An apparatus and method for decoupling temperature and strain based on an ANN algorithm

Abstract

The application discloses a kind of based on ANN algorithm's optical fiber Brillouin distributed temperature, strain decoupling device and method, it is related to optical technology field.The technical points of the present application include: propose a kind of optical fiber Brillouin distributed temperature, strain decoupling device, fusion OTDR and BOTDR technology, OTDR is used to correct the local position loss distortion of sensing optical fiber in the long-term work process of BOTDR, temperature, strain, attenuation data can be measured simultaneously;Based on the above device, a kind of temperature, strain decoupling method based on ANN algorithm is proposed, the data acquisition module of fusion OTDR and BOTDR is used to collect the Brillouin gain spectrum of the optical fiber to be measured;The data fitting of the Brillouin gain spectrum of the optical fiber to be measured is carried out using Lorentz fitting algorithm, the corresponding Brillouin center frequency shift, spectrum width and amplitude are obtained, and input into the pre-trained detection model based on ANN algorithm to obtain the temperature and strain value of the optical fiber to be measured;The method effectively improves the measurement accuracy and monitoring efficiency of optical fiber sensing without increasing hardware cost.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical technology, and in particular to an optical fiber Brillouin distributed temperature and strain decoupling device and method based on an ANN algorithm. BACKGROUND

[0002] A Brillouin optical time-domain reflectometer (BOTDR) utilizes backscattered Brillouin scattering signals in an optical fiber to achieve spatial continuous measurement of temperature and strain of a to-be-measured object along the optical fiber, and has advantages of long measurement distance, distribution, high precision, no electromagnetic interference, intrinsic safety, easy deployment, and no power supply required at the sensing end, etc. The BOTDR has been applied in many fields such as power line monitoring, large infrastructure monitoring (such as bridges, tunnels, pipe galleries), oil and gas pipeline monitoring, and has achieved remarkable results. However, Brillouin scattering is simultaneously affected by temperature and strain, and the system has cross-sensitivity problems in temperature and strain demodulation, which brings difficulties to practical application.

[0003] Traditional demodulation methods for simultaneous measurement of strain and temperature based on Brillouin scattering technology mainly include a reference optical fiber method, a grating auxiliary demodulation method, a multi-parameter demodulation method, a special optical fiber method, a dual-wavelength sensing demodulation method, a Raman / Rayleigh scattering auxiliary demodulation method, and an artificial neural network demodulation algorithm. For existing overhead cables, it is impossible to realize parallel laying of reference optical fibers, gratings, and special optical fibers with the to-be-measured optical fiber, and the Raman scattering auxiliary demodulation method has a limited sensing distance and cannot meet the existing optical cable monitoring requirements; the multi-parameter demodulation scheme utilizes the linear relationship between the frequency shift and peak power of Brillouin spectrum and temperature and strain to construct a binary linear equation system for solving, but in actual application, factors such as bending and connection coupling of the optical fiber will cause instability of the peak power of the Brillouin spectrum, and power fluctuations of the laser, pump pulse signal, and probe signal will also cause fluctuations of the peak value of the Brillouin spectrum, limiting the measurement accuracy of the system; the dual-wavelength sensing demodulation method utilizes two working wavelengths (1550 nm and 850 nm) for dual-parameter simultaneous sensing, but the optical fiber loss at the 850 nm working wavelength is large, resulting in low signal-to-noise ratio of the system and limiting the sensing distance; the temperature and strain dual-parameter measurement method based on Brillouin and Rayleigh scattering technology utilizes a DPP-BOTDA system to measure the Brillouin frequency shift and an OFDR system to measure the Rayleigh backscattering spectrum frequency shift, and demodulates temperature and strain according to a calibration temperature and strain coefficient matrix, but the system cannot be applied to long-distance power line monitoring due to the limited sensing distance of the OFDR; Ruiz-Lombera et al. of the University of Cantabria in Spain utilized an artificial neural network to realize simultaneous measurement of temperature and strain on a single-mode optical fiber, but the temperature and strain resolution obtained by the artificial neural network solution to cross-sensitivity is low.

[0004] The above prior arts effectively solve the problem of simultaneous measurement of strain and temperature in Brillouin fiber sensing, but there are still problems such as that the existing laid optical fiber cannot change the optical fiber hardware facilities, the sensing distance is insufficient, and the measurement accuracy is limited. SUMMARY

[0005] To this end, the present application proposes an optical fiber Brillouin distributed temperature and strain decoupling device and method based on an ANN algorithm, in an attempt to solve or at least alleviate at least one of the above problems.

[0006] According to an aspect of the present application, an optical fiber Brillouin distributed temperature and strain decoupling device based on an ANN algorithm is proposed, which comprises a data acquisition module and a data processing module, the data acquisition module is used to acquire the Brillouin gain spectrum of the optical fiber, comprising: a broadband laser, a narrow spectrum width laser, a first optical fiber coupler, a first dense wavelength division multiplexer, a first electro-optical modulator, a doped fiber amplifier, a random polarization scrambler, a first circulator, a fiber to be measured, a second dense wavelength division multiplexer, a photoelectric detector, an arbitrary function generator, a data acquisition card, a second circulator, an optical fiber grating filter, a microwave source, a second electro-optical modulator, a second optical fiber coupler, a photoelectric balance detector, and a detector; wherein,

