature label, it is possible to traverse all the multiple feature values ​​corresponding to each of the multiple recognition channels, thereby improving the filtering accuracy.

[0166] refer to Figure 8 In another embodiment of this application, the first candidate set includes a first label candidate channel set corresponding to each of the multiple feature labels, the second candidate set includes a second label candidate channel set corresponding to each of the multiple feature labels, and the candidate channel set of the previous time window includes a third label candidate channel set corresponding to each of the multiple feature labels.

[0167] S2003 includes:

[0168] S8001. Perform intersection processing on the first label candidate channel set and the third label candidate channel set under the same feature label respectively to obtain the fourth label candidate channel set corresponding to the same feature label.

[0169] S8002. Based on the candidate channel set of the fourth label corresponding to each of the multiple feature labels, determine the set of related channels.

[0170] S2004 includes:

[0171] S8003. Perform union processing on the fourth label candidate channel set and the second label candidate channel set under the same feature label respectively to obtain the fifth label candidate channel set corresponding to the same label.

[0172] S8004. Count the number of each candidate channel in the candidate channel set of the fifth label corresponding to each of the multiple feature labels, and obtain the count of each candidate channel.

[0173] S8005. Select candidate channels whose statistical count is greater than or equal to a preset number as target channels;

[0174] S8006. Generate a set of candidate channels corresponding to the current time window based on the target channel.

[0175] In another embodiment of this application, by counting the number of candidate channels, if the number of candidate channels is greater than or equal to a preset number, the number of feature values ​​of the candidate channel that pass the dual threshold is greater than the preset number, thereby reducing the channel selection criteria and making the channels that meet the requirements as comprehensive as possible, thus improving the accuracy of vibration event recognition.

[0176] In another embodiment of this application, if the number of each candidate channel is less than a preset number, the candidate channel set corresponding to the current time window is determined to be an empty set.

[0177] In another specific embodiment of this application, reference is made to Figure 9 :

[0178] Obtain multiple recognition channels corresponding to various feature values, as well as the candidate channel set corresponding to the previous time window; initialize the voting values ​​corresponding to each of the multiple recognition channels, i.e., set them to 0; select multiple recognition channels corresponding to any feature value, and obtain the candidate channel set corresponding to any feature value in the previous time window; based on the first threshold value and the second threshold value corresponding to any feature value, determine the first candidate set and the second candidate set corresponding to any feature value respectively; determine whether the first candidate set and the second candidate set are empty sets. Specifically, if the first candidate set and the second candidate set corresponding to any feature value are empty sets, determine that the candidate channel set corresponding to any feature value is an empty set.

[0179] If the first and second candidate sets are not empty, determine whether the candidate channel set corresponding to any feature value in the previous time window is empty. Specifically, if the candidate channel set corresponding to any feature value in the previous time window is empty, the candidate channel set corresponding to any feature value in the current time window is the corresponding second candidate set, and output and cache it; increment the voting value of the recognition channel in the candidate channel set corresponding to any feature value by one.

[0180] In the previous time window, if the candidate channel set corresponding to any feature value is not empty, the first candidate set corresponding to any feature value is intersected with the candidate channel set corresponding to any feature value in the previous time window to determine the corresponding channel set corresponding to any feature value; the corresponding channel set is unioned with the second candidate channel set corresponding to any feature value to determine and cache the candidate channel set corresponding to any feature value in the current time window; the voting value corresponding to the recognition channel in the candidate channel set corresponding to any feature value is incremented by one.

[0181] After the voting calculation, it is determined whether there are other unprocessed feature values. Specifically: if there are no other unprocessed feature values, the candidate channel sets corresponding to each of the multiple feature values ​​are cached, and the recognition channels with voting values ​​greater than or equal to a preset threshold are selected to form the target channel set; otherwise, if there are other unprocessed feature values, the next feature value is processed.

[0182] Given the target channel set, determine whether the target channel set is empty. Specifically, if the target channel set is not empty, determine the target channel set as the candidate channel set corresponding to the current time window; otherwise, if the target channel set is empty, determine the candidate channel set corresponding to the current time window as empty.

[0183] In another embodiment of this application, the screening rate and the vibration event identification rate are improved by making multiple judgments and by using a voting system to achieve data statistics.

[0184] In a specific embodiment of this application, it is assumed that there are three time windows and three recognition channels within the target time period: channel 1, channel 2, and channel 3; the first threshold includes a first threshold value a corresponding to the first feature and a second threshold b corresponding to the first feature, wherein a is less than b; the second threshold includes a first threshold value c corresponding to the first feature and a second threshold value d corresponding to the second feature, wherein c is less than d.

[0185] In the first time window, among the three recognition channels corresponding to the first feature, the feature values ​​corresponding to each of the three recognition channels are all greater than 'a' and 'b'; therefore, the first candidate set corresponding to the first feature is (Channel 1, Channel 2, Channel 3); the second candidate set corresponding to the first feature is (Channel 1, Channel 2, Channel 3); the candidate channel set corresponding to the first feature is the second candidate set, i.e., (Channel 1, Channel 2, Channel 3); among the three recognition channels corresponding to the second feature, the feature values ​​corresponding to Channel 1 and Channel 2 are greater than 'c', and the feature value corresponding to Channel 1 is greater than 'd'; therefore, the first candidate set corresponding to the second feature is (Channel 1, Channel 2); the second candidate set corresponding to the second feature is (Channel 1); the candidate channel set corresponding to the second feature value is the second candidate set, i.e., (Channel 1); where the voting value of Channel 1 is 2, the voting value of Channel 2 is 1, and the voting value of Channel 3 is 1. If the preset voting value is 2, then the candidate channel set corresponding to the first time window is (Channel 1).

[0186] In the second time window, the judgment rules are the same as in the first time window: in the first candidate channel set corresponding to any feature, the feature value corresponding to the identification channel is greater than the first threshold value corresponding to any feature; in the second candidate channel set corresponding to any feature, the feature value corresponding to the identification channel is greater than the second threshold value corresponding to any feature; the first candidate set corresponding to the first feature is (channel 1, channel 3); the second candidate set corresponding to the first feature is (channel 1); then the candidate channel set corresponding to the first feature is the intersection of the first candidate set corresponding to the first feature and the candidate channel set corresponding to the first feature, and then the union of the second candidate set corresponding to the first feature, i.e., (channel 1, channel 3); the first candidate set corresponding to the second feature is (channel 1, channel 2), the second candidate set corresponding to the second feature is (channel 1), and similarly, the candidate channel set corresponding to the second feature value is (channel 1); among them, the voting value of channel 1 is 2, the voting value of channel 2 is 0, and the voting value of channel 3 is 1, then the candidate channel set corresponding to the second time window is (channel 1).

[0187] In the third time window, the judgment rules are the same as in the first time window. The first candidate set corresponding to the first feature is (channel 3); the second candidate set corresponding to the first feature is (channel 3); then the candidate channel set corresponding to the first feature is (channel 3); the first candidate set corresponding to the second feature is (channel 3); the second candidate set corresponding to the second feature is (channel 3); the candidate channel set corresponding to the second feature value is (channel 3), where the voting value of channel 1 is 0, the voting value of channel 2 is 0, and the voting value of channel 3 is 2. Therefore, the candidate channel set corresponding to the third time window is channel 3.

[0188] Therefore, the set of valid channels within the target time period is (Channel 3).

[0189] In this embodiment of the application, the vibration event identification method further includes:

[0190] If there is no previous time window in the current time window, the candidate channel set of the previous time window is determined to be an empty set; if the candidate channel set of the previous time window is an empty set, the second candidate set can be directly determined as the candidate channel set corresponding to the current time window.

[0191] In this embodiment, the determination of the first candidate set, the second candidate set, and the candidate channel set of the previous time window can be added. If the first candidate set or the candidate channel set of the previous time window is empty, the second candidate set is determined to be the candidate channel set corresponding to the current time window. If the first candidate set or the candidate channel set of the previous time window is empty and the second candidate set is empty, the candidate channel set corresponding to the current time window is determined to be empty.

[0192] In this embodiment, it can be determined whether the first candidate set, the second candidate set, and the candidate channel set corresponding to the previous event window are empty sets, thereby avoiding excessive filtering operations, improving the channel filtering rate, and improving the vibration event recognition rate.

[0193] In another embodiment of this application, when the first candidate set or the candidate channel set corresponding to the previous time window is an empty set, the number of each candidate channel in the second candidate set is counted.

[0194] refer to Figure 10 In this embodiment of the application, S1007 includes:

[0195] S10071. Amplitude filtering is performed on the amplitude waves corresponding to each of the multiple valid identification channels in the set of valid channels to obtain the filtered signals corresponding to each of the multiple valid identification channels; amplitude wave refers to the waveform formed by the change of vibration amplitude information over time; amplitude filtering can specifically be low-pass filtering, for example, filtering each valid identification channel through a Butterworth filter to remove the high-frequency noise components present.

[0196] S10072. Extract features from the filtered signals corresponding to each of the multiple effective recognition channels to obtain various feature data corresponding to each of the multiple effective recognition channels; the feature data includes time domain features, frequency domain features, and wavelet threshold features.

[0197] S10073. Normalize and combine the various feature data corresponding to each of the multiple effective recognition channels to obtain the input vectors corresponding to each of the multiple effective recognition channels; specifically, after normalizing the various feature data corresponding to each of the multiple effective channels, sort them in a preset order to obtain the input vectors corresponding to each of the multiple effective recognition channels.

[0198] S10074. Input the input vectors corresponding to each of the multiple valid recognition channels into the vibration event recognition model to identify the event type, thereby obtaining the vibration event type corresponding to each of the multiple valid recognition channels; specifically, the vibration event recognition model is a model obtained by intelligent recognition training based on the sample input vectors and the corresponding vibration event types.

[0199] In this embodiment of the application, by filtering effective recognition channels for recognition, the input data for the vibration event recognition model is reduced, the training difficulty of the vibration event recognition model is reduced, and the accuracy of the recognition results of the vibration event recognition model is improved.

[0200] In this embodiment, the acquisition and storage of fiber optic vibration data, the extraction of effective channel sets, and the identification of vibration events are independent of each other, thereby enabling multi-process parallel computation and improving the vibration event identification rate.

[0201] The vibration event identification method in this application has the following beneficial effects:

[0202] The vibration event recognition method in this application extracts the effective channels within the target time period based on the vibration detection matrix corresponding to the time window, thereby reducing the amount of input data for event recognition, which in turn reduces the amount of recognition data processing, improves the efficiency of vibration event recognition, and reduces the cost of event recognition. Furthermore, by using a time window, the vibration amplitude information and time correlation information within the target time period are segmented for processing, thereby reducing the complexity of data processing and improving the speed of data processing.

[0203] refer to Figure 11 This application also provides a vibration event identification device, which includes:

[0204] The matrix acquisition module 101 is used to acquire the vibration detection matrix corresponding to the current time window; the vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber.

[0205] The channel feature extraction module 102 is used to extract channel features based on the vibration detection matrix corresponding to the current time window, and obtain the channel features corresponding to each of the multiple identification channels under the current time window.

[0206] The candidate channel set determination module 103 is used to determine the candidate channel set corresponding to the current time window based on the channel features corresponding to each of the multiple recognition channels under the current time window and the candidate channel set of the previous time window.

[0207] The current time window determination module 104 is used to take the next time window of the current time window as the current time window;

[0208] The loop module 105 is used to repeatedly execute the following steps: obtain the vibration detection matrix corresponding to the current time window, and then take the next time window of the current time window as the current time window; until the current time window is the last time window in the target time period.

[0209] The effective channel set determination module 106 is used to determine the candidate channel set corresponding to the current time window as the effective channel set corresponding to the target time period;

[0210] The event recognition module 107 is used to identify vibration events and determine the type of vibration event based on the vibration detection matrix corresponding to the effective channel set in the target time period.

[0211] The candidate channel set determination module includes:

[0212] The first candidate unit is used to filter channels based on the channel features corresponding to each of the multiple recognition channels under the current time window and the first threshold to obtain the first candidate set;

[0213] The second candidate unit is used to filter channels based on the channel features corresponding to each of the multiple recognition channels under the current time window and the second threshold to obtain a second candidate set; the second threshold is greater than the first threshold.

[0214] The intersection processing unit is used to perform intersection processing on the first candidate set and the candidate channel set of the previous time window to determine the intersection channel set;

[0215] The candidate channel set determination unit is used to determine the candidate channel set corresponding to the current time window based on the phase traffic channel set and the second candidate set; the identified channels in the candidate channel set corresponding to the current window exist in the phase traffic channel set or the second candidate set.

[0216] The vibration event identification device also includes:

[0217] The sample acquisition module is used to acquire sample channel features corresponding to multiple sample recognition channels;

[0218] The feature statistics module is used to perform feature value statistics based on the sample channel features corresponding to multiple sample recognition channels, and to obtain the maximum and minimum feature values ​​corresponding to the sample channel features.

[0219] The range expansion module is used to expand the numerical range based on the feature minimum and feature maximum values ​​to obtain a first threshold corresponding to the feature minimum and a second threshold corresponding to the feature maximum; the first threshold is less than the feature minimum and the second threshold is greater than the feature maximum.

[0220] The first candidate units include:

[0221] The first channel determination unit is used to determine any identification channel as the first candidate channel when all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the first threshold values ​​corresponding to each of the multiple feature values.

[0222] The first candidate set determination unit is used to generate a first candidate set based on the first candidate channel;

[0223] The first traversal unit is used to traverse multiple feature labels;

[0224] The first determining unit is used to determine the feature values ​​corresponding to each of the multiple recognition channels under the first current feature label;

[0225] The first label candidate determination unit is used to determine the first label candidate channel set corresponding to the first current feature label; the feature value of the recognition channel in the first label candidate channel set under the first current feature label is greater than or equal to the first threshold value corresponding to the first current feature label.

[0226] The third candidate set determination unit is used to determine the first candidate set based on the first label candidate channel set corresponding to each of the multiple feature labels.

[0227] The second candidate units include:

[0228] The second channel determination unit is used to determine any identification channel as a second candidate channel when all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the second threshold values ​​corresponding to each of the multiple feature values.

[0229] The second candidate set determination unit is used to generate a second candidate set based on the second candidate channel;

[0230] The second traversal unit is used to traverse multiple feature labels;

[0231] The second determining unit is used to determine the feature values ​​corresponding to each of the multiple recognition channels under the second current feature label;

[0232] The second label candidate determination unit is used to determine the second label candidate channel set corresponding to the second current feature label; the feature value of the recognition channel in the second label candidate channel set under the second current feature label is greater than or equal to the second threshold value corresponding to the second current feature label.

[0233] The fourth candidate set determination unit is used to determine the second candidate set based on the second label candidate channel set corresponding to each of the multiple feature labels.

[0234] The intersection processing unit includes:

[0235] The intersection unit is used to find the intersection of the first label candidate channel set and the third label candidate channel set under the same feature label, respectively, to obtain the fourth label candidate channel set corresponding to the same feature label.

[0236] The intersection set determination unit is used to determine the intersection channel set based on the fourth label candidate channel set corresponding to each of the multiple feature labels;

[0237] The candidate channel set determination unit includes:

[0238] The separate union processing unit is used to perform union processing on the fourth label candidate channel set and the second label candidate channel set under the same feature label respectively, to obtain the fifth label candidate channel set corresponding to the same label;

[0239] The quantity statistics unit is used to count the number of each candidate channel in the fifth label candidate channel set corresponding to various feature labels, and obtain the statistical count of each candidate channel.

[0240] The target channel determination unit is used to determine candidate channels whose statistical count is greater than or equal to a preset number as target channels;

[0241] The candidate channel set determination unit is used to generate a candidate channel set corresponding to the current time window based on the target channel.

[0242] The vibration event identification device also includes:

[0243] The empty set determination module is used to determine that the candidate channel set of the previous time window is an empty set when the current time window does not have a previous time window.

[0244] The event recognition module includes:

[0245] The filtering unit is used to perform amplitude filtering on the amplitude waves corresponding to each of the multiple valid identification channels in the set of valid channels, so as to obtain the filtered signals corresponding to each of the multiple valid identification channels.

[0246] The feature extraction unit is used to extract features from the filtered signals corresponding to each of the multiple effective recognition channels to obtain multiple feature data corresponding to each of the multiple effective recognition channels.

[0247] The vector generation unit is used to combine normalized data based on the various feature data corresponding to each of the multiple effective recognition channels to obtain the input vectors corresponding to each of the multiple effective recognition channels.

[0248] The event recognition unit is used to input the input vectors corresponding to each of the multiple valid recognition channels into the vibration event recognition model to identify the event type and obtain the vibration event type corresponding to each of the multiple valid recognition channels.

[0249] This application also provides an electronic device, which includes a processor and a memory. The memory stores at least one instruction or at least one program. The processor loads and executes the instruction or program to implement the vibration event recognition method described above.

[0250] Memory is used to store software programs and modules. The processor executes these stored software programs and modules to perform various functional applications and data processing. Memory can primarily consist of a program storage area and a data storage area. The program storage area stores the operating system, application programs required for functionality, etc.; the data storage area stores data created based on device usage, etc. Furthermore, memory can include high-speed random access memory (RAM) and non-volatile memory, such as at least one hard disk drive, flash memory, or other volatile solid-state storage devices. Correspondingly, memory can also include a memory controller to provide the processor with access to the memory.

[0251] The methods and embodiments provided in this application can be executed in electronic devices such as mobile terminals, computer terminals, servers, or similar computing devices. Figure 12 This is the electronic device provided in the embodiments of this application. For example... Figure 12 As shown, the electronic device 900 can vary significantly due to differences in configuration or performance. It may include one or more central processing units (CPUs) 910 (CPUs 910 may include, but are not limited to, microprocessors such as MCUs or programmable logic devices such as FPGAs), a memory 930 for storing data, and one or more storage media 920 (e.g., one or more mass storage devices) for storing application programs 923 or data 922. The memory 930 and storage media 920 may be temporary or persistent storage. The program stored in the storage media 920 may include one or more modules, each module may include a series of instruction operations on the electronic device. Furthermore, the CPU 910 may be configured to communicate with the storage media 920 and execute the series of instruction operations in the storage media 920 on the electronic device 900. Electronic device 900 may also include one or more power supplies 960, one or more wired or wireless network interfaces 950, one or more input / output interfaces 940, and / or one or more operating systems 921, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0252] The input / output interface 940 can be used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the electronic device 900. In one example, the input / output interface 940 includes a network interface controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the input / output interface 940 may be a radio frequency (RF) module for wireless communication with the Internet.

[0253] Those skilled in the art will understand that Figure 12 The structure shown is for illustrative purposes only and does not limit the structure of the electronic device described above. For example, the electronic device 900 may also include... Figure 12 The more or fewer components shown, or having the same Figure 12 The different configurations shown.

[0254] Embodiments of this application also provide a storage medium storing at least one instruction or at least one program, wherein the at least one instruction or at least one program is loaded and executed by a processor to implement the vibration event recognition method as described above.

[0255] The foregoing description has fully disclosed the specific embodiments of this application. It should be noted that any modifications made by those skilled in the art to the specific embodiments of this application do not depart from the scope of the claims. Accordingly, the scope of the claims of this application is not limited to the foregoing specific embodiments.

Claims

1. A vibration event identification method, characterized in that, include: Obtain the vibration detection matrix corresponding to the current time window; The vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber. Based on the vibration detection matrix corresponding to the current time window, channel features are extracted to obtain the channel features corresponding to each of the multiple identification channels under the current time window. Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the candidate channel set of the previous time window, the candidate channel set corresponding to the current time window is determined. The next time window of the current time window is taken as the current time window; Repeat the following steps: obtain the vibration detection matrix corresponding to the current time window, and then use the next time window of the current time window as the current time window; Until the current time window is the last time window within the target time period; The candidate channel set corresponding to the current time window is determined as the effective channel set corresponding to the target time period; Based on the vibration detection matrix corresponding to the target time period of the effective channel set, vibration events are identified and the type of vibration event is determined.

2. The vibration event identification method according to claim 1, characterized in that, The step of determining the candidate channel set corresponding to the current time window based on the channel features corresponding to each of the multiple recognition channels under the current time window and the candidate channel set of the previous time window includes: Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the first threshold, channel filtering is performed to obtain a first candidate set; Based on the channel features corresponding to each of the multiple identification channels under the current time window, and the second threshold, channel filtering is performed to obtain a second candidate set; the second threshold is greater than the first threshold. The intersection of the first candidate set and the candidate channel set of the previous time window is performed to determine the set of interconnected channels; Based on the set of traffic channels in the phase and the second candidate set, a set of candidate channels corresponding to the current time window is determined; the identified channels in the set of candidate channels corresponding to the current time window exist in the set of traffic channels in the phase or the second candidate set.

3. The vibration event identification method according to claim 2, characterized in that, The method further includes: Obtain sample channel features corresponding to multiple sample recognition channels; Based on the sample channel features corresponding to the multiple sample recognition channels, feature value statistics are performed to obtain the maximum and minimum feature values ​​corresponding to the sample channel features; Numerical range expansion processing is performed based on the minimum and maximum feature values ​​to obtain a first threshold corresponding to the minimum feature value and a second threshold corresponding to the maximum feature value; the first threshold is less than the minimum feature value and the second threshold is greater than the maximum feature value.

4. The vibration event identification method according to claim 2, characterized in that, The channel features include multiple feature values, the first threshold includes first threshold values ​​corresponding to each of the multiple feature values, and the second threshold includes second threshold values ​​corresponding to each of the multiple feature values; the process of channel filtering based on the channel features corresponding to each of the multiple recognition channels under the current time window and the first threshold to determine the first candidate set includes: If all the feature values ​​corresponding to any identification channel are greater than or equal to the first threshold value corresponding to each of the feature values, then the identification channel is determined as the first candidate channel. The first candidate set is generated based on the first candidate channel; The step of determining the second candidate set by filtering channels based on the channel features corresponding to each of the multiple identification channels under the current time window and the second threshold includes: If all the multiple feature values ​​corresponding to any identification channel are greater than or equal to the second threshold value corresponding to each of the multiple feature values, then the identification channel is determined as the second candidate channel. The second candidate set is generated based on the second candidate channel.

5. The vibration event identification method according to claim 2, characterized in that, The channel features include feature values ​​corresponding to multiple feature labels, the first threshold includes first threshold values ​​corresponding to each of the multiple feature values, and the second threshold includes second threshold values ​​corresponding to each of the multiple feature values; the process of channel filtering based on the channel features corresponding to each of the multiple recognition channels under the current time window and the first threshold to determine the first candidate set includes: Traverse the various feature labels; Determine the feature values ​​corresponding to each of the multiple recognition channels under the first current feature label; Determine the first label candidate channel set corresponding to the first current feature label; the feature value of the recognition channel in the first label candidate channel set under the first current feature label is greater than or equal to the first threshold value corresponding to the first current feature label. The first candidate set is determined based on the first candidate channel set corresponding to each of the multiple feature labels.

6. The vibration event identification method according to claim 5, characterized in that, The step of determining the second candidate set by filtering channels based on the channel features corresponding to each of the multiple identification channels under the current time window and the second threshold includes: Traverse the various feature labels; Determine the feature values ​​corresponding to each of the multiple recognition channels under the second current feature label; Determine the second label candidate channel set corresponding to the second current feature label; the feature value of the recognition channel in the second label candidate channel set under the second current feature label is greater than or equal to the second threshold value corresponding to the second current feature label. The second candidate set is determined based on the second candidate channel set corresponding to each of the multiple feature labels.

7. The vibration event identification method according to claim 6, characterized in that, The first candidate set includes the first label candidate channel set corresponding to each of the various feature labels; the second candidate set includes the second label candidate channel set corresponding to each of the various feature labels; the candidate channel set of the previous time window includes the third label candidate channel set corresponding to each of the various feature labels; the step of finding the intersection of the first candidate set and the candidate channel set of the previous time window to determine the corresponding channel set includes: The intersection of the first tag candidate channel set under the same feature label and the third tag candidate channel set under the same feature label is performed to obtain the fourth tag candidate channel set corresponding to the same feature label. Based on the fourth label candidate channel set corresponding to each of the multiple feature labels, the corresponding channel set is determined; The step of determining the candidate channel set corresponding to the current time window based on the set of corresponding traffic channels and the second candidate set includes: The fourth label candidate channel set and the second label candidate channel set under the same feature label are respectively subjected to union processing to obtain the fifth label candidate channel set corresponding to the same feature label; The number of each candidate channel in the fifth label candidate channel set corresponding to each of the various feature labels is counted to obtain the statistical number of each candidate channel; Candidate channels whose statistical count is greater than or equal to a preset number are identified as target channels; A set of candidate channels corresponding to the current time window is generated based on the target channel.

8. The vibration event identification method according to claim 1, characterized in that, The method further includes: If the previous time window does not exist in the current time window, the candidate channel set of the previous time window is determined to be an empty set.

9. The vibration event identification method according to claim 1, characterized in that, The vibration event identification based on the vibration detection matrix corresponding to the target time period using the effective channel set, and the determination of the vibration event type, include: Amplitude filtering is performed on the amplitude waves corresponding to each of the multiple valid identification channels in the set of valid channels to obtain the filtered signals corresponding to each of the multiple valid identification channels. Feature extraction is performed on the filtered signals corresponding to each of the multiple effective recognition channels to obtain multiple feature data corresponding to each of the multiple effective recognition channels; Based on the multiple feature data corresponding to each of the multiple effective recognition channels, normalized data combination is performed to obtain the input vector corresponding to each of the multiple effective recognition channels; The input vectors corresponding to each of the multiple effective identification channels are input into the vibration event identification model to identify the event type, thereby obtaining the vibration event type corresponding to each of the multiple effective identification channels.

10. A vibration event identification device, characterized in that, include: The matrix acquisition module is used to acquire the vibration detection matrix corresponding to the current time window; The vibration detection matrix corresponding to the current time window represents the correlation information between the vibration amplitude information and time of multiple identification channels of the target optical fiber. The channel feature extraction module is used to extract channel features based on the vibration detection matrix corresponding to the current time window, so as to obtain the channel features corresponding to each of the multiple identification channels under the current time window. The candidate channel set determination module is used to determine the candidate channel set corresponding to the current time window based on the channel features corresponding to each of the multiple identification channels under the current time window and the candidate channel set of the previous time window. The current time window determination module is used to take the next time window of the current time window as the current time window; The loop module is used to repeatedly execute the following steps: obtain the vibration detection matrix corresponding to the current time window, and then use the next time window of the current time window as the current time window; Until the current time window is the last time window within the target time period; The effective channel set determination module is used to determine the candidate channel set corresponding to the current time window as the effective channel set corresponding to the target time period; The event identification module is used to identify vibration events and determine the type of vibration event based on the vibration detection matrix corresponding to the target time period of the effective channel set.

11. A computer storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, code set, or instruction set, which is loaded and executed by a processor to implement the vibration event recognition method as described in any one of claims 1-9.

12. An electronic device, characterized in that, The device includes a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded and executed by the processor to implement the vibration event recognition method as described in any one of claims 1-9.

Images