Low-altitude and ground unmanned equipment monitoring method and system based on optical fiber sensing

Abstract

The invention relates to the technical field of optical fiber sensing, in particular to a low-altitude and ground unmanned equipment monitoring method and system based on optical fiber sensing, and the method comprises the following steps: laying an anti-interference fading DAS system in a monitoring area to obtain a signal dynamic component and a composite signal in the monitoring area; converting the signal dynamic component into a dynamic characteristic matrix; performing noise suppression on the dynamic feature matrix by using a noise suppression model to obtain a clean dynamic feature map; positioning a vibration source by using the clean dynamic feature map, and further obtaining a vibration signal spectrogram in combination with a vibration signal generation model; and constructing a vibration source classification and recognition model by using a convolutional neural network, and further performing classification and recognition on a vibration source by using the vibration signal spectrogram and the vibration source classification and recognition model. According to the method, the signal stability is improved through the anti-interference fading DAS system, and high-precision monitoring and identification of low-altitude and ground unmanned equipment can be realized in combination with multi-strategy noise suppression, CycleGAN signal extraction and convolutional neural network classification.

Description

Technical Field

[0001] The present invention relates to the field of optical fiber sensing technology, and in particular to a method and system for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing. Background Art

[0002] With the rapid development of the low-altitude economy and unmanned equipment technology, the application of low-altitude drones, unmanned vehicles, and other unmanned equipment in logistics, security, agriculture, and other fields is becoming increasingly widespread. However, they also pose a potential threat to public safety. Therefore, achieving high-precision, all-weather monitoring of low-altitude and ground-based unmanned equipment has become a key requirement for current technological development. Distributed fiber optic acoustic sensing (DAS) systems can achieve real-time positioning and signal acquisition of vibration events by detecting backward Rayleigh scattered light signals in optical fibers, providing basic technical support for unmanned equipment monitoring. However, existing monitoring methods based on fiber optic sensing still face technical bottlenecks such as poor signal quality, strong noise interference, and low vibration source classification accuracy in complex environments, making it difficult to meet the needs of high-reliability monitoring.

[0003] In existing technologies, unmanned equipment monitoring methods based on DAS systems typically employ the following process: acquiring vibration signals through the DAS system, performing signal noise reduction processing using algorithms such as wavelet transform and empirical mode decomposition, extracting time-domain or frequency-domain features, and finally implementing vibration source classification based on machine learning models. However, such methods have significant drawbacks: First, traditional DAS systems are affected by random interference fading caused by Rayleigh scattering, which can easily create detection blind spots in optical fibers, leading to vibration signal loss or positioning errors; second, noise reduction algorithms such as wavelet transform rely on fixed basis functions, which have limited ability to suppress non-stationary noise, and residual noise can interfere with subsequent feature extraction; third, the characteristic expression dimension of one-dimensional vibration signals is single, making it difficult to fully capture the spatial distribution and pattern differences of vibration sources, resulting in classification models easily confusing similar vibration sources; fourth, existing methods often use global signal processing when classifying and locating vibration signals, which can easily lead to computational redundancy, and environmental noise and non-target vibration source signals can reduce classification accuracy. Summary of the Invention

[0004] In view of the defects in the prior art, the present invention provides a method and system for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing.

[0005] In order to achieve the above-mentioned objectives, in a first aspect, the present invention provides a method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing, the method comprising the following steps: deploying an anti-interference fading DAS system in the monitoring area; using the anti-interference fading DAS system to obtain the signal dynamic components and composite signals in the monitoring area; converting the signal dynamic components into a dynamic feature matrix; using a noise suppression model to suppress the noise of the dynamic feature matrix to obtain a clean dynamic feature map; using the clean dynamic feature map to locate the vibration source, and then combining it with a vibration signal generation model to obtain a vibration signal spectrum map; using a convolutional neural network to construct a vibration source classification and recognition model, and then using the vibration signal spectrum map and the vibration source classification and recognition model to classify and identify the vibration source. The present invention improves signal stability through the anti-interference fading DAS system, and combines multi-strategy noise suppression, CycleGAN signal extraction and convolutional neural network classification to achieve high-precision monitoring and identification of low-altitude and ground unmanned equipment.

[0006] Optionally, deploying an anti-interference fading DAS system in the monitoring area includes the following steps: Determine the monitoring area; Planning the buried path of the sensing optical fiber within the monitoring area; The sensing optical fiber is buried according to the buried path, and the supporting DAS system equipment is installed.

