A monitoring and early warning method and system for a distributed fiber optic acoustic sensor

Abstract

The present invention relates to the technical field of distributed optical fiber acoustic wave sensors, and in particular to a monitoring and early warning method and system for distributed optical fiber acoustic wave sensors. The method comprises: acquiring historical acoustic wave data and early warning data within a preset time period from a database; forming a monitoring and early warning sample database; using broadband chirped continuous light as detection light and single-frequency light as local oscillator light, reconstructing the Rayleigh scattered light field of any frequency at any time through coherent detection, and improving the response bandwidth of the distributed optical fiber acoustic wave sensor; constructing a monitoring model, using the trained monitoring model to analyze the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, and identifying normal acoustic wave signals and abnormal acoustic wave signals; introducing a delayed summation beamforming algorithm, and improving the accuracy of the distributed optical fiber acoustic wave sensor in detecting weak high-frequency acoustic wave signals; and analyzing and processing the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor through the early warning model, and identifying vibration events.

Description

Technical Field

[0001] The present invention relates to the technical field of distributed optical fiber acoustic wave sensors, and in particular to a monitoring and early warning method and system for distributed optical fiber acoustic wave sensors. Background Art

[0002] The working principle of the distributed fiber acoustic sensor is as follows: due to the uneven distribution of parameters such as mass and density in the fiber drawing process, light will be scattered when it propagates in the fiber. The scattering direction of light in the fiber is randomly distributed, and part of the scattered light will propagate in the opposite direction of the incident light, so it is called backscattering. The backscattering of optical fiber can be divided into Rayleigh scattering, Brillouin scattering and Raman scattering according to different scattering factors. Among them, DAS is mainly realized by the backscattering of optical fiber. When the optical fiber is affected by external vibration, the phase of the Rayleigh backscattered light will change. According to the elastic-photoelectric effect of the optical fiber, the amplitude of the lateral vibration of the optical fiber is proportional to the magnitude of the axial strain of the optical fiber, and the phase change is linearly related to the change of the external vibration. Therefore, by detecting the change in the phase or intensity of the Rayleigh backscattered light, the acoustic wave signal of the external vibration event can be restored.

[0003] At present, various enterprises mainly monitor safety risks around pipelines through manual inspections. The so-called safety risks include third-party construction, pipeline leakage, etc. Manual inspections have problems such as difficulty in covering all pipelines and inability of inspectors to go to the site for inspections in harsh environments. Distributed Fiber Acoustic Sensing (DAS) uses optical fibers laid in the same trench as the pipeline to measure the vibration of the optical fiber by detecting the phase change of Rayleigh scattering of pulsed light in the optical fiber, thereby realizing timely perception of vibration events around the pipeline (such as third-party construction, pipeline leakage, human damage, etc.), which can effectively solve the shortcomings of manual inspections.

[0004] However, there are still some problems with the DAS system in actual use, such as a high probability of false alarms and the inability to identify small-aperture gas leaks during the leak monitoring process. This is because the sound wave frequency of small-aperture leaks is relatively high, reaching hundreds of kHz, which exceeds the frequency response bandwidth of the DAS. Summary of the invention

[0005] The present invention applies a GAN adversarial network model, generates false alarm sound wave signals through the GAN model, trains the early warning model to identify the corresponding false alarm sound wave signals, reduces the false alarm rate of the monitoring and early warning model, and improves the accuracy of distributed optical fiber acoustic wave sensors in detecting weak high-frequency acoustic wave signals by introducing a delayed-sum beamforming algorithm.