[0007] The optical signal output end of the narrow spectrum width laser is in communication with the optical signal input end of the first optical fiber coupler, the No. 1 optical signal output end of the first optical fiber coupler and the optical signal output end of the broadband laser are respectively and simultaneously in communication with the optical signal input end of the first dense wavelength division multiplexer, the optical signal output end of the first dense wavelength division multiplexer is in communication with the optical signal input end of the first electro-optical modulator, the optical signal output end of the first electro-optical modulator is in communication with the optical signal input end of the doped fiber amplifier, the optical signal output end of the doped fiber amplifier is in communication with the random polarization scrambler, the output end of the random polarization scrambler is in communication with the No. 1 optical signal port of the first circulator, and the No. 2 optical signal port of the first circulator is in communication with the fiber to be measured;

[0008] The No. 3 optical signal port of the first circulator is in communication with the input end of the second dense wavelength division multiplexer, the No. 1 output end of the second dense wavelength division multiplexer is in communication with the optical signal input end of the photoelectric detector, and the electrical signal output end of the photoelectric detector is in communication with the No. 1 electrical signal data acquisition end of the data acquisition card; the No. 2 output end of the second dense wavelength division multiplexer is in communication with the No. 1 optical signal port of the second circulator, the No. 2 optical signal port of the second circulator is in communication with the optical fiber grating filter, and the No. 3 optical signal port of the second circulator is in communication with the No. 1 input end of the second optical fiber coupler;

[0009] The second optical signal output end of the first optical fiber coupler and the signal output end of the microwave source are in communication with the optical signal input end and the microwave signal loading end of the second electro-optical modulator respectively and simultaneously, the optical signal output end of the second electro-optical modulator is in communication with the second input end of the second optical fiber coupler, the two output ends of the second optical fiber coupler are in communication with the two input ports of the balanced detector respectively and simultaneously, the electrical signal output end of the photoelectric balanced detector is in communication with the signal input end of the detector, and the signal output end of the detector is in communication with the second electrical signal data acquisition end of the data acquisition card simultaneously;

[0010] The microwave signal output end of the arbitrary function generator is in communication with the microwave signal loading end of the first electro-optical modulator and the trigger signal input end of the data acquisition card respectively and simultaneously;

[0011] The data acquisition card outputs the Brillouin gain spectrum of the to-be-tested optical fiber to the data processing module.

[0012] Further, the data processing module comprises a data fitting sub-module and a data decoupling sub-module; the data fitting sub-module is configured to perform data fitting on the Brillouin gain spectrum of the to-be-tested optical fiber by using a Lorentz fitting algorithm to obtain a corresponding Brillouin center frequency shift, spectral width and amplitude; and the data decoupling sub-module is configured to input the Brillouin center frequency shift, spectral width and amplitude into a pre-trained detection model based on an ANN algorithm to obtain the temperature and strain value of the to-be-tested optical fiber.

[0013] Further, the data fitting sub-module performs Lorentz data fitting according to the following formula to obtain the Brillouin center frequency shift, spectral width and amplitude:

[0014]

[0015] wherein g0 represents the Brillouin gain, the amplitude P = g0 2 ; v b represents the Brillouin center frequency shift; and Δν represents the full width at half maximum of the Brillouin gain spectrum, i.e. the spectral width.

[0016] Further, the pre-training process of the detection model based on the ANN algorithm in the data decoupling sub-module comprises:

[0017] Step one, obtaining a training data set and a test data set, and pre-processing the data in the data set; specifically including:

[0018] applying different temperatures to the optical fiber, using the data acquisition module to acquire the Brillouin gain spectrum of each point of the optical fiber, and taking the temperature as the corresponding label value; using the Lorentz fitting algorithm to perform data fitting on the Brillouin gain spectrum of each point to obtain a plurality of sets of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different temperatures, and taking them as a first input parameter set;

[0019] Different stresses are applied to the optical fiber, and the Brillouin gain spectrum of each point of the optical fiber is acquired by using the data acquisition module, and the strain is taken as the corresponding label value; the Brillouin gain spectrum of each point is subjected to data fitting by using a Lorentz fitting algorithm, and a plurality of groups of Brillouin center frequency shifts, spectrum widths and amplitudes corresponding to different stresses are obtained, which are taken as a second input parameter set;

[0020] Different temperatures and stresses are applied to the optical fiber at the same time, and the Brillouin gain spectrum of each point of the optical fiber is acquired by using the data acquisition module, and the temperature and the strain are taken as the corresponding label value; the Brillouin gain spectrum of each point is subjected to data fitting by using a Lorentz fitting algorithm, and a plurality of groups of Brillouin center frequency shifts, spectrum widths and amplitudes corresponding to different temperatures and stresses are obtained, which are taken as a third input parameter set;

[0021] The first input parameter set, the second input parameter set and the third input parameter set are combined and divided into a training data set and a test data set;

[0022] The data in the data set is preprocessed, including eliminating unreasonable data and performing normalization processing;

[0023] Step two, inputting the training data set into the detection model based on the ANN algorithm for supervised learning training;

[0024] Step three, inputting the test data set into the detection model based on the ANN algorithm obtained by training for testing, comparing the deviation between the actual output parameter and the output parameter of the test data set, and calculating the maximum error and the average absolute error;

[0025] Step four, judging whether the error is greater than a preset threshold, if yes, adjusting the structure of the detection model based on the ANN algorithm and repeating step three until the error is not greater than the preset threshold to stop training, and obtaining the trained detection model based on the ANN algorithm.

