CN111222743B - Method for judging vertical offset distance and threat level of optical fiber sensing event - Google Patents
Method for judging vertical offset distance and threat level of optical fiber sensing event Download PDFInfo
- Publication number
- CN111222743B CN111222743B CN201911125636.8A CN201911125636A CN111222743B CN 111222743 B CN111222743 B CN 111222743B CN 201911125636 A CN201911125636 A CN 201911125636A CN 111222743 B CN111222743 B CN 111222743B
- Authority
- CN
- China
- Prior art keywords
- optical fiber
- event
- vertical offset
- offset distance
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000013307 optical fiber Substances 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000010801 machine learning Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 34
- 238000012360 testing method Methods 0.000 claims description 22
- 238000007637 random forest analysis Methods 0.000 claims description 16
- 238000012706 support-vector machine Methods 0.000 claims description 16
- 238000013145 classification model Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 10
- 238000001228 spectrum Methods 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 8
- 238000003066 decision tree Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 3
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000007635 classification algorithm Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 abstract description 38
- 238000001514 detection method Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000012544 monitoring process Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 239000002609 medium Substances 0.000 description 3
- 239000002689 soil Substances 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 239000004568 cement Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G06Q50/40—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
- G06F2218/14—Classification; Matching by matching peak patterns
Abstract
The invention discloses a method for judging vertical offset distance and threat level of an optical fiber sensing event, which comprises the steps of sensing and collecting vibration signals along a buried optical cable through a distributed optical fiber vibration sensing system, carrying out frequency domain signal processing on longitudinal time signals of each spatial point distributed along the buried optical fiber within an event influence range to construct an energy attenuation characteristic sequence, learning the spatial distribution characteristic difference of energy attenuation under different threat degrees by using a machine learning method, and identifying the vertical offset distance between a vibration source on the ground and the optical cable, thereby judging the threat degree of the vibration source to the optical cable and giving accurate early warning.
Description
Technical Field
The invention belongs to the technical field of buried photoelectric cables and oil and gas pipeline safety monitoring and boundary line security control, and particularly relates to a method for judging vertical offset distance and threat level of an optical fiber sensing event under a complex buried condition.
Background
The distributed optical fiber vibration sensing system based on the phase-sensitive optical time domain reflection technology is widely applied to the fields of long-distance photoelectric cables, oil and gas pipeline safety monitoring, perimeter security and the like. The existing buried optical cable can be used for sensing various vibration and sound sources along the optical fiber, and real-time detection, identification and positioning can be carried out on dangerous events threatening safety such as mechanical construction, artificial excavation and the like, so that the optical fiber can be prevented from getting ill. With the further application of the technology, maintenance and management personnel are more concerned about the threat level of the dangerous vibration source or the vertical offset distance between the dangerous source and the optical cable, and the threat level is judged according to the distance of the vertical offset distance so as to achieve the purpose of precise attendance or police attendance.
However, in the prior art, various vibration sources are detected and identified, and the longitudinal positions of the vibration sources along the optical fiber are positioned, and methods for estimating the vertical offset distance between the vibration sources and the optical fiber and judging the threat level are rarely involved. The Shanghai optical machine utilizes an antenna array signal processing method MUSIC to process acoustic sensing array signals received by optical fibers and estimate spatial position information of sound source signals in air and water media, and because the transmission media are single in air or water, the spatial position of the sound source is easy to determine according to the arrival relationship of the array signals. Samaneh Azadi et al use buried optical cables to collect signals and determine the orientation of the seismic source by multi-point calculation of time delay differences, which is effective for estimating a long-distance vibration source under buried conditions. However, under the complicated buried condition, the soil medium is more complicated than air and water, the optical cable is generally placed in an underground cable trench, vibration and sound source are transmitted to the optical cable and actually pass through multiple layers of mixed soil medium, cement or metal pipe gallery, air and other media, the transmission process is complicated, and the factors influencing signal receiving delay are many and the fluctuation is large; particularly for short-distance (such as within 10 meters) vibration sources, the propagation time of the vibration source reaching the optical fiber is short, the complex underground propagation path brings large time delay fluctuation to influence the accuracy of the offset distance estimation, and great challenges are brought to the vertical offset distance estimation and threat level judgment of the vibration source.