[0007] Optionally, the step of using the anti-interference fading DAS system to obtain the signal dynamic component and the composite signal within the monitoring area comprises the following steps: The continuous light generated by the laser is modulated into quasi-orthogonal chirped pulses and injected into the sensing fiber, and the beat frequency signal of the backscattered Rayleigh light and the local oscillator light is obtained using a photodetector; The high and low frequency band signals of the quasi-orthogonal chirped pulse are extracted by matched filtering and unmatched filtering, and then the interference fading of the beat frequency signal is suppressed by rotating vector sum synthesis to obtain a synthetic signal; Performing dynamic detection on the synthesized signal to obtain a signal dynamic component, wherein the dynamic detection is time difference; Digital signal processing is performed on the dynamic component of the signal to obtain a composite signal containing a vibration signal.

[0008] This method modulates laser pulses into quasi-orthogonal chirped signals, combines matched filtering with unmatched filtering to separate high- and low-frequency signals, and effectively suppresses interference fading. It uses vector synthesis technology to compensate for random phase changes, solves the signal loss problem caused by fading in traditional DAS, significantly improves signal integrity, and thus improves the accuracy of the composite signal, which is conducive to improving the accuracy of monitoring low-altitude and ground unmanned equipment.

[0009] Optionally, converting the signal dynamic component into a dynamic feature matrix comprises the following steps: Determining the row index and column index of the dynamic feature matrix; The dynamic feature matrix is ​​constructed using the amplitude of the signal dynamic component according to the row index and the column index.

[0010] This method breaks through the limitation of the single characteristic dimension of traditional one-dimensional signal by converting the one-dimensional signal into a three-dimensional feature matrix of time-space-intensity, providing richer spatial pattern information for subsequent noise suppression and vibration source classification, thereby improving the accuracy of monitoring low-altitude and ground unmanned equipment.

[0011] Optionally, the step of using a noise suppression model to suppress noise on the dynamic feature matrix to obtain a clean dynamic feature map comprises the following steps: Constructing a noise suppression model of the dynamic feature matrix; The dynamic feature matrix is ​​input into the noise suppression model to obtain a clean dynamic feature map.

[0012] Optionally, the noise suppression model performs the following steps when running: generating an original solution set of a noise suppression model using a multi-strategy noise suppression scheme according to the dynamic feature matrix; According to each solution in the original solution set, a mutant individual is generated by differential vector perturbation, and then a test vector is generated by crossover operation; Select the mutant individuals and those with high fitness in the test vector as excellent solutions, and use the set of all excellent solutions as the solution set for the next iteration; When the maximum number of iterations is reached or the fitness value converges, the current optimal solution is output, and the optimal solution is reconstructed into the clean dynamic feature map.

[0013] This method combines multiple noise reduction schemes to generate an initial solution set, and then iteratively optimizes the solution set through mutation, crossover and selection operations, dynamically searches for the optimal noise suppression strategy, solves the residual noise problem of traditional fixed basis function noise reduction, significantly improves signal purity, and adapts to complex noise scenarios.

[0014] Optionally, generating an original solution set of a noise suppression model using a multi-strategy noise suppression solution according to the dynamic feature matrix comprises the following steps: Gray-scaling the dynamic feature matrix and performing noise reduction processing on the dynamic feature matrix using multiple noise suppression schemes, thereby obtaining a set of noise-reduced image sets; Each denoised image in the denoised image set is flattened into a one-dimensional vector, thereby obtaining an original solution set of the noise suppression model.

[0015] Optionally, the method of locating the vibration source by using the clean dynamic characteristic graph and then obtaining the vibration signal spectrum graph in combination with a vibration signal generation model comprises the following steps: Locating the vibration source using the clean dynamic characteristic map; intercepting the composite signal according to the result of locating the vibration source to obtain composite signal segments; CycleGAN is used to build a vibration signal generation model, and then a vibration signal spectrum is obtained based on the composite signal segment.

[0016] This method first locates the vibration source position, and then intercepts the composite signal at the corresponding position to obtain composite signal fragments. This not only alleviates the computational redundancy caused by global signal processing when classifying vibration sources, but also reduces classification errors caused by high-frequency noise. The vibration signal generation model constructed using CycleGAN is applicable to various types of noise and signals, and can solve the problem that traditional methods such as wavelet transform cannot effectively process non-stationary noise in vibration signals.