[0006] The technical solution proposed by the present invention is: a monitoring and early warning method of a distributed optical fiber acoustic wave sensor, the method comprising:

[0007] Obtain historical sound wave data and warning data within a preset time period from the database;

[0008] A voiceprint data sample library is formed based on the historical sound wave data obtained, and the voiceprint data sample library and the early warning data sample library are combined to form a monitoring and early warning sample database;

[0009] Using coherent detection technology, the frequency range of the acoustic wave signal collected by the distributed optical fiber acoustic wave sensor is expanded;

[0010] Construct a monitoring model and use the voiceprint data sample library to train the monitoring model; use the trained monitoring model to analyze the acoustic signals collected by the distributed optical fiber acoustic sensor to identify normal acoustic signals and abnormal acoustic signals;

[0011] Detect whether the actually collected sound wave signal contains an abnormal signal; if it does not contain an abnormal signal, proceed to the next step, otherwise, return to the previous step;

[0012] The weak high-frequency sound wave signals in the actually collected sound wave signals are analyzed by using a delay-sum beamforming algorithm;

[0013] The trained early warning model is used to analyze and process the acoustic wave signals collected by the distributed fiber optic acoustic wave sensor, identify vibration events, and issue early warning information.

[0014] Preferably, the method further includes obtaining actual operation data and false alarm feedback information, and extracting false alarm area location information and false alarm time period information;

[0015] The adversarial network GAN model is applied to generate false alarm sound wave signals through the GAN model, and the early warning model is trained to identify the corresponding false alarm sound wave signals, thereby reducing the false alarm rate of the monitoring and early warning model.

[0016] Preferably, the forming of a voiceprint data sample library based on the acquired historical sound wave data comprises the following steps:

[0017] Acquire historical sound wave data within a preset time period from a database, extract the sound wave acquisition time from the historical sound wave data, and pre-process the sound wave data to remove noise and interference;

[0018] Convert the pre-processed historical sound wave data into the frequency domain and extract the key features in the frequency domain;

[0019] According to the key features in the frequency domain, the sound wave data is divided into multiple voiceprint sample sets for representing different vibration events, and a type label and a collection time label are added to each voiceprint sample set.

[0020] Preferably, the step of combining the voiceprint data sample library and the early warning data sample library to form a monitoring and early warning sample database comprises the following steps:

[0021] Extract the warning record data, warning processing log and the sound wave signal that generates the warning record from the warning data sample library; extract the warning event type from the warning processing log;

[0022] Compare the sound wave signal that generates the warning record with the sound wave signal in each voiceprint sample set to determine the similarity;

[0023] If the similarity is within the set similarity threshold, the voiceprint signal in the sound wave sample set is associated with the warning time type and stored in the local database;

[0024] Otherwise, an extended voiceprint sample set is added to the voiceprint data sample library, wherein the extended voiceprint sample set includes the sound wave signal for generating the warning record, the unique hot encoding of the warning event type and the acquisition timestamp; and the extended voiceprint sample library is stored in the local database.

[0025] Preferably, if the similarity is within a set similarity threshold range, the voiceprint signal in the sound wave sample set is associated with the warning time type and stored in a local database, comprising the following steps:

[0026] Reading sound wave sample data, wherein the sound wave sample data includes the sound wave signal that generates the warning record and the sound wave signal in the voiceprint sample set;

[0027] Use filters or noise reduction algorithms to remove noise from sound wave signals;

[0028] Normalize the sound wave sample data after noise removal to make the data range consistent;

[0029] Use audio analysis and processing algorithms to extract corresponding frequency domain feature vectors from the sound wave signals in the sound wave sample data and, ;in, represents the dimension of the frequency domain feature vector, Indicates The frequency domain feature vector of the acoustic signal that generates the warning record, Indicates The frequency domain feature vector of the sound wave signal in the voiceprint sample set;

[0030] Calculated using the cosine similarity formula and Similarity ,

[0031] Set a similarity threshold a, if ,but and The corresponding voiceprint signals are similar, and the voiceprint signals in the sound wave sample set are associated with the corresponding types of warning events;

[0032] An embedded relational database is used to store voiceprint signal data associated with warning events.

[0033] Preferably, the method of reconstructing the Rayleigh scattered light field of any frequency at any time by coherent detection using broadband chirped continuous light as the detection light and single-frequency light as the local oscillator light comprises the following steps:

[0034] Assume that the detection light and the local oscillator light are:

[0035] ;

[0036] ;

[0037] in, and Represent the angular frequencies of the detection tube and the local oscillator light respectively;

[0038] The mixed signal is:

[0039] , expand the above formula to obtain:

[0040] ;

[0041] The low-frequency component, the beat frequency signal, is extracted through a low-pass filter:

[0042] ;

[0043] The frequency of the beat signal is:

[0044] ;

[0045] Perform Fourier transform on the beat frequency signal to obtain frequency domain information:

[0046] ;

[0047] In this way, the Rayleigh scattered light field with different frequency components is reconstructed.