Disclosure of Invention
The invention aims to: the method solves the problems that in the existing method, only various dangerous vibration sources are detected and identified and are positioned along the longitudinal position of an optical fiber, the distance between the vibration source and the optical fiber cannot be judged, and further the threat degree cannot be judged, and the vertical distance of the vibration source from the buried optical fiber is difficult to accurately estimate due to the complex buried condition, a plurality of influence factors of the short-distance vibration source reaching the optical fiber propagation delay and large fluctuation, and the vertical offset distance and the threat level of the optical fiber sensing event are judged.
The technical scheme adopted by the invention is as follows:
a method for judging vertical offset distance and threat level of an optical fiber sensing event comprises the following steps:
collecting longitudinal time signals of each space point distributed along the buried optical fiber to obtain all-line space-time signals distributed along the optical fiber;
determining the central position of a vibration event distributed along a buried optical fiber, selecting a plurality of longitudinal time signals distributed along the optical fiber space to form a signal array according to the vibration event influence range on the left and right of the central position, performing frequency domain signal processing on each space point time signal in the selected area to obtain the energy characteristic of the space point time signal, and splicing the energy characteristic values of different space points according to the spatial sequence to construct an energy attenuation characteristic sequence of the event signal distributed along the optical fiber;
During off-line training, acquiring event signals under different vertical offset distances when a certain vibration event occurs, constructing an event sample data set, calculating to obtain an energy attenuation characteristic sequence of each event signal sample distributed along the optical fiber as a training set, inputting the energy attenuation characteristic sequence into a machine learning model integrated by a plurality of classifiers for training to obtain a classification model of the vertical offset distance between the vibration source and the optical fiber;
during on-line testing, the energy attenuation characteristic sequence distributed along the optical fiber within a certain vibration event signal influence range is input into a constructed and trained classification model, the vertical offset distance between the vibration source and the optical fiber is identified, and then the threat level corresponding to the vertical offset distance is obtained.
Further, as described above, the method for determining the vertical offset distance and threat level of the optical fiber sensing event includes the specific steps of:
s2.1 time domain to frequency domain conversion: respectively carrying out Fourier transform on the time signals of each space point to obtain a power spectrum on a frequency domain;
s2.2, frequency domain filtering: filtering out noise components outside the useful frequency range in the frequency domain;
S2.3, selecting a main frequency component interval and calculating the energy value: after filtering, searching a peak value area in a frequency domain, and calculating a signal energy characteristic value of the frequency band power spectrum of the area near a plurality of peak values;
s2.4, splicing signal energy characteristic values obtained after the signals of each space point are subjected to S2.1-S2.3 processing according to a space distribution sequence, and constructing an energy distribution characteristic vector as an energy attenuation characteristic sequence of the event signals distributed along the optical fiber.
Further, as described above, the method for determining the vertical offset distance and threat level of the optical fiber sensing event includes the specific steps of calculating the signal energy characteristic value of the optical fiber sensing event from the frequency band power spectrum of the area near the plurality of peak values: and calculating the average power value in a frequency interval range near each peak value, and calculating the average value of the obtained average power values near all the peak values as the signal energy characteristic value.
Further, as mentioned above, in the method for judging the vertical offset distance and threat level of the optical fiber sensing event, the machine learning model is a classification model obtained by fusing two models, namely the random forest and the support vector machine, based on the secondary Stacking, 4-fold cross validation is performed by using 4 support vector machine classifiers to output a prediction result, and the result is combined into a new feature to be trained by using the random forest classifier.
Further, according to the method for judging the vertical offset distance and the threat level of the optical fiber sensing event, in the classification model, the support vector machine classification algorithm maps low-dimensional data to a high-dimensional space by using a kernel function, the low-dimensional data are distinguished by hyperplane segmentation, the random forest classification algorithm classifies a plurality of parallel decision trees respectively by constructing the parallel decision trees, and then majority voting is carried out to obtain a final classification result.
Further, as mentioned above, in the method for determining the vertical offset distance and threat level of the optical fiber sensing event, the off-line construction and training process of the two-stage Stacking fusion of the random forest model and the support vector machine model specifically includes:
s6.1, for each model obtained by training the support vector machine, dividing a training set D into k parts, taking each part as a test set, training the model by using the residual data set, and predicting the result of the part;
s6.2, repeating the S6.1 until each model obtains a prediction result, and obtaining a secondary model training set;
s6.3, averaging k test sets to obtain a test set of the secondary model;
s6.4, taking the random forest classifier as a secondary model, and respectively training and testing the secondary model by using the training set and the testing set obtained in S6.2 and S6.3;
And S6.5, after the training of the secondary model in the S6.4 is finished, saving the two-stage model as a vertical offset distance classifier.