[0017] Optionally, the use of a convolutional neural network to construct a vibration source classification and identification model, and then using the vibration signal spectrum and the vibration source classification and identification model to classify and identify the vibration source, comprises the following steps: Construct a vibration signal spectrum dataset; Using the vibration signal spectrum dataset to train and validate a convolutional neural network to obtain a vibration source classification and recognition model; The vibration signal spectrum is input into the vibration source classification and identification model to achieve classification and identification of the vibration source.

[0018] This method utilizes the local pattern extraction capability of CNN for time-frequency features to fully capture the frequency components and temporal dynamics of different vibration sources, solves the problem of easy confusion in traditional one-dimensional feature classification, and improves the accuracy of vibration source classification.

[0019] In a second aspect, the present invention provides a fiber-optic sensing-based low-altitude and ground-based unmanned equipment monitoring system. The system comprises a data acquisition device, a data output device, a processor, and storage. The storage comprises a computer-readable storage medium storing a computer program comprising program instructions that, when executed by the processor, cause the processor to implement the fiber-optic sensing-based low-altitude and ground-based unmanned equipment monitoring method provided by the present invention. This system can improve the practicality of the method and facilitate its dissemination. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 Schematic diagram of the process of a method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to an embodiment of the present invention; Figure 2 A partial diagram of the slotting situation in the monitoring area according to an embodiment of the present invention; Figure 3 A partial diagram of trench backfilling in a monitoring area according to an embodiment of the present invention; Figure 4 A partially clean dynamic feature map obtained based on the noise suppression model according to an embodiment of the present invention; Figure 5 This is a dynamic feature map after noise reduction using wavelet transform according to an embodiment of the present invention; Figure 6 Schematic diagram of the framework of the low-altitude and ground unmanned equipment monitoring system based on optical fiber sensing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the present invention. In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0023] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, those of ordinary skill in the art will appreciate that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0024] It should be noted in advance that, in an optional embodiment, except for independent explanations, the same symbols or letters appearing in all formulas have the same meanings and values.

[0025] In an alternative embodiment, see Figure 1 The present invention provides a method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing, the method comprising the following steps: S1. Deploy an anti-interference fading DAS system in the monitoring area.

[0026] Wherein, step S1 specifically includes the following steps: S11. Determine the monitoring area.

[0027] Specifically, in this embodiment, the selected monitoring area is a temporarily designated rectangular monitoring experimental field, covering an area of ​​about 0.5 square kilometers, and 10 cm thick cement is paved in the experimental field.

[0028] S12: planning a buried path for the sensing optical fiber within the monitoring area.

[0029] Specifically, in this embodiment, the monitoring boundaries are first defined by determining the range of drones and unmanned vehicles. Then, key monitoring areas, such as the start and stop locations of drones and unmanned vehicles, are identified within the monitoring area. Fiber density is increased in key monitoring areas, while density can be reduced in less-demand areas to control costs.

[0030] S13: burying the sensing optical fiber according to the burying path and installing the matching DAS system equipment.

[0031] Specifically, in this embodiment, a trench is excavated in the monitoring area according to the planned buried path, with a depth of 2 cm, a width of 2 cm, and a total length of 1.4 km. After placing the optical cable in the trench, cement of the same material as the road surface is used to backfill and compact it to establish a good coupling with the road surface. The partial diagram of the trench and the partial diagram of the trench after backfilling are shown as follows: Figure 2 and Figure 3 shown.

[0032] After the burial of the sensing fiber is completed, the DAS equipment needs to be installed at the ground monitoring center, including narrow linewidth lasers, optical detectors, couplers, erbium-doped fiber amplifiers, filters, circulators, and data acquisition cards.

[0033] S2. Use the anti-interference fading DAS system to obtain the signal dynamic component and composite signal in the monitoring area.

[0034] Wherein, step S2 specifically includes the following steps: S21. Modulate the continuous light generated by the laser into a quasi-orthogonal chirped pulse and inject it into the sensing optical fiber, and use a light detector to obtain the beat frequency signal of the backward Rayleigh scattered light and the local oscillator light.

[0035] Specifically, in this embodiment, the continuous light generated by a narrow-linewidth laser is divided into a first light path and a second light path through a coupler. The first light path is modulated into quasi-orthogonal chirped light pulses comprising positively chirped pulses and negatively chirped pulses through a Mach-Zehnder modulator. The first light path is modulated into quasi-orthogonal chirped light pulses through a Mach-Zehnder modulator. The quasi-orthogonal chirped light pulses are then input into an erbium-doped fiber amplifier to increase the optical power. After filtering out the positive first-order sideband through an optical filter, the quasi-orthogonal chirped light pulses enter the sensor optical fiber through a circulator. The second light path is used as the local oscillator light to beat with the backscattered Rayleigh light output by the circulator and input into a photodetector. The photodetector then performs photoelectric conversion on the received optical signal, and the beat signal is obtained after sampling by a data acquisition card.