[0026] Further, the wideband light source of the wideband laser has an output power of 20 mW, a wavelength of 1550 nm and a spectrum width of 100 GHz; the narrow spectrum width light source of the narrow spectrum width laser has an output power of 40 mW, a wavelength of 1550 nm and a spectrum width of 1 MHz; the coupling ratio of the first optical fiber coupler is 90:10 or 80:20 or 70:30, and the coupling ratio of the second optical fiber coupler is 50:50; the center wavelength of the fiber grating filter is 1550 nm, and the bandwidth is 0.1 nm; and the detection bandwidth range of the photoelectric balance detector is 200 MHz-500 MHz.

[0027] According to another aspect of the present application, a kind of based on ANN algorithm's optical fiber Brillouin distributed temperature, strain decoupling method is provided, and the method is based on above-mentioned one based on ANN algorithm's optical fiber Brillouin distributed temperature, strain decoupling device is realized, the method includes the following steps:

[0028] acquire the Brillouin gain spectrum of the optical fiber to be measured by using the data acquisition module fusing OTDR and BOTDR;

[0029] perform data fitting on the Brillouin gain spectrum of the optical fiber to be measured by using a Lorentz fitting algorithm to obtain the corresponding Brillouin central frequency shift, spectral width and amplitude;

[0030] input the Brillouin central frequency shift, spectral width and amplitude into a pre-trained detection model based on ANN algorithm to obtain the temperature and strain value of the optical fiber to be measured.

[0031] Further, the Lorentz data fitting is performed according to the following formula to obtain the Brillouin central frequency shift, spectral width and amplitude:

[0032]

[0033] wherein g0 represents the Brillouin gain, the amplitude P = g0 2 ; v b represents the Brillouin central frequency shift; Δν represents the full width at half maximum of the Brillouin gain spectrum, i.e. the spectral width.

[0034] Further, the pre-training process of the detection model based on ANN algorithm comprises:

[0035] Step one, obtain a training data set and a test data set, and pre-process the data in the data set; specifically including:

[0036] apply different temperatures to the optical fiber, acquire the Brillouin gain spectrum of each point of the optical fiber by using the data acquisition module, and use the temperature as the corresponding label value; perform data fitting on the Brillouin gain spectrum of each point by using the Lorentz fitting algorithm to obtain multiple sets of Brillouin central frequency shift, spectral width and amplitude corresponding to different temperatures, which are used as the first input parameter set;

[0037] apply different stresses to the optical fiber, acquire the Brillouin gain spectrum of each point of the optical fiber by using the data acquisition module, and use the strain as the corresponding label value; perform data fitting on the Brillouin gain spectrum of each point by using the Lorentz fitting algorithm to obtain multiple sets of Brillouin central frequency shift, spectral width and amplitude corresponding to different stresses, which are used as the second input parameter set;

[0038] apply different temperatures and stresses to the optical fiber simultaneously, acquire the Brillouin gain spectrum of each point of the optical fiber by using the data acquisition module, and use the temperature and strain as the corresponding label value; perform data fitting on the Brillouin gain spectrum of each point by using the Lorentz fitting algorithm to obtain multiple sets of Brillouin central frequency shift, spectral width and amplitude corresponding to different temperatures and stresses, which are used as the third input parameter set;

[0039] The first input parameter set, the second input parameter set and the third input parameter set are combined and divided into a training data set and a test data set;

[0040] The preprocessing of the data in the data set comprises: eliminating unreasonable data and performing normalization processing;

[0041] Step two, inputting the training data set into the detection model based on the ANN algorithm for supervised learning training;

[0042] Step three, inputting the test data set into the detection model based on the ANN algorithm obtained by training for testing, comparing the deviation between the actual output parameter and the output parameter of the test data set, and calculating the maximum error and the average absolute error;

[0043] Step four, judging whether the error is greater than a preset threshold, if yes, adjusting the structure of the detection model based on the ANN algorithm and repeating step three until the error is not greater than the preset threshold to stop training, and obtaining the trained detection model based on the ANN algorithm.

[0044] The beneficial technical effects of the present application are:

[0045] The present application provides a kind of optical fiber Brillouin distributed temperature, strain decoupling device, fusion OTDR and BOTDR technology, OTDR is used to correct the local position loss distortion of sensing optical fiber in the long-term working process of BOTDR system, and a single system can realize temperature, strain, attenuation measurement simultaneously;Randomly replace orthogonal winding with winding, reduce the intensity fluctuation caused by orthogonal winding, reduce the error of temperature and strain demodulation.

[0046] The present application also provides a kind of optical fiber Brillouin distributed temperature, strain decoupling method based on ANN algorithm, utilizes ANN algorithm to combine Brillouin center frequency shift, spectral width, amplitude to realize temperature, strain simultaneous measurement, under the premise of not increasing hardware cost, effectively improve the measurement precision and monitoring efficiency of optical fiber sensing. BRIEF DESCRIPTION OF DRAWINGS

[0047] The above and other objects, features and advantages of the exemplary embodiments of the present application will be more apparent from the following detailed description read in conjunction with the accompanying drawings, in which several embodiments of the present application are shown by way of example, and not limitation. In the drawings:

[0048] Figure 1 It is a structure schematic view of a kind of optical fiber Brillouin distributed temperature, strain decoupling device based on ANN algorithm for the embodiment of the present application;

[0049] Figure 2 It is the relationship diagram of Brillouin frequency shift, amplitude and spectral width obtained by temperature test and temperature change in the embodiment of the present application;

[0050] Figure 3 The relationship diagram of Brillouin frequency shift, amplitude and spectral width and temperature change obtained by stress test in the embodiment of the application;

[0051] Figure 4 The flow chart of the optical fiber Brillouin distributed temperature and strain decoupling method based on the ANN algorithm in the embodiment of the application. DETAILED DESCRIPTION

[0052] The principles and spirits of the present application will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are given only to enable those skilled in the art to better understand and implement the present application, and do not limit the scope of the present application in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0053] Those skilled in the art know that the embodiments of the present application can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure can be embodied in the form of a complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. In this context, it should be understood that any number of elements in the drawings are used for illustration only, not limitation, and any naming is only for distinction, not any limiting meaning.