Further, according to the method for judging the vertical offset distance and the threat level of the optical fiber sensing event, the distributed optical fiber vibration sensing system is adopted to collect longitudinal time signals of each spatial point distributed along the buried optical fiber in the influence range of the vibration event.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. the method comprises the steps of sensing and collecting vibration signals along the buried optical cable through a distributed optical fiber sensing system, carrying out frequency domain signal processing on longitudinal time signals of all spatial points distributed along the buried optical fiber within an event influence range, constructing an energy attenuation characteristic sequence, learning the spatial distribution characteristic difference of energy attenuation under different threat degrees by using a machine learning method, and identifying the vertical offset distance between a vibration source on the ground and the optical cable, so that the threat degree of the vibration source to the optical cable is judged, and accurate early warning is given. The detection positioning method can be realized under complex burying conditions and propagation paths, namely, a vibration source is propagated to the optical cable through multiple layers of mixed soil media, cement or metal pipe galleries, air and other media. Under the complex buried condition, when the vertical offset distance between the vibration source and the optical fiber is short (such as < 10m), the far-field condition is not satisfied, the positioning error of the existing time delay estimation method is large or is not applicable, and the method is less influenced.
2. According to the method, on the basis of the functions of detecting, identifying and positioning the vibration source, the estimation of the vertical offset distance between the vibration source and the optical cable and the judgment of the threat degree of the optical cable are added, the accuracy and the intelligent level of system alarming can be further improved, and the efficiency of accurate attendance or police attendance is greatly improved.
3. The method is carried out based on the spatial energy attenuation characteristic of the distributed optical fiber sensing system receiving array signals in the influence range of the vibration source, and the problem of inaccurate estimation of relative time delay of signals in a complex buried propagation path or inaccurate positioning caused by other methods can be solved by utilizing the spatial correlation of the distributed optical fiber sensing system along an optical fiber acquisition signal array and estimating the influence degree of a dangerous source on an optical cable in the spatial distribution characteristic. The method overcomes the influences of complex buried conditions and large receiving delay errors, not only positions the dangerous vibration source along the longitudinal position of the optical fiber, but also can judge the vertical distance of the vibration source deviating from the optical fiber under the conditions of uneven buried medium and complex buried environment, and further improves the accuracy and the intelligent level of system alarm.
4. The method adopts a data-driven learning method, learns the mapping relation between the spatial distribution characteristic difference and the distance of the energy attenuation of the vibration source under different vertical offset distances by constructing a sample database and a suitable machine learning network, classifies the spatial distribution characteristic according to the array signal during actual measurement, determines the vertical distance of the dangerous vibration source deviating from the optical cable according to the classification result, and avoids the defects of over dependence on human experience and manual participation.
5. The method has strong timeliness, avoids the problem of searching and calculating time redundancy in the traditional MUSIC and other methods, and can improve the online monitoring and operating efficiency of the system.
6. In the method, the classification model adopts a two-stage Stacking-based method to realize the fusion of the identification results of the two classifiers of the support vector machine and the random forest, thereby further improving the accuracy of the estimation of the vertical offset distance.
7. In the method, the signal energy characteristic value of the frequency band power spectrum in the areas near a plurality of peak values is calculated, and particularly, the average value of the average power values of all the obtained peak values is calculated to be used as the signal energy characteristic value of the frequency band power spectrum by calculating the average power value in a frequency interval range near each peak value. By searching the concentrated part of the signal energy, namely the peak area, the problem of inaccurate estimation caused by the attenuation difference of different frequency components in underground propagation energy can be avoided.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present invention and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained according to the drawings without inventive efforts.
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a diagram of excavator construction data sets collected at different vertical offset distances at an actual site in accordance with an embodiment of the present invention;
FIG. 3 is a graph of energy attenuation characteristics for the same construction event at different vertical offsets in accordance with an embodiment of the present invention;
FIG. 4 is a classification model architecture diagram of the present invention;
FIG. 5 is a diagram illustrating the results of classifying confusion matrices and vertical offset distance classifications in an exemplary embodiment of the present invention;
fig. 6 is a diagram illustrating the structure and operation of a distributed optical fiber vibration sensing system according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the detailed description and specific examples, while indicating the preferred embodiment of the invention, are intended for purposes of illustration only and are not intended to limit the scope of the invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
It is noted that relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The features and properties of the present invention are described in further detail below with reference to examples.