[0036] In this embodiment, the chirp rate of the quasi-orthogonal chirped optical pulse is 20 MHz / μs, the pulse width is 5 μs, and the sampling rate is 500 MS / s. In other optional embodiments, the pulse width can be further shortened to improve the multi-target detection capability.

[0037] S22. Extract high and low frequency band signals of the quasi-orthogonal chirped pulse through matched filtering and unmatched filtering, and then suppress interference fading of the beat frequency signal through rotation vector sum synthesis to obtain a synthesized signal.

[0038] Specifically, in this embodiment, when the interference fading of the beat frequency signal is suppressed by rotating the vector sum synthesis, the synthesis phase is adjusted in real time by using closed-loop phase feedback calibration. This step is a conventional technique and will not be described in detail here.

[0039] S23. Perform dynamic detection on the synthesized signal to obtain a signal dynamic component, wherein the dynamic detection is time difference.

[0040] Specifically, in this embodiment, the dynamic detection of the composite signal can be expressed using the following relationship:

[0041] in, is the dynamic component of the signal, is the synthetic signal, For delay The synthesized signal after .

[0042] S24. Perform digital signal processing on the dynamic component of the signal to obtain a composite signal including a vibration signal.

[0043] Specifically, in this embodiment, the composite signal satisfies the following relationship:

[0044] in, is a composite signal, j is an imaginary unit, is the Hilbert transform.

[0045] This embodiment modulates the laser pulse into a quasi-orthogonal chirp signal, combines matched filtering with non-matched filtering to separate high and low frequency signals, and effectively suppresses Interference fading; vector synthesis technology is used to compensate for random phase changes, solve the signal loss problem caused by fading in traditional DAS, significantly improve signal integrity, and thus improve the accuracy of the composite signal, which is conducive to improving the accuracy of monitoring low-altitude and ground unmanned equipment.

[0046] S3. Convert the signal dynamic component into a dynamic feature matrix.

[0047] Wherein, step S3 specifically includes the following steps: S31. Determine the row index and column index of the dynamic feature matrix.

[0048] Specifically, in this embodiment, the spatial position in the optical fiber is used as the row index of the dynamic characteristic matrix, and the time is used as the column index of the dynamic characteristic matrix.

[0049] S32. Construct the dynamic feature matrix according to the row index and the column index using the amplitude of the signal dynamic component.

[0050] Specifically, in this embodiment, each time a laser pulse is triggered, the data acquisition card records the precise emission timestamp. Based on the reception timestamp of the backscattered Rayleigh light, the position of the signal in the optical fiber is calculated using the formula for the propagation speed of the light pulse in the optical fiber:

[0051] Where z is the position of the signal in the optical fiber, c is the speed of light, is the group refractive index of the sensing fiber, , The receiving timestamp.

[0052] Furthermore, by emitting a sequence of laser pulses with a repetition frequency of 500 Hz and correlating the position of the signal in the optical fiber with the signal amplitude, a time-amplitude mapping covering all spatial positions can be constructed, thereby obtaining the amplitude at different positions at different times.

[0053] This embodiment breaks through the limitation of the single characteristic dimension of traditional one-dimensional signal by converting the one-dimensional signal into a three-dimensional feature matrix of time-space-intensity, providing richer spatial pattern information for subsequent noise suppression and vibration source classification, thereby improving the accuracy of monitoring low-altitude and ground unmanned equipment.

[0054] S4. Use a noise suppression model to suppress noise on the dynamic feature matrix to obtain a clean dynamic feature map.

[0055] S41. Constructing a noise suppression model of the dynamic feature matrix.

[0056] Specifically, in this embodiment, the noise suppression model performs the following steps when running: 1. Based on the dynamic feature matrix, a multi-strategy noise suppression scheme is used to generate an original solution set of the noise suppression model.

[0057] Step 1 specifically includes the following steps: (1) The dynamic feature matrix is ​​grayscaled and denoised using a variety of noise suppression schemes to obtain a set of denoised image sets.