[0048] Preferably, the building of the monitoring model includes:

[0049] Use multi-layer convolutional neural networks to build monitoring models;

[0050] The multi-layer convolutional neural network includes an input layer, multiple convolutional layers and an output layer;

[0051] Receiving a voiceprint data sample through the input layer;

[0052] Outputting the judgment result of the sound wave signal through the output layer, wherein the judgment result includes normal and abnormal;

[0053] The method uses the trained early warning model to analyze and process the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, identifies vibration events, and issues early warning information, including:

[0054] A convolutional neural network (CNN) is used to construct an early warning model; an output layer of the early warning model outputs a probability value of the occurrence of the identified vibration event;

[0055] Use the monitoring and early warning sample database to train the early warning model;

[0056] The received sound wave signal is classified through the trained early warning model, the type of vibration event is identified, and the predicted value of the probability of the corresponding vibration event is output;

[0057] Set the confidence threshold c. If the predicted value is greater than c, a warning signal is generated.

[0058] The type of vibration event and the warning signal are sent to the host computer.

[0059] Preferably, the introduction of a delay-sum beamforming algorithm to improve the accuracy of the distributed optical fiber acoustic wave sensor in detecting weak high-frequency acoustic wave signals comprises the following steps:

[0060] Obtaining a distributed optical fiber acoustic wave sensor receiving an acoustic wave signal;

[0061] Acquiring an acoustic wave signal received by a communication optical fiber near a distributed optical fiber acoustic wave sensor;

[0062] According to the positions of the distributed optical fiber acoustic wave sensor and the communication optical fiber and the direction of the target signal, the distributed optical fiber acoustic wave sensor and the communication optical fiber are delayed accordingly;

[0063] Performing weighted summation on the delayed acoustic wave signals of the distributed optical fiber acoustic wave sensor and the communication optical fiber to obtain a final output signal;

[0064] Specifically:

[0065] Features Distributed fiber optic acoustic sensors and communication fibers;

[0066] Set up The signal received by a distributed fiber optic acoustic sensor or communication fiber is ;

[0067] Assume the direction of the target sound source is , the location of the distributed fiber optic acoustic wave sensor is ;

[0068] The first The signal of the communication fiber needs to be delayed ,in represents the speed of sound; the output signal of the delay-sum beamformer is ,in represents the weight coefficient, Acquired according to the array antenna algorithm; draw the time domain waveform of the original signal and the beam signal.

[0069] The present invention also provides a monitoring and early warning system for a distributed optical fiber acoustic wave sensor, wherein the system is used to execute the monitoring and early warning method for a distributed optical fiber acoustic wave sensor.

[0070] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the monitoring and early warning method of the distributed optical fiber acoustic wave sensor.

[0071] Beneficial effects of the present invention:

[0072] 1. The present invention constructs a monitoring model to analyze the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, identify normal acoustic wave signals, abnormal acoustic wave signals and vibration events, and issue early warning information.

[0073] 2. The present invention improves the accuracy of distributed optical fiber acoustic wave sensors in detecting weak high-frequency acoustic wave signals by introducing a delayed-sum beamforming algorithm. For example, in the field of pipeline leak detection, the ability of distributed optical fiber acoustic wave sensors to identify small-aperture leaks can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 The present invention is a flow chart of a monitoring and early warning method of a distributed optical fiber acoustic wave sensor. DETAILED DESCRIPTION

[0075] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0076] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0077] refer to Figure 1 The technical solution provided by the present invention is: a monitoring and early warning method of a distributed optical fiber acoustic wave sensor, comprising the following steps:

[0078] 1. Obtain historical sound wave data and warning data within a preset time period from the database. This is achieved through the following steps:

[0079] Through structured query language SQL query statements, historical sound wave data within a specified time period are extracted from the database, and the sound wave acquisition time is extracted from the historical sound wave data, and the sound wave data is preprocessed including but not limited to filtering and noise reduction;

[0080] The pre-processed historical sound wave data is converted into frequency domain signals using Fourier transform (FFT), and key features are extracted;

[0081] According to the key features in the frequency domain, the sound wave data is divided into multiple voiceprint sample sets for representing different vibration events, and a type label and a collection time label are added to each voiceprint sample set.