In the embodiment, taking the safety monitoring application of the communication optical cable as an example, one fiber core of the existing buried communication optical cable is used as a detection optical fiber, the total length is 15km, the time sampling rate is set to be 5kHz, the space sampling interval is 5.16m, and the construction simulation of the excavator is carried out near the optical cable which is about 8km away from the monitoring center, and the operation is divided into two events of excavating and knocking.
The method flow is as shown in figure 1, longitudinal time signals of each space point distributed along the buried optical fiber are collected according to the setting, and full-line space-time signals distributed along the optical fiber are obtained;
determining the central position of a vibration event along the distribution of a buried optical fiber, selecting a plurality of longitudinal time signals distributed along the optical fiber space to form a signal array according to the influence range of the vibration event on the left and the right of the central position, carrying out frequency domain signal processing on each space point time signal in the selected area and constructing an energy attenuation characteristic sequence of the event signal distributed along the optical fiber;
during off-line training, acquiring event signals under different vertical offset distances when a certain vibration event occurs, constructing an event sample data set, calculating to obtain an energy attenuation characteristic sequence of each event signal sample distributed along the optical fiber as a training set, inputting the energy attenuation characteristic sequence into a machine learning model integrated by a plurality of classifiers for training to obtain a classification model of the vertical offset distance between the vibration source and the optical fiber;
during online testing, the energy attenuation characteristic sequences distributed along the optical fiber within a certain vibration event signal influence range are input into a constructed and trained classification model, the vertical offset distance between the vibration source and the optical fiber is identified, and then the threat level corresponding to the vertical offset distance is obtained.
In a specific embodiment, the data collection and database construction process is as follows:
when a vibration event is detected to occur at a certain position, a collection point at the center of the event occurrence position is taken as a center, 25 point positions of 12 points are respectively taken from the left and the right, namely the space influence range along the optical fiber 129m, the length of an event sample is selected to be 30s, and a space-time signal sample of a single event with the size of (25, 150000) is constructed. A data set constructed based on the field data collected in this embodiment is shown in fig. 2. In addition, three levels of type I (0-4m), type II (5-10m) and type III (> 10m) can be preset according to the threat degree, and the training set and the test set are divided according to the ratio of 85: 15. The threat degree classification of the time sampling rate, the space sampling interval, the acquired space point, the event sample length and the vertical distance can be set according to the actual situation.
In a specific embodiment, a specific method for processing a spatio-temporal signal within an event influence range and constructing an energy attenuation characteristic sequence of the event signal distributed along an optical fiber is as follows:
s2.1 time domain to frequency domain conversion: respectively carrying out Fourier transform on the time signals of each space point to obtain a power spectrum on a frequency domain;
S2.2 frequency domain filtering: noise components outside the useful frequency range are filtered out in the frequency domain by filtering. Empirically, the mechanical vibration signals generated by the construction machine are in a low frequency region below 100Hz, and the time signal of each acquisition point is filtered by a low-pass filter of 100Hz to remove high-frequency noise.
S2.3, selecting a main frequency component interval and calculating an energy characteristic value: after filtering, in order to avoid the problem of inaccurate estimation caused by the attenuation difference of different frequency components in underground propagation energy, a peak area, which is a concentrated part of signal energy, needs to be searched, and the energy characteristics of the peak area are calculated by frequency band power spectrums of areas near several peaks. In this embodiment, the first three spectral peaks are searched, the average power values P (1), P (2), and P (3) within 2Hz range near the 3 spectral peaks are calculated, and then the average value P of the three values is obtained i As a signal energy characteristic value obtained at this point, (p (1) + p (2) + p (3))/3. The selection of the 2Hz frequency interval is set according to the actual situation, and other intervals such as 3Hz, 4Hz and 5Hz can be set.
S2.4, the energy characteristic value obtained by processing each space point signal by S2.1-S2.3,and (3) splicing according to the spatial distribution sequence to construct an energy distribution characteristic vector with the length of 25: p ═ P 1 ，P 2 ，...，P 25 And the energy attenuation characteristic sequence is prepared for next classification and identification.