[0058] Each value in the dynamic feature matrix is ​​mapped to a grayscale value range to grayscale the dynamic feature matrix, resulting in a dynamic feature grayscale image. This image is then subjected to denoising using wavelet transform, empirical mode decomposition, and Gaussian filtering, respectively, to produce a set of denoised images. This process is conventional and will not be described in detail here.

[0059] (2) Each denoised image in the denoised image set is flattened into a one-dimensional vector, thereby obtaining an original solution set of the noise suppression model.

[0060] First, each denoised image in the denoised image set is flattened into a one-dimensional vector in row order, resulting in a basic solution set. Then, for any two one-dimensional vectors in the basic solution set, the values ​​at the same position are randomly swapped to obtain a new one-dimensional vector and add it to the basic solution set. This operation is repeated until the number of one-dimensional vectors in the basic solution set reaches 30. Finally, the final basic solution set is used as the original solution set for the differential evolution algorithm, where each one-dimensional vector is considered an individual.

[0061] 2. Based on each solution in the original solution set, a mutant individual is generated by differential vector perturbation, and then a test vector is generated by crossover operation.

[0062] First, for the The i-th target individual in generation , it is necessary to randomly select three different individuals to generate mutation vectors. Meanwhile, considering that traditional differential evolution algorithms have problems such as insufficient balance between exploration and exploitation in mutation operations, slow convergence due to fixed scaling factors and crossover factors, easy trapping in local optimality, and difficulty in achieving a balance between optimization efficiency and accuracy, this embodiment makes the following improvements to the traditional differential evolution algorithm: Considering the important influence of basis vectors on the convergence speed of differential evolution algorithm, a variation coefficient is introduced in the mutation operation of traditional differential evolution algorithm. , the coefficient of variation can measure the number of iterations until the maximum number of iterations. Multiplying the coefficient of variation with the basis vector can improve the effectiveness of the algorithm's local search while maintaining the algorithm's global search capability, thereby increasing the algorithm's convergence speed and reducing the possibility of the algorithm falling into a local optimum; The scaling factor and crossover factor of the traditional differential evolution algorithm are improved so that the scaling factor is larger and the crossover factor is smaller at the beginning of the iteration, thereby enhancing the global search ability of the algorithm. As the number of iterations increases, the scaling factor gradually decreases and the crossover factor gradually increases, thereby increasing the local search ability of the algorithm while retaining the excellent information in the population, improving the accuracy and convergence speed of the algorithm.

[0063] In the improved differential evolution algorithm, the scaling factor, crossover factor, and mutation vector satisfy the following relationships respectively:

[0064]

[0065]

[0066] in, For the The mutation vector of the i-th individual in generation, is the maximum number of iterations, For the The r1th individual in the generation, F is the scaling factor, For the The r2th individual in the generation, For the The r3th individual in the generation. When using the original solution set .

[0067] Then, a binomial crossover is performed on the mutation vector and the target individual using the improved crossover factor to generate the test vector.

[0068] 3. Select the mutant individuals and those with high fitness in the test vector as excellent solutions, and use the set of all excellent solutions as the solution set for the next iteration.

[0069] The fitness function used in this embodiment is:

[0070] in, is the fitness value, n is the number of values ​​in the one-dimensional vector, and is the total variation regularization term and fidelity The balance weight, is the value at the j+1th position in the mutant individual or test vector, is the value at the jth position in the mutant individual or test vector, is the variant individual or test vector, is the first individual in the original solution set. and Take 0.55 and 0.45 respectively.

[0071] 4. When the maximum number of iterations is reached or the fitness value converges, the current optimal solution is output, and the optimal solution is reconstructed into the clean dynamic feature map.

[0072] In this embodiment, when the maximum number of iterations is reached or the fitness value converges, the current optimal solution, i.e., the optimal individual, is output. Otherwise, step 2 is returned to enter the next iteration. However, the solution set used in step 2 in the next iteration is the solution set obtained in step 3 during the previous iteration. The maximum number of iterations is set to 200, and the fitness value convergence condition is that the optimal individual meets the requirements for 5 consecutive generations. .

[0073] Since each individual in the original solution set in the differential evolution algorithm is obtained by flattening the corresponding denoised image in the denoised image set, after obtaining the optimal individual, the optimal solution can be reconstructed into a grayscale image, that is, a clean dynamic feature map.

[0074] S42: Input the dynamic feature matrix into the noise suppression model to obtain a clean dynamic feature map.