[0082] 2. Constructing a voiceprint data sample library based on the historical sound wave data obtained, and combining the voiceprint data sample library and the early warning data sample library to form a monitoring and early warning sample database; including the following steps:

[0083] Connect to the warning data sample library and execute the embedded relational database query command to extract the required warning record data, warning processing log and sound wave signals related to the warning record. The extracted data includes key information such as the type of warning event and the timestamp of the sound wave signal;

[0084] Use text processing or pattern matching techniques to extract the type of warning event. Use regular expressions or other string matching techniques to identify specific event descriptions.

[0085] Compare the sound wave signal that generates the warning record with the sound wave signal in each voiceprint sample set to determine the similarity;

[0086] If the similarity is within the set similarity threshold, the voiceprint signal in the sound wave sample set is associated with the warning time type and stored in the local database, specifically:

[0087] Read the sound wave sample data, which includes the sound wave signal that generates the warning record and the sound wave signal in the voiceprint sample set; use a filter or noise reduction algorithm to remove the noise of the sound wave signal; normalize the sound wave sample data after noise removal to make the data range consistent; use the audio analysis processing algorithm (Librosa library) to extract the corresponding frequency domain feature vectors from the sound wave signals in the sound wave sample data and, ;in, represents the dimension of the frequency domain feature vector, Indicates The frequency domain feature vector of the acoustic signal that generates the warning record, Indicates The frequency domain feature vector of the sound wave signal in the voiceprint sample set; calculated using the cosine similarity formula and Similarity , set a similarity threshold a, if ,but and The corresponding voiceprint signals are similar, and the voiceprint signals in the sound wave sample set are associated with the corresponding types of warning events; an embedded relational database (SQLite) is used to store the voiceprint signal data associated with the warning events.

[0088] Otherwise, an extended voiceprint sample set is added to the voiceprint data sample library, wherein the extended voiceprint sample set includes the sound wave signal for generating the warning record, the unique hot encoding of the warning event type and the acquisition timestamp; and the extended voiceprint sample library is stored in the local database.

[0089] Because the distributed fiber optic acoustic wave sensor is a device used to detect and measure sound waves. Through coherent detection technology, the Rayleigh scattered light field of any frequency at any time can be reconstructed, thereby improving the response bandwidth of the sensor. High-frequency sound wave signals can be detected. The details are as follows:

[0090] 3. Using broadband chirped continuous light as the detection light and single-frequency light as the local oscillator light, the Rayleigh scattered light field of any frequency at any time is reconstructed through coherent detection to improve the response bandwidth of the distributed fiber optic acoustic wave sensor. The specific principle is:

[0091] Assume that the detection light and the local oscillator light are:

[0092] ;

[0093] ;

[0094] in, and Represent the angular frequencies of the detection tube and the local oscillator light respectively;

[0095] The mixed signal is:

[0096] , expand the above formula to obtain:

[0097] ;

[0098] The low-frequency component, the beat frequency signal, is extracted through a low-pass filter:

[0099] ;

[0100] The frequency of the beat signal is:

[0101] ;

[0102] Perform Fourier transform on the beat frequency signal to obtain frequency domain information:

[0103] ; In this way, the Rayleigh scattered light field with different frequency components can be reconstructed.

[0104] 4. Build a monitoring model and use the voiceprint data sample library to train the monitoring model; use the trained monitoring model to analyze the acoustic signals collected by the distributed optical fiber acoustic sensor to identify normal acoustic signals and abnormal acoustic signals; detect whether the actual collected acoustic signals contain abnormal signals.

[0105] Specifically: a monitoring model is constructed using a multi-layer convolutional neural network; the multi-layer convolutional neural network includes an input layer, multiple convolutional layers and an output layer; the voiceprint data sample is received through the input layer; the judgment result of the sound wave signal is output through the output layer, and the judgment result includes normal and abnormal. The Softmax function is usually used to convert the output into a probability distribution, indicating the probability of belonging to different categories.

[0106] 5. Analyze the weak high-frequency sound wave signal in the actual collected sound wave signal through the delay-sum beamforming algorithm. Specifically, the following steps are included:

[0107] Obtain an acoustic signal received by a distributed optical fiber acoustic wave sensor; obtain an acoustic signal received by a communication optical fiber near the distributed optical fiber acoustic wave sensor; perform a corresponding time delay on the distributed optical fiber acoustic wave sensor and the communication optical fiber according to the positions of the distributed optical fiber acoustic wave sensor and the communication optical fiber and the direction of the target signal; perform a weighted summation on the delayed acoustic signals of the distributed optical fiber acoustic wave sensor and the communication optical fiber to obtain a final output signal;

[0108] Specifically:

[0109] Features Distributed fiber optic acoustic sensors and communication fibers;

[0110] Set up The signal received by a distributed fiber optic acoustic sensor or communication fiber is ;

[0111] Assume the direction of the target sound source is , the location of the distributed fiber optic acoustic wave sensor is ;

[0112] The first The signal of the communication fiber needs to be delayed ,in represents the speed of sound; the output signal of the delay-sum beamformer is ,in represents the weight coefficient, Acquired according to the array antenna algorithm; draw the time domain waveform of the original signal and the beam signal.

[0113] 6. Use the trained warning model to analyze and process the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, identify vibration events, and issue warning information, which specifically includes the following steps:

[0114] Convolutional neural network (CNN) is used to construct an early warning model; the output layer of the early warning model outputs the probability value of the vibration event that has been identified; the early warning model is trained using a monitoring early warning sample database; the received sound wave signal is classified by the trained early warning model, the type of vibration event is identified, and the predicted value of the probability of the corresponding vibration event is output; a confidence threshold c is set, and if the predicted value is greater than c, an early warning signal is generated; the type of vibration event and the early warning signal are sent to a host computer. The training method of the early warning model in this embodiment is a commonly used training method of a convolutional neural network model, and will not be described in detail.

[0115] 7. Extract the location information of the false alarm area and the false alarm time period information from the actual operation data and the false alarm feedback information.

[0116] In this embodiment, in order to reduce the false alarm rate of the warning model, a generative adversarial network (GAN) is used to generate false alarm sound wave signals. GAN consists of two main parts: the generator and the discriminator. The generator is responsible for generating realistic false alarm sound wave signals. The discriminator is responsible for distinguishing between real sound wave signals and generated false alarm sound wave signals. Specifically:

[0117] First, the location information (latitude and longitude) of the false alarm area and the false alarm time period information are extracted from the false alarm feedback information and the actual operation data. This information will be used to train the Generative Adversarial Network (GAN), and then the timestamps will be converted into numerical features.

[0118] Constructing the generator model: A multi-layer perceptron MLP is used, each layer contains a fully connected layer, an activation function (LeakyReLU), and a batch normalization layer. The input dimension is set to 100-dimensional random noise. The output dimension is set to 1024-dimensional generated data.

[0119] Constructing the discriminator: In this embodiment, a multi-layer perceptron MLP is also used, each layer of which includes a fully connected layer, a LeakyReLU activation function, a Dropout layer, and a batch normalization layer.

[0120] Set the input dimension to 1024-dimensional real data or generated data. Set the output dimension to 1-dimensional probability value.

[0121] The GAN model is composed of a generator model and a discriminator. During the training process, the generator tries to generate realistic data to deceive the discriminator, while the discriminator tries to distinguish between real data and generated data.

[0122] The loss function uses the binary cross entropy loss function. The gradient descent optimizer (Adam optimizer) is used with a learning rate of 0.0002 and a momentum of 0.5.

[0123] In each training cycle, the following steps are performed:

[0124] Randomly extract half of the data from the real dataset as real samples. Generate fake samples with the same number of real samples;

[0125] Update the discriminator: train the discriminator with real samples and fake samples respectively, and calculate the loss of the discriminator; Update the generator: train the generator with the generated fake samples, and calculate the loss of the generator;

[0126] Print the loss value and accuracy of the current cycle. Save or visualize the generated data at regular intervals.

[0127] Through adversarial training, the generator will continuously improve its generation capabilities, making the generated false alarm sound wave signals more and more realistic, thereby helping the early warning model to better identify false alarms.

[0128] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the monitoring and early warning method of the distributed optical fiber acoustic wave sensor.

[0129] The embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. The embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above-mentioned functions defined in the method of the present application are executed. It should be noted that the above-mentioned computer-readable medium of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of an electrical, magnetic, optical, electromagnetic, infrared segment, or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wire segments, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, electrical wire, optical cable, RF, etc., or any suitable combination of the foregoing.

[0130] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0131] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments. Without departing from the principles, the implementation methods of the present invention may be deformed or modified in any way.

Claims

1. A monitoring and early warning method for a distributed optical fiber acoustic wave sensor, characterized in that: The method comprises: Obtain historical sound wave data and warning data within a preset time period from the database; A voiceprint data sample library is formed based on the historical sound wave data obtained, and the voiceprint data sample library and the early warning data sample library are combined to form a monitoring and early warning sample database; Using coherent detection technology, the frequency range of the acoustic wave signal collected by the distributed optical fiber acoustic wave sensor is expanded; Construct a monitoring model and use the voiceprint data sample library to train the monitoring model; use the trained monitoring model to analyze the acoustic signals collected by the distributed optical fiber acoustic sensor to identify normal acoustic signals and abnormal acoustic signals; Detect whether the actually collected sound wave signal contains an abnormal signal; if it does not contain an abnormal signal, proceed to the next step, otherwise, return to the previous step; The weak high-frequency sound wave signals in the actually collected sound wave signals are analyzed by using a delay-sum beamforming algorithm; Use the trained warning model to analyze and process the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, identify vibration events, and issue warning information; The method shown also includes: obtaining actual operation data and false alarm feedback information, extracting false alarm area location information and false alarm time period information; Apply the adversarial network GAN model, generate false alarm sound wave signals through the GAN model, train the early warning model to identify the corresponding false alarm sound wave signals, and reduce the false alarm rate of the monitoring and early warning model; The method of combining the voiceprint data sample library and the early warning data sample library to form a monitoring and early warning sample database includes the following steps: Extract the warning record data, warning processing log and the sound wave signal that generates the warning record from the warning data sample library; extract the warning event type from the warning processing log; Compare the sound wave signal that generates the warning record with the sound wave signal in each voiceprint sample set to determine the similarity; If the similarity is within the set similarity threshold, the voiceprint signal in the sound wave sample set is associated with the warning time type and stored in the local database; Otherwise, adding an extended voiceprint sample set to the voiceprint data sample library, wherein the extended voiceprint sample set includes the sound wave signal for generating the warning record, the unique hot encoding of the warning event type, and the acquisition timestamp; storing the extended voiceprint sample library in the local database; If the similarity is within the set similarity threshold range, the voiceprint signal in the sound wave sample set is associated with the warning time type and stored in the local database, including the following steps: Reading sound wave sample data, wherein the sound wave sample data includes the sound wave signal that generates the warning record and the sound wave signal in the voiceprint sample set; Use filters or noise reduction algorithms to remove noise from sound wave signals; Normalize the sound wave sample data after noise removal to make the data range consistent; Use audio analysis and processing algorithms to extract corresponding frequency domain feature vectors from the sound wave signals in the sound wave sample data and, ;in, represents the dimension of the frequency domain feature vector, Indicates The frequency domain feature vector of the acoustic signal that generates the warning record, Indicates The frequency domain feature vector of the sound wave signal in the voiceprint sample set; Calculated using the cosine similarity formula and Similarity , set a similarity threshold a, if ,but and The corresponding voiceprint signals are similar, and the voiceprint signals in the sound wave sample set are associated with the corresponding types of warning events; Use an embedded relational database to store voiceprint signal data associated with warning events; The method of utilizing the coherent detection technology to expand the frequency range of the acoustic wave signal collected by the distributed optical fiber acoustic wave sensor comprises the following steps: Assume that the detection light and the local oscillator light are: ； ； in, and represent the angular frequencies of the probe light and the local oscillator light respectively; The mixed signal is: , expand the above formula to obtain: ； The low-frequency component, the beat frequency signal, is extracted through a low-pass filter: ； The frequency of the beat signal is: ； Perform Fourier transform on the beat frequency signal to obtain frequency domain information: ； In this way, the Rayleigh scattered light field with different frequency components is reconstructed; The method of analyzing the weak high-frequency sound wave signal in the actually collected sound wave signal by using the delay-sum beamforming algorithm comprises the following steps: Obtaining a distributed optical fiber acoustic wave sensor receiving an acoustic wave signal; Acquiring an acoustic wave signal received by a communication optical fiber near a distributed optical fiber acoustic wave sensor; According to the positions of the distributed optical fiber acoustic wave sensor and the communication optical fiber and the direction of the target signal, the distributed optical fiber acoustic wave sensor and the communication optical fiber are delayed accordingly; Performing weighted summation on the delayed acoustic wave signals of the distributed optical fiber acoustic wave sensor and the communication optical fiber to obtain a final output signal; Specifically: Features Distributed fiber optic acoustic sensors and communication fibers; Set up The signal received by a distributed fiber optic acoustic sensor or communication fiber is ; Assume the direction of the target sound source is , the location of the distributed fiber optic acoustic wave sensor is ; The first The signal of the communication fiber needs to be delayed ,in represents the speed of sound; the output signal of the delay-sum beamformer is ,in represents the weight coefficient, Acquired according to array antenna algorithm; Plot the time domain waveforms of the original signal and the beamed signal.

2. A monitoring and early warning method of a distributed optical fiber acoustic wave sensor according to claim 1, characterized in that: The process of forming a voiceprint data sample library based on the acquired historical sound wave data includes the following steps: Acquire historical sound wave data within a preset time period from a database, extract the sound wave acquisition time from the historical sound wave data, and pre-process the sound wave data to remove noise and interference; After converting the pre-processed historical sound wave data into the frequency domain, the key features in the frequency domain are extracted; According to the key features in the frequency domain, the sound wave data is divided into multiple voiceprint sample sets for representing different vibration events, and a type label and a collection time label are added to each voiceprint sample set.

3. The monitoring and early warning method of a distributed optical fiber acoustic wave sensor according to claim 1 is characterized in that: The construction of the monitoring model includes: Use multi-layer convolutional neural networks to build monitoring models; The multi-layer convolutional neural network includes an input layer, multiple convolutional layers and an output layer; Receiving a voiceprint data sample through the input layer; Outputting the judgment result of the sound wave signal through the output layer, wherein the judgment result includes normal and abnormal; The method uses the trained early warning model to analyze and process the acoustic wave signals collected by the distributed optical fiber acoustic wave sensor, identifies vibration events, and issues early warning information, including: A convolutional neural network (CNN) is used to construct an early warning model; an output layer of the early warning model outputs a probability value of the occurrence of the identified vibration event; Use the monitoring and early warning sample database to train the early warning model; The received sound wave signal is classified through the trained early warning model, the type of vibration event is identified, and the predicted value of the probability of the corresponding vibration event is output; Set the confidence threshold c. If the predicted value is greater than c, a warning signal is generated. The type of vibration event and the warning signal are sent to the host computer.

4. A distributed optical fiber acoustic sensor monitoring and early warning system, characterized in that: The system is used to execute the monitoring and early warning method of a distributed optical fiber acoustic wave sensor as described in any one of claims 1-3.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement a monitoring and early warning method for a distributed optical fiber acoustic wave sensor as described in any one of claims 1 to 3.

Images