Fig. 3 is an energy attenuation characteristic curve of two types of mechanical events obtained based on the method, fig. 3(a) and (b) are energy attenuation characteristic curves obtained from mechanical knock event signals under different vertical distances, and fig. 3(c) is an energy attenuation characteristic curve obtained from excavator digging event signals under different vertical distances. It can be seen that the attenuation law of the energy attenuation curve corresponding to different distances has better discrimination.
In a specific embodiment, in order to further improve the accuracy of the estimation of the vertical offset distance, the machine learning model is a classification model obtained by two models, namely a Random Forest (RF) model and a Support Vector Machine (SVM) model based on two-stage Stacking fusion, 4-fold cross validation is performed by using 4 SVM classifiers to output a prediction result, and the result is combined into a new feature to be trained by using the RF classifier.
Further, in a specific embodiment, the process of offline building and training the classification model fused by the secondary Stacking specifically comprises:
s6.1, for each model obtained by training of the support vector machine, dividing a training set D into k parts (wherein k is 4 and corresponds to 4-fold intersection), taking each part as a test set, training the model by using a residual data set, and predicting the result of the part;
S6.2, repeating the S6.1 until each model obtains a prediction result, and obtaining a secondary model training set;
s6.3, averaging k test sets to obtain a test set of the secondary model;
s6.4, taking the random forest classifier as a secondary model, and respectively training and testing the secondary model by using the training set and the testing set obtained in S6.2 and S6.3;
and S6.5, after the training of the secondary model in the S6.4 is finished, saving the two-stage model as a vertical offset distance classifier.
Further, in a specific implementation manner, the random forest classification algorithm in the classification model is used for constructing a plurality of parallel decision trees for classification respectively, then majority voting is performed to obtain a final classification result, the support vector machine classification algorithm is used for mapping low-dimensional data onto a high-dimensional space by using a kernel function, and the classification is performed by hyperplane segmentation.
In the actual online test, the test set in the data set is respectively subjected to spatial energy attenuation feature extraction and model classification, and the obtained classification result, i.e., the confusion matrix, is shown in fig. 5. And calculating the accuracy of distance estimation of the samples in the corresponding positioning precision range according to the confusion matrix. Taking 5m events as an example, counting the number of classified samples within the range of the positioning accuracy +/-1 m, namely the number of event samples classified into 4 meters, 5 meters and 6 meters, and dividing the number by the total number of the real 5m event samples to obtain the distance judgment accuracy within the range of the positioning accuracy +/-1 m. As can be calculated from fig. 5, for the excavator knocking event, the event distance judgment accuracy within the range of the positioning accuracy +/-1 m is 92.25%, and the event distance judgment accuracy within the range of the positioning accuracy +/-2 m is 100%; for test data actually mined by an excavator, the event distance judgment accuracy within the range of positioning accuracy +/-1 m is 83.5%, and the event distance judgment accuracy within the range of positioning accuracy +/-2 m is 86.7%. Further, according to the range of the vertical offset distance of the event in embodiment 1, the event is divided into three sections, for example, I (0 to 5m), II (6 to 10m), and III (10m or more), in order from near to far, and the threat level is discriminated based on the range of the section to which the distance estimation result belongs. The threat level classification accuracy of the event is calculated from fig. 5, with a classification accuracy of 99.06% for the tap event and 82.03% for the mine event.