[0075] This embodiment combines multiple noise reduction schemes to generate an initial solution set, and then iteratively optimizes the solution set through mutation, crossover and selection operations, dynamically searching for the optimal noise suppression strategy, solving the residual noise problem of traditional fixed basis function noise reduction, significantly improving signal purity, and adapting to complex noise scenarios.

[0076] S5. Use the clean dynamic feature map to locate the vibration source, and then combine it with the vibration signal generation model to obtain a vibration signal spectrum map.

[0077] Wherein, step S5 specifically includes the following steps: S51. Locate the vibration source using the clean dynamic feature map.

[0078] Specifically, in this embodiment, a hovering experiment of a drone is conducted at 890m of the sensing fiber, with a hovering height of 7m. The partial clean dynamic feature map obtained by using the noise suppression model is as follows: Figure 4 As shown. Figure 4 It is not difficult to see that the vibration caused by the drone hovering causes the intensity of the backscattered Rayleigh light of the sensing optical fiber at 890m to change periodically, and the high-frequency components are attenuated under the concrete cover, so only the main stripes of the fundamental frequency are observed, that is, the bright narrow-band straight stripes at 890m.

[0079] Furthermore, in order to verify the superiority of this solution, this embodiment uses wavelet transform to compare with the noise suppression model. The dynamic feature map after noise reduction is obtained using conventional wavelet transform. Figure 5 As shown. It is not difficult to see that Figure 5 There is also a bright straight line at 890m, but it is different from Figure 4 compared to, Figure 5 At other locations, a large number of short-range artifact streaks with random orientations remain. This is a typical residual noise pattern caused by the inability of the fixed basis function to adapt to non-stationary noise. In some cases, these short-range artifact streaks may be mistaken for vibration sources, further leading to misclassification of the vibration source. This shows that the noise suppression model provided by this embodiment can better locate the vibration source, laying the foundation for subsequent vibration source classification.

[0080] S52: intercepting the composite signal according to the positioning result of the vibration source to obtain composite signal segments.

[0081] Specifically, in this embodiment, after the vibration source is located, the signal segments at corresponding positions in the composite signal can be intercepted according to the timestamps to obtain composite signal segments.

[0082] S53. Use CycleGAN to build a vibration signal generation model, and then obtain a vibration signal spectrum diagram according to the composite signal segment.

[0083] Specifically, in this embodiment, multiple unmanned vehicle or drone operation experiments are first conducted within the monitoring area. Noisy signals generated by the drone or drone operation during the experiments are collected using optical fiber sensors. Simultaneously, a laser Doppler vibrometer is used to measure vibration signals from the ground directly below the unmanned vehicle or drone to obtain a clean signal. The noisy and clean signals require time synchronization. Next, the vibration signal is intercepted from the clean signal to obtain a clean vibration signal. Simultaneously, the noisy signal is intercepted according to the timestamp of the clean vibration signal to obtain the corresponding noisy vibration signal. When intercepting the clean vibration signal, the clean vibration signal should have the same time span as the composite signal segment without compromising the integrity of the vibration signal. Next, a short-time Fourier transform is performed on the clean and noisy vibration signals to obtain clean and noisy vibration signal spectrograms. These spectrograms are then scaled to 256×256 pixels and used to construct a training and validation dataset for CycleGAN (cycle-consistent generative adversarial network). Finally, the obtained dataset was divided into a training set and a validation set in a ratio of 7:3 to complete the training and validation of CycleGAN. This enabled CycleGAN to generate a corresponding clean vibration signal spectrogram based on the input noisy vibration signal spectrogram. Ultimately, the CycleGAN generator was used as a vibration signal generation model.

[0084] In this example, the CycleGAN generator uses a U-Net network structure, the discriminator uses a 70×70 PatchGAN, and the CycleGAN loss function includes a generative adversarial loss and a cycle consistency loss. These are existing technical methods and will not be described in detail here.

[0085] After the vibration signal generation model is obtained, a short-time Fourier transform is performed on the composite signal segment to obtain a composite signal spectrum diagram and the composite signal spectrum diagram is input into the vibration signal generation model to obtain a vibration signal spectrum diagram.

[0086] This embodiment first locates the vibration source position, and then intercepts the composite signal at the corresponding position to obtain a composite signal segment. This not only alleviates the computational redundancy caused by global signal processing when classifying vibration sources in the existing technology to a certain extent, but also reduces classification errors caused by high-frequency noise. The vibration signal generation model constructed using CycleGAN is applicable to various types of noise and signals, and can solve the problem that traditional methods such as wavelet transform cannot effectively process non-stationary noise in vibration signals.

[0087] S6. Use a convolutional neural network to build a vibration source classification and identification model, and then use the vibration signal spectrum diagram and the vibration source classification and identification model to classify and identify the vibration source.

[0088] Wherein, step S6 specifically includes the following steps: S61. Construct a vibration signal spectrum dataset.

[0089] Specifically, in this embodiment, the clean vibration signal spectrum obtained in step S53 is annotated to distinguish the source of the vibration signal, thereby constructing a vibration signal spectrum data set.

[0090] S62. Use the vibration signal spectrum dataset to train and verify a convolutional neural network to obtain a vibration source classification and recognition model.

[0091] Specifically, in this embodiment, the vibration signal data set is divided into a training set and a validation set in a ratio of 7:3 to complete the training and validation of the convolutional neural network and obtain a vibration source classification and recognition model.

[0092] S63: Input the vibration signal spectrum diagram into the vibration source classification and identification model to achieve classification and identification of the vibration source.

[0093] This embodiment utilizes CNN's ability to extract local patterns of time-frequency features to fully capture the frequency components and temporal dynamics of different vibration sources, solves the problem of easy confusion in traditional one-dimensional feature classification, and improves the accuracy of vibration source classification.

[0094] It should be noted that, in some cases, the actions described in the specification can be performed in a different order and still achieve the desired results. In this embodiment, the order of steps given is only to make the embodiment appear clearer and easier to explain, rather than to limit it.

[0095] In an alternative embodiment, see Figure 6 In order to improve the practicality of the present method and facilitate the promotion of the present method, the present invention also provides a low-altitude and ground unmanned equipment monitoring system based on optical fiber sensing. The low-altitude and ground unmanned equipment monitoring system based on optical fiber sensing includes: a data acquisition device 1, a data output device 2, a processor 3 and a storage 4. The storage 4 includes a computer-readable storage medium. The computer-readable storage medium stores a computer program. The computer program includes program instructions. When the program instructions are executed by the processor 3, the processor 3 implements the contents described in steps S1 to S6.

[0096] In summary, this solution has at least the following beneficial effects: 1. This method uses optical fiber to sense the start and stop of drones or unmanned vehicles, so that the drones or unmanned vehicles can still be sensed in the presence of electromagnetic interference.

[0097] 2. This method modulates the laser pulse into a quasi-orthogonal chirp signal, combines matched filtering with non-matched filtering to separate high and low frequency signals, and effectively suppresses Interference fading; vector synthesis technology is used to compensate for random phase changes, solve the signal loss problem caused by fading in traditional DAS, significantly improve signal integrity, and thus improve the accuracy of the composite signal, which is conducive to improving the accuracy of monitoring low-altitude and ground unmanned equipment.

[0098] 3. This method breaks through the limitation of the single characteristic dimension of traditional one-dimensional signal by converting the one-dimensional signal into a three-dimensional feature matrix of time, space and intensity, and then combines multiple noise reduction schemes and improves the differential evolution algorithm to achieve accurate positioning of the vibration source.

[0099] 4. This method first locates the vibration source position, and then intercepts the composite signal at the corresponding position to obtain the composite signal segment. This method can not only alleviate the computational redundancy caused by global signal processing in the existing technology when classifying vibration sources to a certain extent, but also reduce the classification errors caused by high-frequency noise. 5. This method uses CycleGAN to construct a vibration signal generation model, which is applicable to various types of noise and signals and can solve the problem that traditional methods such as wavelet transform cannot effectively process non-stationary noise in vibration signals.

[0100] 6. This method utilizes CNN's ability to extract local patterns of time-frequency features to fully capture the frequency components and temporal dynamics of different vibration sources, solve the problem of easy confusion in traditional one-dimensional feature classification, and improve the accuracy of vibration source classification.

[0101] 7. A system compatible with the method is provided, which can improve the practicability of the method and facilitate the promotion of the method.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing, characterized in that: The steps include: Deploy anti-interference fading DAS system in the monitoring area; Using the anti-interference fading DAS system to obtain the signal dynamic components and composite signals within the monitoring area; Converting the signal dynamic component into a dynamic feature matrix; Using a noise suppression model to suppress noise on the dynamic feature matrix to obtain a clean dynamic feature map; Using the clean dynamic characteristic graph to locate the vibration source, and then combining it with the vibration signal generation model to obtain a vibration signal spectrum graph; A convolutional neural network is used to construct a vibration source classification and identification model, and then the vibration source is classified and identified using the vibration signal spectrum diagram and the vibration source classification and identification model.

2. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1 is characterized in that: The deployment of the anti-interference fading DAS system in the monitoring area includes the following steps: Determine the monitoring area; Planning the buried path of the sensing optical fiber within the monitoring area; The sensing optical fiber is buried according to the buried path, and the supporting DAS system equipment is installed.

3. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1, characterized in that: The method of using the anti-interference fading DAS system to obtain the signal dynamic component and composite signal in the monitoring area includes the following steps: The continuous light generated by the laser is modulated into quasi-orthogonal chirped pulses and injected into the sensing fiber, and the beat frequency signal of the backscattered Rayleigh light and the local oscillator light is obtained using a photodetector; The high and low frequency band signals of the quasi-orthogonal chirped pulse are extracted by matched filtering and unmatched filtering, and then the interference fading of the beat frequency signal is suppressed by rotating vector sum synthesis to obtain a synthetic signal; Performing dynamic detection on the synthesized signal to obtain a signal dynamic component, wherein the dynamic detection is time difference; Digital signal processing is performed on the dynamic component of the signal to obtain a composite signal containing a vibration signal.

4. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 3 is characterized in that: The step of converting the signal dynamic component into a dynamic feature matrix comprises the following steps: Determining the row index and column index of the dynamic feature matrix; The dynamic feature matrix is ​​constructed using the amplitude of the signal dynamic component according to the row index and the column index.

5. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1 is characterized in that: The method of using a noise suppression model to suppress noise on the dynamic feature matrix to obtain a clean dynamic feature map includes the following steps: Constructing a noise suppression model of the dynamic feature matrix; The dynamic feature matrix is ​​input into the noise suppression model to obtain a clean dynamic feature map.

6. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1, characterized in that: The noise suppression model performs the following steps when running: generating an original solution set of a noise suppression model using a multi-strategy noise suppression scheme according to the dynamic feature matrix; According to each solution in the original solution set, a mutant individual is generated by differential vector perturbation, and then a test vector is generated by crossover operation; Select the mutant individuals and those with high fitness in the test vector as excellent solutions, and use the set of all excellent solutions as the solution set for the next iteration; When the maximum number of iterations is reached or the fitness value converges, the current optimal solution is output, and the optimal solution is reconstructed into the clean dynamic feature map.

7. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 6, characterized in that: The method of generating an original solution set of a noise suppression model using a multi-strategy noise suppression scheme according to the dynamic feature matrix comprises the following steps: Gray-scaling the dynamic feature matrix and performing noise reduction processing on the dynamic feature matrix using multiple noise suppression schemes, thereby obtaining a set of noise-reduced image sets; Each denoised image in the denoised image set is flattened into a one-dimensional vector, thereby obtaining an original solution set of the noise suppression model.

8. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1, characterized in that: The method of locating the vibration source by using the clean dynamic characteristic graph and then obtaining the vibration signal spectrum graph in combination with the vibration signal generation model includes the following steps: Locating the vibration source using the clean dynamic characteristic map; intercepting the composite signal according to the result of locating the vibration source to obtain composite signal segments; CycleGAN is used to build a vibration signal generation model, and then a vibration signal spectrum is obtained based on the composite signal segment.

9. The method for monitoring low-altitude and ground unmanned equipment based on optical fiber sensing according to claim 1, characterized in that: The method of using a convolutional neural network to construct a vibration source classification and identification model, and then using the vibration signal spectrum diagram and the vibration source classification and identification model to classify and identify the vibration source, includes the following steps: Construct a vibration signal spectrum dataset; Using the vibration signal spectrum dataset to train and validate a convolutional neural network to obtain a vibration source classification and recognition model; The vibration signal spectrum is input into the vibration source classification and identification model to achieve classification and identification of the vibration source.

10. The low-altitude and ground unmanned equipment monitoring system based on optical fiber sensing is characterized by: The low-altitude and ground unmanned equipment monitoring system based on optical fiber sensing includes: a data acquisition device, a data output device, a processor and a storage device, the storage device includes a computer-readable storage medium, the computer-readable storage medium stores a computer program, the computer program includes program instructions, and when the program instructions are executed by the processor, the processor implements the low-altitude and ground unmanned equipment monitoring method based on optical fiber sensing as described in any one of claims 1-9.

Images