[0054] The embodiment of the present application provides an optical fiber Brillouin distributed temperature and strain decoupling device based on an ANN algorithm, which comprises a data acquisition module and a data processing module, wherein the data acquisition module is used for acquiring the Brillouin gain spectrum of the optical fiber, and comprises a wideband laser 1, a narrow spectrum width laser 2, a first optical fiber coupler 3, a first dense wavelength division multiplexer 4, a first electro-optical modulator 5, a doped fiber amplifier 6, a random polarization scrambler 7, a first circulator 8, a fiber to be measured 9, a second dense wavelength division multiplexer 10, a photoelectric detector 11, an arbitrary function generator 12, a data acquisition card 13, a second circulator 14, an optical fiber grating filter 15, a microwave source 16, a second electro-optical modulator 17, a second optical fiber coupler (four-port) 18, a photoelectric balance detector 19, and a detector 20.

[0055] The light signal output end of the narrow spectrum width laser 2 is communicated with the light signal input end of the first optical fiber coupler 3, the No.1 light signal output end of the first optical fiber coupler 3 and the light signal output end of the wideband laser 1 are respectively and simultaneously communicated with the light signal input end of the first dense wavelength division multiplexer 4, the light signal output end of the first dense wavelength division multiplexer 4 is communicated with the light signal input end of the first electro-optical modulator 5, the light signal output end of the first electro-optical modulator 5 is communicated with the light signal input end of the doped fiber amplifier 6, the light signal output end of the doped fiber amplifier 6 is communicated with the random polarization scrambler 7, the output end of the random polarization scrambler 7 is communicated with the No.1 light signal port 8-1 of the first circulator 8, the No.2 light signal port 8-2 of the first circulator 8 is communicated with the optical fiber to be measured 9, the No.3 light signal port 8-3 of the first circulator 8 is communicated with the input end of the second dense wavelength division multiplexer 10, the No.1 output end of the second dense wavelength division multiplexer 10 is communicated with the light signal input end of the photoelectric detector 11, the electrical signal output end of the photoelectric detector 11 is communicated with the No.1 electrical signal data acquisition end of the data acquisition card 13; the No.2 output end of the second dense wavelength division multiplexer 10 is communicated with the No.1 light signal port 14-1 of the second circulator 14, the No.2 light signal port 14-2 of the second circulator 14 is communicated with the fiber grating filter 15, the No.3 light signal port 14-3 of the second circulator 14 is communicated with the No.1 input end of the second optical fiber coupler (four-port) 18, the No.2 light signal output end of the first optical fiber coupler 3 and the signal output end of the microwave source 16 are respectively and simultaneously communicated with the light signal input end and the microwave signal loading end of the second electro-optical modulator 17, the light signal output end of the second electro-optical modulator 17 is communicated with the No.2 input end of the second optical fiber coupler (four-port) 18, the two output ends of the second optical fiber coupler (four-port) 18 are respectively and simultaneously communicated with the two input ports of the balanced detector 19, the electrical signal output end of the photoelectric balanced detector 19 is simultaneously communicated with the signal input end of the detector 20, the signal output end of the detector 20 is simultaneously communicated with the No.2 electrical signal data acquisition end of the data acquisition card 13, the microwave signal output end of the arbitrary function generator 12 is respectively and simultaneously communicated with the microwave signal loading end of the first electro-optical modulator 5 and the trigger signal input end of the data acquisition card 13.

[0056] In the embodiment, preferably, the wideband light source output power of the wideband laser 1 is 20 mW, the wavelength is 1550 nm, and the spectrum width is 100 GHz. The narrow spectrum width light source output power of the narrow spectrum width laser 2 is 40 mW, the wavelength is 1550 nm, and the spectrum width is 1 MHz. The coupling ratio of the first optical fiber coupler is 90:10 or 80:20 or 70:30, and the coupling ratio of the second optical fiber coupler is 50:50. The center wavelength of the fiber grating filter 15 is 1550 nm, and the bandwidth is 0.1 nm. The detection bandwidth of the photoelectric balanced detector 19 is 200 MHz-500 MHz.

[0057] The above data acquisition principle is that the narrow spectrum wide light source output continuous light of the narrow spectrum wide laser 2 is divided into upper and lower two branches through the first optical fiber coupler 3: the continuous light of the upper branch and the broadband light source output continuous light of the broadband laser 1 are combined after the first dense wavelength division multiplexer 4, modulated into probe pulse light by the first electro-optical modulator 5, amplified by the doped fiber amplifier 6, adjusted by the polarization controller, and then injected into the measured optical fiber through the first circulator 8, to generate Rayleigh scattering signals and Brillouin scattering signals with a frequency shift of v b The backscattering signals in the measured optical fiber are separated by the second dense wavelength division multiplexer 10, and the Rayleigh scattering signals are received by the photodetector 11 and recorded by the data acquisition card 13, realizing the OTDR system for detecting the line loss of the measured optical fiber.