Further, in particular embodiments, a distributed optical fiber vibration sensing system may be employed in the method to collect longitudinal time signals at various spatial points distributed along the buried optical fiber within the range of influence of the vibration event. The distributed optical fiber vibration sensing system comprises a processing host, optical signal demodulation equipment and a detection optical cable which are sequentially connected, wherein the optical signal demodulation equipment specifically comprises a narrow-band laser, an acousto-optic modulator, an optical amplifier, an isolator, a circulator, a filter, a first coupler, an interferometer, a second coupler, a photoelectric detector and an analog-to-digital converter which are sequentially connected. The basic architecture and operating principle of the system are shown in fig. 6. The detection optical cable generally adopts a common single-mode communication optical fiber, can directly utilize a spare fiber core of the communication optical cable laid along a road or a pipeline, or can be buried along an underground pipeline or a town road for sensing external invasion. The optical signal demodulation device is used as a core part of the system. One path of continuous coherent optical signals generated by the narrow linewidth laser is modulated by the acousto-optic modulator to become narrow pulse light, and the narrow pulse light is amplified by the optical amplifier and then injected from one end of the detection optical cable through the isolator, the port 1 of the circulator and the port 2 of the circulator in sequence. When an event occurs, backward Rayleigh scattering optical signals generated in the process of transmitting the optical pulse signals along the detection optical cable return along the optical cable and are received by ports 2 and 3 of the circulator, the backward Rayleigh scattering optical signals are sequentially filtered by the optical filter and coupled by the first coupler and then injected into the Michelson interferometer to generate interference, phase change information introduced by external disturbance is output by the 3 x 3 second coupler, and signals generated by the action of vibration or sound waves on the optical fiber can be acquired. The optical signal is converted into an electric signal by the photoelectric detector, and then the signal is synchronously acquired by the synchronous trigger analog-to-digital converter controlled by the acquisition card and transmitted to the signal processing host in real time. The signal processing host serves as a system terminal and is used for analyzing and processing optical fiber array signals, detecting a vibration source and identifying the position of the vibration source along an optical fiber, estimating the vertical offset distance, and giving an accurate alarm based on an identification result and the threat level of the vibration source.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (7)
1. A method for judging vertical offset distance and threat level of an optical fiber sensing event is characterized by comprising the following steps: the method comprises the following steps:
collecting longitudinal time signals of each space point distributed along the buried optical fiber to obtain all-line space-time signals distributed along the optical fiber;
determining the central position of a vibration event distributed along a buried optical fiber, selecting a plurality of longitudinal time signals distributed along the optical fiber space to form a signal array according to the vibration event influence range on the left and right of the central position, performing frequency domain signal processing on each space point time signal in the signal array to obtain the energy characteristic of the signal array, and splicing the energy characteristic values of different space points according to the spatial sequence to construct an energy attenuation characteristic sequence of the event signal distributed along the optical fiber;
during off-line training, acquiring event signals under different vertical offset distances when a certain vibration event occurs, constructing an event sample data set, calculating to obtain an energy attenuation characteristic sequence of each event signal sample distributed along the optical fiber as a training set, inputting the energy attenuation characteristic sequence into a machine learning model integrated by a plurality of classifiers for training to obtain a classification model of the vertical offset distance between the vibration source and the optical fiber;
During on-line testing, the energy attenuation characteristic sequence distributed along the optical fiber within a certain vibration event signal influence range is input into a constructed and trained classification model, the vertical offset distance between the vibration source and the optical fiber is identified, and then the threat level corresponding to the vertical offset distance is obtained.
2. The method as claimed in claim 1, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: the specific method for constructing the energy attenuation characteristic sequence of the event signal distributed along the optical fiber comprises the following steps:
s2.1 time domain to frequency domain conversion: respectively carrying out Fourier transform on the time signals of each space point to obtain a power spectrum on a frequency domain;
s2.2, frequency domain filtering: filtering out noise components outside the useful frequency range in the frequency domain;
s2.3, selecting a main frequency component interval and calculating an energy characteristic value: after filtering, searching a peak value area in a frequency domain, and calculating a signal energy characteristic value of the frequency band power spectrum of the area near a plurality of peak values;
s2.4, splicing signal energy characteristic values obtained after the signals of each space point are subjected to S2.1-S2.3 processing according to a space distribution sequence, and constructing an energy distribution characteristic vector as an energy attenuation characteristic sequence of the event signals distributed along the optical fiber.
3. The method as claimed in claim 2, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: the specific method for calculating the signal energy characteristic value by the frequency band power spectrum of the areas near the plurality of peak values comprises the following steps: and calculating the average power value in a frequency interval range near each peak value, and calculating the average value of the obtained average power values near all the peak values as the signal energy characteristic value.
4. The method as claimed in claim 1, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: the machine learning model is a classification model obtained by fusing a random forest model and a support vector machine model based on two-stage Stacking, 4-fold cross validation is carried out by using 4 support vector machine classifiers to output a prediction result, the result is combined into new features, and the new features are trained by using the random forest classifier.
5. The method as claimed in claim 4, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: in the classification model, a support vector machine classification algorithm maps low-dimensional data to a high-dimensional space by using a kernel function, the low-dimensional data are distinguished by hyperplane segmentation, a random forest classification algorithm is used for classifying a plurality of parallel decision trees respectively by constructing the parallel decision trees, and then a final classification result is obtained by majority voting.
