Roadway deformation monitoring system based on distributed optical fiber sensors

Abstract

The invention belongs to the field of optical fiber sensors, and particularly discloses a roadway deformation monitoring system based on distributed optical fiber sensors, which comprises a signal acquisition module, a signal processing module and a roadway deformation monitoring module. According to the invention, by arranging a reference interferometer and a sensing interferometer, carrying out mathematical modeling on phase noise, introducing an auxiliary frequency modulation technology, extracting phase modulation depth by using orthogonal detection, and dynamically calculating the ratio of the noise to the optical path difference of the interferometer, the dynamic elimination of the noise in a sensing signal is realized, and the change of environmental noise is tracked in real time; the method comprises the following steps: denoising a phase signal, converting the denoised phase signal into strain component information through a signal demodulator, generating a time-frequency spectrogram, constructing a roadway deformation detection model, introducing a k-NN algorithm to replace Softmax to classify feature vectors, and realizing millisecond-level real-time monitoring of roadway deformation through deep feature learning of a convolutional neural network and local neighborhood classification advantages of the k-NN algorithm.

Description

Technical Field

[0001] The present invention relates to the field of optical fiber sensors, and in particular to a tunnel deformation monitoring system based on distributed optical fiber sensors. Background Art

[0002] Tunnel deformation monitoring refers to the use of various measurement technologies and instruments to conduct long-term, systematic, and continuous observations of changes in geometric parameters such as the shape, size, and position of underground tunnels over time. Through deformation monitoring, abnormal increases in deformation rates or deformation exceeding safety thresholds can be detected in a timely manner, and early warning signals can be issued in advance so that appropriate safety measures can be taken. In the complex construction environment of mine tunnels, factors such as electromagnetic interference and mechanical vibration can cause severe phase noise in the optical signals collected by fiber optic sensors. Traditional denoising methods are unable to effectively eliminate dynamic noise, resulting in the monitoring data being unable to accurately reflect the actual deformation of the tunnel and even potentially causing safety accidents due to misjudgment. Traditional monitoring technologies usually use manual analysis or simple algorithms to process optical signals. They are unable to quickly demodulate multi-dimensional strain components such as circumferential, radial, and axial strains, and have difficulty capturing subtle changes in tunnel deformation. By the time deformation reaches the dangerous threshold, the optimal time for disposal has often been missed. Summary of the Invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a tunnel deformation monitoring system based on distributed optical fiber sensors. In view of the complex construction environment of mine tunnels, electromagnetic interference, mechanical vibration and other factors will cause serious phase noise in the optical signals collected by optical fiber sensors. Traditional denoising methods are difficult to effectively eliminate dynamic noise, resulting in the monitoring data being unable to accurately reflect the actual deformation of the tunnel, and may even cause safety accidents due to misjudgment. This solution sets a reference interferometer and a sensor interferometer, mathematically models the phase noise, introduces auxiliary frequency modulation technology to generate a calibration phase signal, extracts the phase modulation depth through orthogonal detection, and dynamically calculates the ratio of noise to the optical path difference of the interferometer to achieve dynamic elimination of noise in the sensor signal. It can track environmental noise changes in real time, accurately eliminate various interference signals, and ensure that the system is collected. The collected optical signals truly reflect the deformation characteristics of the tunnel; traditional monitoring technology usually uses manual analysis or simple algorithms to process optical signals, which cannot quickly demodulate multi-dimensional strain components such as circumferential, radial, and axial strains, and it is difficult to capture subtle changes in tunnel deformation. When the deformation reaches the dangerous threshold, the best time for disposal is often missed. The solution uses a signal demodulator to convert the denoised phase signal into strain component information, generate a time-frequency spectrum, and construct a tunnel deformation detection model. At the same time, the k-NN algorithm is introduced to replace Softmax to classify feature vectors, which can automatically extract deformation features in the time-frequency spectrum. Through the deep feature learning of convolutional neural networks and the local neighborhood classification advantages of the k-NN algorithm, millisecond-level real-time monitoring of tunnel deformation is achieved, and an alarm is issued in time at the early stage of tunnel deformation, buying precious time for personnel evacuation and engineering reinforcement.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides a tunnel deformation monitoring system based on a distributed optical fiber sensor, which specifically includes a signal acquisition module, a signal processing module and a tunnel deformation monitoring module;

[0005] The signal acquisition module embeds the distributed optical fiber sensor into the mine tunnel wall through a pipe to collect the optical signal of the optical fiber sensor;

[0006] The signal processing module sets a reference phase for the optical signal of the optical fiber sensor collected by the signal collection module, performs adaptive denoising on the optical signal, and obtains a denoised phase signal;

[0007] The tunnel deformation monitoring module converts the denoised phase signal into strain component information, constructs a tunnel deformation detection model through a convolutional neural network, and introduces the k-NN algorithm to replace Softmax in the convolutional neural network to achieve real-time monitoring of tunnel deformation.

[0008] The signal processing module processes the optical signal using an adaptive denoising method, wherein the adaptive denoising method specifically includes the following steps:

[0009] Step A1: Set up a sensor interferometer to measure the lane light signal in real time, set up a reference interferometer to measure the signal denoising reference phase, and perform mathematical modeling of the phase noise. The formula used is as follows: ; ;

[0010] Where, is the phase difference between the reference interferometer and the sensing interferometer, is the fiber refractive index, is the wavelength of the light source under vacuum conditions, is the physical path difference between the reference interferometer and the sensing interferometer, is the light frequency of the light source, is the speed of light in a vacuum, is the phase noise caused by the light source, is the frequency drift of the light source;

[0011] Step A2: Introduce auxiliary frequency modulation in the sensing interferometer. The formula used is as follows: ;

[0012] Where, is the light source frequency modulation function, is the maximum frequency deviation of the light source, is the auxiliary modulation frequency, is the current time step;

[0013] Step A3: Generate a calibration phase signal in the sensor interferometer through auxiliary frequency modulation. The formula used is as follows: ;

[0014] Where, is the additional phase signal caused by modulation, is the phase modulation depth;

[0015] Step A4: Perform orthogonal detection on the auxiliary modulation frequency to extract the phase modulation depth, and calculate the ratio of the noise to the optical path difference of the interferometer in real time. The formula used is as follows: ;

[0016] Where, is the ratio of the noise to the optical path difference of the interferometer, is the modulation depth of the sensing interferometer, is the modulation depth of the reference interferometer, is the optical path difference between the noise and the sensor interferometer, is the optical path difference between the noise and reference interferometers;

[0017] Step A5: Dynamically eliminate the noise in the sensor interferometer signal to obtain a denoised phase signal. The formula used is as follows: ;

[0018] Where, is the total output phase after noise cancellation, is the total output phase of the sensor interferometer, is the total output phase of the reference interferometer.

[0019] The tunnel deformation monitoring module uses a hybrid method of introducing k-NN into a convolutional neural network to monitor the tunnel deformation. The hybrid method of introducing k-NN into a convolutional neural network specifically includes the following steps:

[0020] Step B1: Use a signal demodulator to convert the denoised phase signal into strain component information and generate a time-frequency spectrum. The formula used is as follows: ; ; ; ;

[0021] Where, is the hoop strain, is the radial strain, is the axial strain, is the radial displacement of the pipe, is the pipe radius, is the pipe deformation function, is the elastic modulus of the pipe, is Poisson's ratio, The internal pressure of the pipe, and are the inner and outer radii of the pipe, respectively;

[0022] Step B2: Establish and initialize a convolutional neural network as a roadway deformation detection model, wherein the convolutional neural network includes two convolutional layers, two subsampling layers, and one fully connected layer;

[0023] Step B3: Introduce the k-NN algorithm to replace Softmax in the convolutional neural network, classify the feature vectors output by the fully connected layer, and optimize the k-NN algorithm. Specifically, this includes using a genetic algorithm to dynamically select the optimal k value and using Euclidean distance as the distance metric.

[0024] Step B4: Input the time-frequency spectrum into the convolutional neural network for processing, output the strain level and confidence level, select the strain level with the highest confidence level to calculate the final deformation level, and preset the deformation threshold. When the final deformation level exceeds the deformation threshold, the system will issue an alarm message and organize relevant personnel to take corresponding measures.

[0025] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0026] (1) In view of the technical problem that in the complex construction environment of mine tunnels, electromagnetic interference, mechanical vibration and other factors will cause serious phase noise in the optical signals collected by optical fiber sensors, and traditional denoising methods are difficult to effectively eliminate dynamic noise, resulting in the monitoring data being unable to accurately reflect the actual deformation of the tunnel, and may even cause safety accidents due to misjudgment. This scheme sets up a reference interferometer and a sensor interferometer, mathematically models the phase noise, introduces auxiliary frequency modulation technology to generate a calibration phase signal, extracts the phase modulation depth through orthogonal detection, and dynamically calculates the ratio of noise to the optical path difference of the interferometer to achieve dynamic elimination of noise in the sensor signal. It can track the changes in environmental noise in real time, accurately eliminate various interference signals, and ensure that the collected optical signal truly reflects the deformation characteristics of the tunnel;

[0027] (2) In view of the technical problem that traditional monitoring technology usually uses manual analysis or simple algorithms to process optical signals, it is unable to quickly demodulate the multi-dimensional strain components such as circumferential, radial, and axial strains, and it is difficult to capture the subtle changes in tunnel deformation. When the deformation reaches the dangerous threshold, the best time for disposal is often missed. The solution converts the denoised phase signal into strain component information through a signal demodulator, generates a time-frequency spectrum, and constructs a tunnel deformation detection model. At the same time, the k-NN algorithm is introduced to replace Softmax to classify the feature vectors, which can automatically extract the deformation features in the time-frequency spectrum. Through the deep feature learning of the convolutional neural network and the local neighborhood classification advantage of the k-NN algorithm, the millisecond-level real-time monitoring of tunnel deformation is achieved, and an alarm is issued in time at the early stage of tunnel deformation, which buys valuable time for personnel evacuation and engineering reinforcement. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a module connection diagram of a tunnel deformation monitoring system based on distributed optical fiber sensors provided by the present invention.

[0029] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0031] Example 1, see Figure 1 The present invention provides a tunnel deformation monitoring system based on a distributed optical fiber sensor, which specifically includes a signal acquisition module, a signal processing module and a tunnel deformation monitoring module;

[0032] The signal acquisition module embeds the distributed optical fiber sensor into the mine tunnel wall through a pipe to collect the optical signal of the optical fiber sensor;

[0033] The signal processing module sets a reference phase for the optical signal of the optical fiber sensor collected by the signal collection module, performs adaptive denoising on the optical signal, and obtains a denoised phase signal;

[0034] The tunnel deformation monitoring module converts the denoised phase signal into strain component information, constructs a tunnel deformation detection model through a convolutional neural network, and introduces the k-NN algorithm to replace Softmax in the convolutional neural network to achieve real-time monitoring of tunnel deformation.

[0035] Example 2, see Figure 1 This embodiment is based on the above embodiment. The signal processing module processes the optical signal using an adaptive denoising method. The adaptive denoising method specifically includes the following steps:

[0036] Step A1: Set up a sensor interferometer to measure the lane light signal in real time, set up a reference interferometer to measure the signal denoising reference phase, and perform mathematical modeling of the phase noise. The formula used is as follows: ; ;

[0037] Where, is the phase difference between the reference interferometer and the sensing interferometer, is the fiber refractive index, is the wavelength of the light source under vacuum conditions, is the physical path difference between the reference interferometer and the sensing interferometer, is the light frequency of the light source, is the speed of light in a vacuum, is the phase noise caused by the light source, is the frequency drift of the light source;

[0038] Step A2: Introduce auxiliary frequency modulation in the sensing interferometer. The formula used is as follows: ;

[0039] Where, is the light source frequency modulation function, is the maximum frequency deviation of the light source, is the auxiliary modulation frequency, is the current time step;

[0040] Step A3: Generate a calibration phase signal in the sensor interferometer through auxiliary frequency modulation. The formula used is as follows: ;

[0041] Where, is the additional phase signal caused by modulation, is the phase modulation depth;

[0042] Step A4: Perform orthogonal detection on the auxiliary modulation frequency to extract the phase modulation depth, and calculate the ratio of the noise to the optical path difference of the interferometer in real time. The formula used is as follows: ;

[0043] Where, is the ratio of the noise to the optical path difference of the interferometer, is the modulation depth of the sensing interferometer, is the modulation depth of the reference interferometer, is the optical path difference between the noise and the sensor interferometer, is the optical path difference between the noise and reference interferometers;

[0044] Step A5: Dynamically eliminate the noise in the sensor interferometer signal to obtain a denoised phase signal. The formula used is as follows: ;

[0045] Where, is the total output phase after noise cancellation, is the total output phase of the sensor interferometer, is the total output phase of the reference interferometer.

[0046] Example 3, see Figure 1 This embodiment is based on the above embodiment. The tunnel deformation monitoring module uses a hybrid method of introducing k-NN into a convolutional neural network to monitor the tunnel deformation. The hybrid method of introducing k-NN into a convolutional neural network specifically includes the following steps:

[0047] Step B1: Use a signal demodulator to convert the denoised phase signal into strain component information and generate a time-frequency spectrum. The formula used is as follows: ; ; ; ;

[0048] Where, is the hoop strain, is the radial strain, is the axial strain, is the radial displacement of the pipe, is the pipe radius, is the pipe deformation function, is the elastic modulus of the pipe, is Poisson's ratio, The internal pressure of the pipe, and are the inner and outer radii of the pipe, respectively;

[0049] Step B2: Establish and initialize a convolutional neural network as a roadway deformation detection model, wherein the convolutional neural network includes two convolutional layers, two subsampling layers, and one fully connected layer;

[0050] Step B3: Introduce the k-NN algorithm to replace Softmax in the convolutional neural network, classify the feature vectors output by the fully connected layer, and optimize the k-NN algorithm. Specifically, this includes using a genetic algorithm to dynamically select the optimal k value and using Euclidean distance as the distance metric.

[0051] Step B4: Input the time-frequency spectrum into the convolutional neural network for processing, output the strain level and confidence level, select the strain level with the highest confidence level to calculate the final deformation level, and preset the deformation threshold. When the final deformation level exceeds the deformation threshold, the system will issue an alarm message and organize relevant personnel to take corresponding measures.

[0052] Example 4, see Figure 1 This embodiment is based on the above embodiment. In the third embodiment, the k-NN algorithm is optimized, specifically including using a genetic algorithm to dynamically select the optimal k value. In practical applications, the average accuracy is the highest when k is 100.

[0053] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0054] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

[0055] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A tunnel deformation monitoring system based on distributed optical fiber sensors, characterized in that: Specifically includes signal acquisition module, signal processing module and tunnel deformation monitoring module; The signal acquisition module embeds the distributed optical fiber sensor into the mine tunnel wall through a pipe to collect the optical signal of the optical fiber sensor; The signal processing module sets a reference phase for the optical signal of the optical fiber sensor collected by the signal collection module, performs adaptive denoising on the optical signal, and obtains a denoised phase signal; The tunnel deformation monitoring module converts the denoised phase signal into strain component information, constructs a tunnel deformation detection model through a convolutional neural network, and introduces the k-NN algorithm to replace Softmax in the convolutional neural network to achieve real-time monitoring of tunnel deformation.

2. A tunnel deformation monitoring system based on distributed optical fiber sensors according to claim 1, characterized in that: The signal processing module processes the optical signal using an adaptive denoising method, wherein the adaptive denoising method specifically includes the following steps: Step A1: Set up a sensor interferometer to measure the lane light signal in real time, set up a reference interferometer to measure the signal denoising reference phase, and perform mathematical modeling of the phase noise. The formula used is as follows: ； ； Where, is the phase difference between the reference interferometer and the sensing interferometer, is the fiber refractive index, is the wavelength of the light source under vacuum conditions, is the physical path difference between the reference interferometer and the sensing interferometer, is the light frequency of the light source, is the speed of light in a vacuum, is the phase noise caused by the light source, is the frequency drift of the light source; Step A2: Introduce auxiliary frequency modulation in the sensing interferometer. The formula used is as follows: ； Where, is the light source frequency modulation function, is the maximum frequency deviation of the light source, is the auxiliary modulation frequency, is the current time step; Step A3: generating a calibration phase signal in the sensing interferometer by auxiliary frequency modulation; Step A4: performing orthogonal detection on the auxiliary modulation frequency to extract the phase modulation depth, and calculating the ratio of the noise to the optical path difference of the interferometer in real time; Step A5: Dynamically eliminate the noise in the sensor interferometer signal to obtain a denoised phase signal.

3. The tunnel deformation monitoring system based on distributed optical fiber sensors according to claim 1 is characterized in that: The tunnel deformation monitoring module uses a hybrid method of introducing k-NN into a convolutional neural network to monitor the tunnel deformation. The hybrid method of introducing k-NN into a convolutional neural network specifically includes the following steps: Step B1: Use a signal demodulator to convert the denoised phase signal into strain component information and generate a time-frequency spectrum. The formula used is as follows: ； ； ； ； Where, is the hoop strain, is the radial strain, is the axial strain, is the radial displacement of the pipe, is the pipe radius, is the pipe deformation function, is the elastic modulus of the pipe, is Poisson's ratio, The internal pressure of the pipe, and are the inner and outer radii of the pipe, respectively; Step B2: Establish and initialize a convolutional neural network as a roadway deformation detection model, wherein the convolutional neural network includes two convolutional layers, two subsampling layers, and one fully connected layer; Step B3: Introduce the k-NN algorithm to replace Softmax in the convolutional neural network, classify the feature vectors output by the fully connected layer, and optimize the k-NN algorithm. Specifically, this includes using a genetic algorithm to dynamically select the optimal k value and using Euclidean distance as the distance metric. Step B4: Input the time-frequency spectrum into the convolutional neural network for processing, output the strain level and confidence level, select the strain level with the highest confidence level to calculate the final deformation level, and preset the deformation threshold. When the final deformation level exceeds the deformation threshold, the system will issue an alarm message and organize relevant personnel to take corresponding measures.

Images