[0058] The Brillouin scattering light signals are injected into the optical fiber grating filter 15 through the second circulator 14 to filter out the Stokes light signals and carrier signals; the continuous light of the lower branch is modulated by the second electro-optical modulator 17 to generate a frequency shift of v L The anti-Stokes light signals output by the second circulator 14 are subjected to coherent frequency mixing with the second optical fiber coupler (four-port) 18, and the output signal of the second optical fiber coupler (four-port) 18 is connected to the photobalanced detector 19, and the signal frequency output by the balanced detector 19 is |v b -v L | and envelope extraction and recording are performed by the detector 20 and the data acquisition card 13 to detect the Brillouin gain spectrum, realize the BOTDR sensing system, and detect the strain and temperature; the arbitrary function generator 12 controls the first electro-optical modulator 5 and the data acquisition card 13, and the microwave source 16 controls the second electro-optical modulator 17.

[0059] The data processing module includes a data fitting submodule and a data decoupling submodule; the data fitting submodule is used for data fitting of the Brillouin gain spectrum of the measured optical fiber by using a Lorentz fitting algorithm to obtain the corresponding Brillouin center frequency shift, spectral width and amplitude; the data decoupling submodule is used for inputting the Brillouin center frequency shift, spectral width and amplitude into a pre-trained detection model based on an ANN algorithm to obtain the temperature and strain value of the measured optical fiber. Specifically as follows.

[0060] After the Brillouin gain spectrum of the measured optical fiber is acquired, curve fitting is performed to obtain a Lorentz fitting curve, that is, the Brillouin gain spectrum g b (v) conforms to the Lorentz spectrum, as shown in formula (1):

[0061]

[0062] Wherein, g0 represents the Brillouin gain; v bdenotes the Brillouin central frequency shift; Δν is the full width at half maximum (FWHM) of the Brillouin spectrum.

[0063] Therefore, the Brillouin gain spectrum can be characterized by the Brillouin frequency shift, amplitude and spectral width, etc., wherein the amplitude P = g0 2 The amplitude and spectral width of the Brillouin scattering spectrum are combined with the Brillouin frequency shift respectively to solve the matrix equation, and the temperature coefficient and strain coefficient corresponding to the Brillouin frequency shift, amplitude and spectral width are obtained respectively, so as to realize the separation of temperature and strain.

[0064] The frequency shift v b of the Brillouin scattering spectrum and the amplitude P jointly create a matrix equation as shown in equation (2):

[0065]

[0066] Wherein, Δν b is the change amount of the central frequency shift of the Brillouin scattering spectrum, is the ratio of the change amount of the amplitude of the Brillouin scattering spectrum to the amplitude, and are the temperature coefficient and strain coefficient of the Brillouin frequency shift v b , and are the temperature coefficient and strain coefficient of the amplitude of the Brillouin scattering spectrum.

[0067] The frequency shift v b of the Brillouin scattering spectrum and the Brillouin spectral width jointly create a matrix equation as shown in equation (3):

[0068]

[0069] Wherein, Δν b is the change amount of the central frequency shift of the Brillouin scattering spectrum, and Δv is the change amount of the spectral width of the Brillouin scattering spectrum, and are the temperature coefficient and strain coefficient of the Brillouin frequency shift v b , and are the temperature coefficient and strain coefficient of the spectral width of the Brillouin scattering spectrum.

[0070] Through temperature test and stress test, the Brillouin spectrum under different temperatures and different stresses is obtained, and the temperature coefficient b and the strain coefficient of the Brillouin frequency shift v are calculated respectively. The temperature coefficient and the strain coefficient of the amplitude P of the Brillouin scattering spectrum are calculated respectively. The temperature coefficient and the strain coefficient of the Brillouin frequency shift are calculated respectively. The temperature coefficient and strain coefficient

[0071] The data acquisition module in the embodiment includes two parts of OTDR and BOTDR functions, wherein the OTDR measures the attenuation data of the optical fiber along the line, and compensates the amplitude change of the BOTDR due to the attenuation of the optical fiber, thereby improving the accuracy of temperature and strain decoupling.

[0072] The temperature test of the to-be-tested optical fiber is performed by using the above device, and the relationship between the Brillouin central frequency shift, amplitude and spectral width and the temperature change of the optical fiber is analyzed, as shown in FIG. 2. Figure 2 When the sensing optical fiber is locally heated, the Brillouin spectrum central frequency moves to the right, as shown in (a); the Brillouin spectrum amplitude increases with the increase of the temperature, as shown in (b); and the Brillouin spectrum spectral width does not change with the temperature, as shown in (c); which indicates that the temperature is positively correlated with the Brillouin frequency shift and the amplitude, and is not correlated with the spectral width.

[0073] The strain test of the to-be-tested optical fiber is performed by using the above device, and the relationship between the Brillouin central frequency shift, amplitude and spectral width and the strain change of the optical fiber is analyzed, as shown in FIG. 3. Figure 3 When the sensing optical fiber is locally stretched, the Brillouin spectrum central frequency moves to the right; the Brillouin spectrum amplitude decreases with the increase of the stretching amount; and the Brillouin spectrum spectral width increases with the increase of the stretching amount, which indicates that the strain is positively correlated with the Brillouin frequency shift and the spectral width, and is negatively correlated with the amplitude of the Brillouin spectrum.

[0074] It can be concluded that the formula (3) created by the frequency shift and the spectral width of the Brillouin scattering spectrum can be rewritten as:

[0075]

[0076] Therefore, based on the above temperature and strain demodulation principle, according to the three parameters of the Brillouin frequency shift, amplitude and spectral width, the artificial neural network (Neural Network, ANN) algorithm is used to realize the high-precision decoupling of the fiber Brillouin distributed temperature and strain, as shown in FIG. 4. The process specifically includes: Figure 4

[0077] Step one: training data sample collection stage; specifically including:

[0078] S101: different temperatures are applied to the to-be-tested optical fiber, and the Brillouin scattering spectrum of each point of the optical fiber is obtained by using the fiber Brillouin distributed temperature and strain decoupling sensing device, and the temperature is taken as the label value (T i , 0) of the ANN algorithm;

[0079] ​S102: The Brillouin scattering spectrum of each point is fitted by using a Lorentz fitting algorithm to obtain Brillouin center frequency shift data, spectral width, and amplitude, which are used as the input parameter set of the ANN;

[0080] S103: Different stresses are applied to the optical fiber to be measured, and the Brillouin scattering spectrum of each point of the optical fiber is obtained by using the optical fiber Brillouin distributed temperature and strain decoupling sensing device, and the strain is used as the label value (S i ) of the ANN algorithm;

[0081] S104: The Brillouin scattering spectrum of each point is fitted by using a Lorentz fitting algorithm to obtain Brillouin center frequency shift data, spectral width, and amplitude, which are used as the input parameter set of the ANN;

[0082] S105: Different temperatures and stresses are applied to the optical fiber to be measured, and the Brillouin scattering spectrum of each point of the optical fiber is obtained by using the optical fiber Brillouin distributed temperature and strain decoupling sensing device, and the temperature and strain are used as the label value (T i , S i ) of the ANN algorithm;

[0083] S106: The Brillouin scattering spectrum of each point is fitted by using a Lorentz fitting algorithm to obtain Brillouin center frequency shift data, spectral width, and amplitude, which are used as the input parameter set of the ANN;

[0084] S107: The data sets obtained in S101-S106 are preprocessed, unreasonable data is removed, and normalization processing is performed to obtain ANN learning sample data, and the sample data is divided into a training set, a validation set, and a test set.

[0085] Step two: ANN learning model training phase; specifically including:

[0086] S201: The Brillouin center frequency shift, spectral width, and amplitude in the training set obtained in step one are used as inputs, and the temperature and strain data (T i , S i ) are used as labels to input the ANN deep learning model for supervised learning training, and the validation set is used for verification;

[0087] S202: The input and output parameters of the test set are used to test the trained ANN deep learning network, the deviation between the actual output parameters and the output parameters of the test set is compared, and the maximum error and the average absolute error are calculated;

[0088] S203: Determine whether the error is greater than a preset threshold;

[0089] S204: If the error exceeds the preset range, adjust the ANN deep learning model structure and return to S202;

[0090] S205: If the bias is within the preset range, the training is complete, and a temperature and strain simultaneous measurement model based on ANN deep learning is obtained.

[0091] Step three: temperature and strain data measurement stage; specifically comprising:

[0092] S301: Obtain the Brillouin scattering spectrum of the optical fiber to be measured by using the optical fiber Brillouin distributed temperature and strain decoupling sensing device;

[0093] S302: The Brillouin scattering spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain the Brillouin center frequency shift, spectral width and amplitude data of each point;

[0094] S303: The Brillouin center frequency shift, spectral width and amplitude data here are input into the model obtained in step two as the input of ANN, and the temperature and strain values are obtained.

[0095] Another embodiment of the present application provides an optical fiber Brillouin distributed temperature and strain decoupling method based on ANN algorithm, which is realized based on the optical fiber Brillouin distributed temperature and strain decoupling device in the above embodiment, and comprises the following steps:

[0096] The data acquisition module fusing OTDR and BOTDR is used to collect the Brillouin gain spectrum of the optical fiber to be measured;

[0097] The Brillouin gain spectrum of the optical fiber to be measured is fitted by using the Lorentz fitting algorithm to obtain the corresponding Brillouin center frequency shift, spectral width and amplitude;

[0098] The Brillouin center frequency shift, spectral width and amplitude are input into the pre-trained detection model based on ANN algorithm to obtain the temperature and strain values of the optical fiber to be measured.

[0099] In the embodiment, preferably, the Lorentz data fitting is performed according to the following formula to obtain the Brillouin center frequency shift, spectral width and amplitude:

[0100]

[0101] Wherein, g0 represents the Brillouin gain, the amplitude P=g0 2 ;v b represents the Brillouin center frequency shift; Δν represents the full width at half maximum of the Brillouin gain spectrum, that is, the spectral width.

[0102] In the embodiment, preferably, the pre-training process of the detection model based on ANN algorithm comprises:

[0103] Step one, obtain the training data set and the test data set, and preprocess the data in the data set; specifically including:

[0104] Different temperatures are applied to the optical fiber, the Brillouin gain spectrum of each point of the optical fiber is collected by using the data acquisition module, and the temperature is taken as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different temperatures, which are taken as a first input parameter set;

[0105] Different stresses are applied to the optical fiber, the Brillouin gain spectrum of each point of the optical fiber is collected by using the data acquisition module, and the strain is taken as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different stresses, which are taken as a second input parameter set;

[0106] Different temperatures and stresses are applied to the optical fiber at the same time, the Brillouin gain spectrum of each point of the optical fiber is collected by using the data acquisition module, and the temperature and the strain are taken as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different temperatures and stresses, which are taken as a third input parameter set;

[0107] The first input parameter set, the second input parameter set and the third input parameter set are combined and divided into a training data set and a test data set;

[0108] The preprocessing of the data in the data set includes: eliminating unreasonable data and performing normalization processing;

[0109] Step two, input the training data set into the detection model based on the ANN algorithm for supervised learning training;

[0110] Step three, input the test data set into the detection model based on the ANN algorithm trained to test, compare the deviation between the actual output parameter and the output parameter of the test data set, and calculate the maximum error and the average absolute error;

[0111] Step four, judge whether the error is greater than the preset threshold, if greater, adjust the structure of the detection model based on the ANN algorithm and repeat step three until the error is not greater than the preset threshold to stop training, and obtain the trained detection model based on the ANN algorithm.

[0112] The function of the optical fiber Brillouin distributed temperature and strain decoupling method based on the ANN algorithm described in this embodiment can be explained by the foregoing optical fiber Brillouin distributed temperature and strain decoupling device based on the ANN algorithm, therefore, the part not described in detail in this embodiment is referred to the above device embodiment.

[0113] It should be noted that, although several units, modules or sub-modules are mentioned in the foregoing detailed description, such a division is merely exemplary and not mandatory. Indeed, according to an embodiment of the application, the features and functionalities of two or more modules described above can be embodied in one module. Conversely, the features and functionalities of one module described above can be further divided into modules.

[0114] Furthermore, although the operations of the method of the application are described in a particular, sequential order, this order is not meant to be a limitation and is not intended to imply that the described operations are the only ones that can be performed or that they should be performed in the order described. Additionally or alternatively, certain of the steps can be handled in parallel, separate steps can be combined, and / or certain steps can be handled chronologically before, after, or in parallel with other steps.

[0115] While the spirit and principles of the application have been described with reference to several specific embodiments, it is to be understood that the application is not limited to the precise embodiments disclosed and that various modifications can be made in aspects without departing from the scope of the application. The application is therefore not limited to the exact construction and arrangement of parts as described and as illustrated in the attached drawings.

Claims

1. An ANN algorithm-based fiber Brillouin distributed temperature and strain decoupling device, characterized in that, The application relates to a Brillouin gain spectrum data acquisition and processing device, which comprises a data acquisition module and a data processing module. The light signal output end of the narrow-line-width laser (2) is communicated with the light signal input end of the first optical fiber coupler (3), the No.1 light signal output end of the first optical fiber coupler (3) and the light signal output end of the broadband laser (1) are simultaneously communicated with the light signal input end of the first dense wavelength division multiplexer (4) respectively, the light signal output end of the first dense wavelength division multiplexer (4) is communicated with the light signal input end of the first electro-optic modulator (5), the light signal output end of the first electro-optic modulator (5) is communicated with the light signal input end of the doped fiber amplifier (6), the light signal output end of the doped fiber amplifier (6) is communicated with the random polarization scrambler (7), the output end of the random polarization scrambler (7) is communicated with the No.1 light signal port (8-1) of the first optical circulator (8), the No.2 light signal port (8-2) of the first optical circulator (8) is communicated with the measured optical fiber (9); The No.3 light signal port (8-3) of the first optical circulator (8) is communicated with the input end of the second dense wavelength division multiplexer (10), the No.1 output end of the second dense wavelength division multiplexer (10) is communicated with the light signal input end of the photodetector (11), the electric signal output end of the photodetector (11) is communicated with the No.1 electric signal data acquisition end of the data acquisition card (13); the No.2 output end of the second dense wavelength division multiplexer (10) is communicated with the No.1 light signal port (14-1) of the second optical circulator (14), the No.2 light signal port (14-2) of the second optical circulator (14) is communicated with the fiber grating filter (15), and the No.3 light signal port (14-3) of the second optical circulator (14) is communicated with the No.1 input end of the second optical fiber coupler (18); The No.2 light signal output end of the first optical fiber coupler (3) and the signal output end of the microwave source (16) are simultaneously communicated with the light signal input end and the microwave signal loading end of the second electro-optic modulator (17) respectively, the light signal output end of the second electro-optic modulator (17) is communicated with the No.2 input end of the second optical fiber coupler (18), the two output ends of the second optical fiber coupler (18) are simultaneously communicated with the two input ports of the balanced photodetector (19) respectively, the electric signal output end of the balanced photodetector (19) is communicated with the signal input end of the detector (20), and the signal output end of the detector (20) is simultaneously communicated with the No.2 electric signal data acquisition end of the data acquisition card (13). The microwave signal output end of the arbitrary function generator (12) is in communication with the microwave signal loading end of the first electro-optical modulator (5) and the trigger signal input end of the data acquisition card (13) at the same time; The data acquisition card (13) outputs the Brillouin gain spectrum of the to-be-tested optical fiber (9) to the data processing module.

2. The fiber Brilouin distributed temperature and strain decoupling device based on ANN algorithm according to claim 1, characterized in that, The data processing module comprises a data fitting submodule and a data decoupling submodule; the data fitting submodule is configured to perform data fitting on the Brillouin gain spectrum of the to-be-tested optical fiber by using a Lorentz fitting algorithm to obtain the corresponding Brillouin center frequency shift, spectral width and amplitude; The data decoupling submodule is configured to input the Brillouin center frequency shift, spectral width and amplitude into a pre-trained detection model based on an ANN algorithm to obtain the temperature and strain value of the to-be-tested optical fiber.

3. The fiber Brilouin distributed temperature and strain decoupling device based on ANN algorithm according to claim 2, characterized in that, The data fitting submodule performs Lorentz data fitting according to the following formula to obtain the Brillouin center frequency shift, spectral width and amplitude: ； wherein, represents the Brillouin gain, the magnitude ; represents the Brillouin central frequency shift; represents the full width at half maximum of the Brillouin gain spectrum, i.e. the spectral width.

4. The fiber Brilouin distributed temperature and strain decoupling device based on ANN algorithm according to claim 3, characterized in that, The pre-training process of the detection model based on the ANN algorithm in the data decoupling submodule comprises: Step one, obtaining a training data set and a test data set, and preprocessing the data in the data set; specifically including: applying different temperatures to the optical fiber, using the data acquisition module to collect the Brillouin gain spectrum of each point of the optical fiber, and taking the temperature as the corresponding label value; using the Lorentz fitting algorithm to perform data fitting on the Brillouin gain spectrum of each point to obtain multiple sets of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different temperatures, which are taken as a first input parameter set; applying different stresses to the optical fiber, using the data acquisition module to collect the Brillouin gain spectrum of each point of the optical fiber, and taking the strain as the corresponding label value; using the Lorentz fitting algorithm to perform data fitting on the Brillouin gain spectrum of each point to obtain multiple sets of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different stresses, which are taken as a second input parameter set; applying different temperatures and stresses to the optical fiber at the same time, using the data acquisition module to collect the Brillouin gain spectrum of each point of the optical fiber, and taking the temperature and strain as the corresponding label value; using the Lorentz fitting algorithm to perform data fitting on the Brillouin gain spectrum of each point to obtain multiple sets of Brillouin center frequency shifts, spectral widths and amplitudes corresponding to different temperatures and stresses, which are taken as a third input parameter set; combining the first input parameter set, the second input parameter set and the third input parameter set, and dividing them into a training data set and a test data set; the preprocessing of the data in the data set includes eliminating unreasonable data and performing normalization processing; Step two, inputting the training data set into the detection model based on the ANN algorithm for supervised learning training; Step three, inputting the test data set into the detection model based on the ANN algorithm obtained by training for testing, comparing the deviation between the actual output parameter and the output parameter of the test data set, and calculating the maximum error and the average absolute error; Step four, judging whether the error is greater than a preset threshold, if yes, adjusting the structure of the detection model based on the ANN algorithm and repeating step three until the error is not greater than the preset threshold to stop training, and obtaining the trained detection model based on the ANN algorithm.

5. The ANN algorithm based fiber optic Brillouin distributed temperature and strain decoupling apparatus according to any one of claims 1-4, characterized in that, The broadband light source output power of the broadband laser (1) is 20 mW, the wavelength is 1550 nm, and the line width is 100 GHz; the narrow line width light source output power of the narrow line width laser (2) is 40 mW, the wavelength is 1550 nm, and the line width is 1 MHz; the coupling ratio of the first fiber coupler (3) is 90:10 or 80:20 or 70:30, and the coupling ratio of the second fiber coupler (18) is 50:50; the center wavelength of the fiber grating filter (15) is 1550 nm, and the bandwidth is 0.1 nm; the detection bandwidth of the photoelectric balance detector (19) is 200 MHz-500 MHz.

6. A method for decoupling temperature and strain based on an ANN algorithm for fiber Brillouin distributed temperature and strain, characterized in that, The method is implemented based on the ANN algorithm-based fiber Brillouin distributed temperature and strain decoupling device in claim 1, and the method comprises the following steps: Brillouin gain spectrum of the fiber to be measured is collected by the data acquisition module integrating OTDR and BOTDR; Brillouin center frequency shift, spectrum width and amplitude corresponding to the Brillouin gain spectrum of the fiber to be measured are obtained by using a Lorentz fitting algorithm for data fitting; The Brillouin center frequency shift, spectrum width and amplitude are input into a pre-trained detection model based on an ANN algorithm to obtain the temperature and strain value of the fiber to be measured.

7. The method of claim 6, wherein the method is based on an ANN algorithm. Lorentz data fitting is performed according to the following formula to obtain the Brillouin center frequency shift, spectrum width and amplitude: ； wherein, represents the Brillouin gain, the magnitude ; represents the Brillouin central frequency shift; represents the full width at half maximum of the Brillouin gain spectrum, i.e. the spectral width.

8. The method of claim 7, wherein the method is based on an ANN algorithm. The pre-training process of the detection model based on the ANN algorithm comprises: Step one, obtaining a training data set and a test data set, and pre-processing data in the data set; specifically comprising: different temperatures are applied to the fiber, and the Brillouin gain spectrum of each point of the fiber is collected by the data acquisition module, and the temperature is used as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectrum widths and amplitudes corresponding to different temperatures, which are used as a first input parameter set; different stresses are applied to the fiber, and the Brillouin gain spectrum of each point of the fiber is collected by the data acquisition module, and the strain is used as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectrum widths and amplitudes corresponding to different stresses, which are used as a second input parameter set; different temperatures and stresses are applied to the fiber simultaneously, and the Brillouin gain spectrum of each point of the fiber is collected by the data acquisition module, and the temperature and strain are used as the corresponding label value; the Brillouin gain spectrum of each point is fitted by using the Lorentz fitting algorithm to obtain a plurality of groups of Brillouin center frequency shifts, spectrum widths and amplitudes corresponding to different temperatures and stresses, which are used as a third input parameter set; The first input parameter set, the second input parameter set and the third input parameter set are combined and divided into a training data set and a test data set; The pre-processing of the data in the data set comprises: eliminating unreasonable data and performing normalization processing; Step two, inputting the training data set into the detection model based on the ANN algorithm for supervised learning training; Step three, input the test data set into the trained detection model based on ANN algorithm for testing, compare the deviation between the actual output parameters and the output parameters of the test data set, and calculate the maximum error and the average absolute error; Step four, judge whether the error is greater than the preset threshold, if greater than, adjust the structure of the detection model based on ANN algorithm and repeat step three until the error is not greater than the preset threshold to stop training, and obtain the trained detection model based on ANN algorithm.

Images