6. The method as claimed in claim 4, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: the off-line construction and training process of the two models of the random forest and the support vector machine based on the two-stage Stacking fusion specifically comprises the following steps:
s6.1, training to obtain each support vector machine model: dividing the training set D into k parts, firstly taking each part as a test set, training a model by using a residual data set, and predicting the result of the part;
s6.2, repeating the S6.1 until each model obtains a prediction result, and obtaining a secondary model training set;
s6.3, averaging k test sets to obtain a test set of the secondary model;
s6.4, taking the random forest classifier as a secondary model, and respectively training and testing the secondary model by using the training set and the testing set obtained in S6.2 and S6.3;
and S6.5, after the training of the secondary model in the S6.4 is finished, saving the two-stage model as a vertical offset distance classifier.
7. The method as claimed in claim 1, wherein the method for determining the vertical offset distance and threat level of the optical fiber sensing event comprises: in the method, a distributed optical fiber vibration sensing system is adopted to collect longitudinal time signals of each spatial point distributed along a buried optical fiber within the influence range of a vibration event.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911125636.8A CN111222743B (en) | 2019-11-15 | 2019-11-15 | Method for judging vertical offset distance and threat level of optical fiber sensing event |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911125636.8A CN111222743B (en) | 2019-11-15 | 2019-11-15 | Method for judging vertical offset distance and threat level of optical fiber sensing event |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111222743A CN111222743A (en) | 2020-06-02 |
| CN111222743B true CN111222743B (en) | 2022-07-29 |
Family
ID=70825801
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911125636.8A Active CN111222743B (en) | 2019-11-15 | 2019-11-15 | Method for judging vertical offset distance and threat level of optical fiber sensing event |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111222743B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111912513B (en) * | 2020-07-14 | 2022-03-22 | 国家电网有限公司 | Identification method of excavator construction event along optical cable based on phi-OTDR |
| CN111917466B (en) * | 2020-07-24 | 2021-11-09 | 中山水木光华电子信息科技有限公司 | Optical fiber fault point monitoring and identifying system and method |
| CN112069912B (en) * | 2020-08-13 | 2023-05-12 | 国家电网有限公司 | Method for identifying construction threat event of optical cable channel based on phi-OTDR |
| CN112067113B (en) * | 2020-09-08 | 2022-03-25 | 武汉理工光科股份有限公司 | Method and device for determining propagation velocity of Rayleigh waves in soil |
| CN112464777B (en) * | 2020-11-20 | 2023-04-18 | 电子科技大学 | Intelligent estimation method for vertical distance of optical fiber vibration source |
| CN112504430B (en) * | 2020-11-26 | 2022-03-01 | 南京大学 | Tower structure health monitoring method based on vibration sensitive optical fiber sensing technology |
| CN114166331A (en) * | 2021-12-03 | 2022-03-11 | 武汉理工光科股份有限公司 | Non-threat signal identification method and device, electronic equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5191557A (en) * | 1986-12-30 | 1993-03-02 | Gas Research Institute | Signal processing to enable utilization of a rig reference sensor with a drill bit seismic source |
| JP2013072800A (en) * | 2011-09-28 | 2013-04-22 | Chugoku Electric Power Co Inc:The | Vibration detection system |
| CN105510959A (en) * | 2015-11-30 | 2016-04-20 | 山东康威通信技术股份有限公司 | Method for recognizing type of vibration source of tunnel, and vibration source positioning method |
| CN106404161A (en) * | 2016-08-19 | 2017-02-15 | 山东康威通信技术股份有限公司 | Calculation method for precisely positioning vibration position in tunnel construction |
| CN108932480A (en) * | 2018-06-08 | 2018-12-04 | 电子科技大学 | The study of distributing optical fiber sensing signal characteristic and classification method based on 1D-CNN |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109520692A (en) * | 2018-12-20 | 2019-03-26 | 泉州亿兴电力有限公司 | A kind of vibration signal propagation test equipment and method |
-
2019
- 2019-11-15 CN CN201911125636.8A patent/CN111222743B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5191557A (en) * | 1986-12-30 | 1993-03-02 | Gas Research Institute | Signal processing to enable utilization of a rig reference sensor with a drill bit seismic source |
| JP2013072800A (en) * | 2011-09-28 | 2013-04-22 | Chugoku Electric Power Co Inc:The | Vibration detection system |
| CN105510959A (en) * | 2015-11-30 | 2016-04-20 | 山东康威通信技术股份有限公司 | Method for recognizing type of vibration source of tunnel, and vibration source positioning method |
| CN106404161A (en) * | 2016-08-19 | 2017-02-15 | 山东康威通信技术股份有限公司 | Calculation method for precisely positioning vibration position in tunnel construction |
| CN108932480A (en) * | 2018-06-08 | 2018-12-04 | 电子科技大学 | The study of distributing optical fiber sensing signal characteristic and classification method based on 1D-CNN |
Non-Patent Citations (7)
| Title |
|---|
| Huijuan Wu 等.Vertical Offset-Distance Estimation and Threat Level Prediction of Vibrations With DAS.《IEEE Access》.2020,第8卷177245 - 177254. * |
| Oliver Heirich 等.Velocity and location information from onboard vibration measurements of rail vehicles.《Proceedings of the 16th International Conference on Information Fusion》.2013,1835-1840. * |
| Yu Hen Hu 等.Energy based collaborative source localization using acoustic micro-sensor array.《2002 IEEE Workshop on Multimedia Signal Processing》.2003,371-375. * |
| 刘路等.管道光纤安全预警技术中的威胁事件识别算法研究.《石油科技论坛》.2016,第35卷(第S1期), * |
| 路豪.基于集成学习模型的DAS振动源二维定位方法研究.《中国优秀博硕士学位论文全文数据库(硕士)信息科技辑》.2022,(第01期),I136-627. * |
| 郑丁 等.基于DVS的振动源与传感通道垂直距离的计算方法.《http://kns.cnki.net/kcms/detail/45.1160.tn.20190701.1004.002.html》.2019,1-8. * |
| 郭学彬 等.论爆破地震波在传播过程中的衰减特性.《中国矿业》.2006,(第03期),51-53+57. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111222743A (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111222743B (en) | Method for judging vertical offset distance and threat level of optical fiber sensing event | |
| US11562224B2 (en) | 1D-CNN-based distributed optical fiber sensing signal feature learning and classification method | |
| CN109344195B (en) | HMM model-based pipeline security event recognition and knowledge mining method | |
| CN107256635B (en) | Vehicle identification method based on distributed optical fiber sensing in intelligent traffic | |
| CN201191221Y (en) | Disturbance signal determination module of distributed optical fiber vibration sensor system | |
| Bai et al. | Detection and identification of external intrusion signals from 33 km optical fiber sensing system based on deep learning | |
| CN102563361A (en) | Device and method for detecting and positioning leakage of gas transmission pipeline based on conventional data and sound wave signals | |
| CN101556724A (en) | Safety management system of optical fiber perimeter and pattern recognition method thereof | |
| CN109190272B (en) | Concrete structure defect detection method based on elastic waves and machine learning | |
| CN111537056A (en) | Pipeline along-line third-party construction dynamic early warning method based on SVM and time-frequency domain characteristics | |
| US20220065977A1 (en) | City-scale acoustic impulse detection and localization | |
| Yang et al. | Pipeline safety early warning by multifeature-fusion CNN and LightGBM analysis of signals from distributed optical fiber sensors | |
| CN112985574A (en) | High-precision classification identification method for optical fiber distributed acoustic sensing signals based on model fusion | |
| CN101299635B (en) | Apparatus and method for discriminating front and back of wide field complete-optical fiber disturbing signal | |
| CN111006849A (en) | Method and system for judging laying state of oil-gas pipeline accompanying optical cable | |
| JP2023550091A (en) | Vertical distance prediction of vibrations using distributed fiber optic sensing | |
| CN110361164A (en) | A kind of submarine optical fiber cable breakpoint location and disturbance location analysis method and system | |
| CN116026449B (en) | Vibration positioning monitoring system based on single-core optical fiber sensing | |
| CN105222885A (en) | Optical fiber vibration detection method and device | |
| CN116186642B (en) | Distributed optical fiber sensing event early warning method based on multidimensional feature fusion | |
| CN105046859A (en) | Optical fiber intrusion detection method based on vibration signal space-time two-dimensional sparse representation K-S detection | |
| US20230024104A1 (en) | Identification of false transformer humming using machine learning | |
| CN116046145A (en) | System and method for detecting consistency of traffic road optical cable areas | |
| CN114719950A (en) | Step-by-step optical fiber multi-dimensional characteristic parameter acquisition and extraction method | |
| CN215569800U (en) | Railway water supply pipe network leak detection system based on time domain reflection technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |