A method and system for earthquake prediction based on optical fiber micro-vibration monitoring

Abstract

The application discloses a kind of based on optical fiber microvibration monitoring earthquake prediction method and system, comprising: S1: using dynamic self-adapting acquisition frequency adjustment mechanism to collect microvibration signal, data acquisition is carried out with frequency 1kHz in distributed optical fiber sensing network, simultaneously obtain data fluctuation, set a fluctuation threshold, when microvibration signal fluctuation in continuous m time windows continuously exceeds fluctuation threshold, acquisition frequency is promoted to 10kHz to carry out data acquisition, and all microvibration signal data are counted;S2: the cross-correlation coefficient matrix of microvibration signal data between different acquisition frequencies is constructed by all microvibration signal data, the direction information and the propagation velocity information between different monitoring points are obtained based on the corresponding value in cross-correlation coefficient matrix, and space characteristic parameter is formed, the space propagation characteristics of signal are comprehensively captured, and the one-sidedness problem of traditional method is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of earthquake prediction technology, and in particular to an earthquake prediction method and system based on fiber optic micro-vibration monitoring. Background Technology

[0002] Traditional earthquake monitoring sensors, such as seismographs, have played a crucial role in monitoring seismic activity. However, they are typically discretely distributed, making it difficult to continuously monitor large areas. This means that earthquake information may not be acquired in a timely manner in areas outside the sensor's coverage. Furthermore, pre-earthquake micro-vibration signals are often extremely weak, and traditional monitoring equipment, due to its limited sensitivity, struggles to effectively capture these subtle signals. Yet, these weak signals may contain crucial information about an impending earthquake. The emergence of fiber optic sensing technology has brought new hope to earthquake monitoring. Fiber optics possess extremely high sensitivity, capable of detecting even the smallest vibrations. Simultaneously, distributed fiber optic sensing technology enables continuous monitoring along the fiber optic line, effectively compensating for the limited coverage of traditional sensors. This allows for real-time and comprehensive monitoring of micro-vibrations in the area, providing rich data support for earthquake prediction.

[0003] Currently, pre-earthquake micro-vibration signals are often very weak and complex, encompassing a variety of frequency components and spatial propagation characteristics from extremely low frequencies (e.g., 0.01Hz-1Hz) to high frequencies (10Hz). Traditional methods extract features only from a single perspective in the time or frequency domain, which is one-sided and cannot fully present the complete picture of the signal, making it difficult to capture the characteristic patterns of pre-earthquake anomalous signals. According to relevant research statistics, in actual earthquake monitoring, due to the limitations of traditional methods, about 60% of the characteristics of weak pre-earthquake anomalous signals are not effectively identified. Furthermore, traditional earthquake prediction models lack effective correlation processing and analysis capabilities when faced with high-dimensional and complex earthquake-related data. Therefore, this paper proposes an earthquake prediction method and system based on fiber optic micro-vibration monitoring. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution:

[0005] An earthquake prediction method based on fiber optic micro-vibration monitoring includes:

[0006] S1: A dynamic adaptive acquisition frequency adjustment mechanism is used to acquire micro-vibration signals. Data acquisition is performed at a frequency of 1kHz in a distributed optical fiber sensor network. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed.

[0007] S2: Construct a cross-correlation matrix of micro-vibration signal data between different acquisition frequencies using all micro-vibration signal data, and obtain spatial feature parameters based on the corresponding values ​​in the cross-correlation matrix to obtain the directional information and propagation speed information between different monitoring points.

[0008] S3: The analysis window length is determined based on the frequency range of the seismic waves using time-frequency domain representation. The time-frequency micro-vibration signal is converted into a time-frequency domain representation based on the determined window length using short-time Fourier transform. The time-frequency characteristic parameters are then determined using the time-frequency domain representation of the micro-vibration signal.

[0009] S4: Encode the collected spatial and temporal characteristic parameters into qubits to obtain a qubit sequence, construct an earthquake prediction model based on a quantum neural network, input the qubit sequence into the earthquake prediction model, and output the earthquake prediction results.

[0010] The dynamic adaptive acquisition frequency adjustment mechanism includes a normal acquisition mechanism and an acquisition frequency boosting mechanism.

[0011] The specific implementation process of the dynamic adaptive acquisition frequency adjustment mechanism is as follows:

[0012] Micro-vibration monitoring is performed using a distributed optical fiber sensor network, and the acquisition device converts continuous micro-vibration signals into electrical signals.

[0013] Micro-vibration signal data under normal conditions are collected using a normal acquisition mechanism at a frequency of 1 kHz, represented as V = (v1, v2, v3, ..., v...). n1 ), where n1 is the number of signal data points acquired at a frequency of 1kHz;

[0014] Calculate the fluctuation of the micro-vibration signal at each monitoring point in the signal data acquired by the normal acquisition mechanism;

[0015] A fluctuation threshold γ is set based on the fluctuation situation. The fluctuation situation monitored in real time is compared with the fluctuation threshold. The time window of the micro vibration signal when the amplitude frequency A is greater than the fluctuation threshold γ is counted. When the amplitude frequency A of the micro vibration signal is greater than the fluctuation threshold γ for m consecutive time windows, the current signal fluctuation is determined to be abnormal, and the acquisition frequency enhancement mechanism is triggered.

[0016] The acquisition frequency is increased to 10kHz for data acquisition. The micro-vibration signal data acquired at a frequency of 10kHz is represented as S=(s1, s2, s3, ..., s n2 ), where n2 is the number of signal data points acquired at a frequency of 10kHz.

[0017] The process for determining the fluctuation situation is as follows:

[0018] Set a time window length T, and plot the waveform of the micro-vibration signal collected by the distributed optical fiber sensor network over time within each time window.

[0019] The amplitude frequency A of the waveform is continuously recorded for each time window length. The amplitude frequency A within each time window length T reflects the fluctuation of the micro-vibration signal at each monitoring point.

[0020] The process of constructing the cross-correlation matrix is ​​as follows:

[0021] Discretize the micro-vibration signal data collected at the same monitoring point using different sampling frequencies to obtain discrete signals V[δ] and S[δ], where [δ] is the sampling point number;

[0022] The cross-correlation coefficients of discrete signals collected from different monitoring points are calculated using a computer program. For spatial discrete signals V[δ] and S[δ], their cross-correlation coefficients are calculated using a formula.

[0023]

[0024] Where N is the total number of sampling points for the signal;

[0025] As m consecutive time windows change, a series of cross-correlation values ​​are obtained. All the obtained cross-correlation values ​​are arranged in order to form a correlation coefficient matrix R.

[0026] The process of obtaining the spatial feature parameters is as follows:

[0027] Observe the corresponding values ​​in the cross-correlation matrix;

[0028] The propagation delay τ of the signal is determined by the peak position of the cross-correlation coefficient matrix;

[0029] The direction of signal propagation is obtained by comparing the propagation delay of signals between multiple monitoring points;

[0030] By combining the actual distance between two points, using the formula Calculate the signal propagation speed u and compare the signal propagation speed in different directions;

[0031] The directional and propagation speed information between different monitoring points is statistically analyzed. This data constitutes spatial characteristic parameters. For monitoring point i, the set of spatial characteristic parameters is represented as K = [u...]. i1 , τ i1 u i2 , τ i2 ....].

[0032] The process of obtaining the time-frequency characteristic parameters is as follows:

[0033] Determine the window function length M;

[0034] The window function length in the short-time Fourier transform is determined based on the window length M;

[0035] The short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation, specifically a discretized time-domain signal. A continuous micro-vibration signal within a time domain is determined, and it is sampled to obtain a time-domain discrete signal G[δ], where [δ] is the sampling point number.

[0036] The short-time Fourier transform of the discrete-time signal G[δ] is expressed as follows:

[0037]

[0038] Where t is the time index, f is the frequency index, M is the window function length, and w[tm] is the window function. These are the parameters for the Fourier transform process;

[0039] As the time index t changes, the above calculations are performed on signal segments at different locations to obtain a series of results. These values ​​form a two-dimensional array, which is the time-frequency domain representation of the micro-vibration signal. One dimension corresponds to time t, and the other dimension corresponds to frequency f.

[0040] The time-frequency characteristic parameters are determined by the time-frequency domain representation of the micro-vibration signal. The time-frequency characteristic parameters include the peak frequency f in the time domain, the energy centroid frequency, and the energy proportion E in a specific frequency range.

[0041] The time-frequency characteristic parameters of monitoring point i are expressed as follows:

[0042] The process of obtaining the quantum bit sequence is as follows:

[0043] Normalization preprocessing is performed on spatial feature parameters and time-frequency feature parameters;

[0044] The precision of the qubit encoding is determined based on the preprocessed characteristic parameters;

[0045] The principle of quantum superposition is used for encoding, and each feature parameter is mapped to a different state of the qubit. For feature parameters with values ​​in the range [0,1], they are represented as the superposition of the probabilities of the qubit being in the |0> state and the |1> state. Each parameter is encoded in turn to form a qubit sequence.

[0046] The process of constructing the earthquake prediction model is as follows:

[0047] The number of qubits is determined based on the length of the encoded qubit sequence and the structural requirements of the quantum neural network;

[0048] Earthquake prediction models consist of an input layer, a hidden layer, and an output layer.

[0049] The number of qubits in the input layer is consistent with the total number of qubits in the encoded feature parameters;

[0050] The hidden layer uses quantum gate operations to perform quantum computing;

[0051] The qubits in the output layer correspond to the earthquake prediction results. One qubit is used to represent whether an earthquake has occurred, and the resulting qubit state is mapped to the probability of an earthquake occurring.

[0052] The input layer result of the earthquake prediction model is: the qubit is in a superposition state of |ψ>=α|0>+β|1>, where α represents the probability amplitude of the probability that the earthquake will not occur, β represents the probability amplitude of the probability that the earthquake will occur, and |α| 2 |β| represents the probability that an earthquake will not occur. 2 This indicates the probability of an earthquake occurring.

[0053] An earthquake prediction system based on fiber optic micro-vibration monitoring includes:

[0054] Dynamic signal acquisition module: It adopts a dynamic adaptive acquisition frequency adjustment mechanism to acquire micro-vibration signals. In the distributed optical fiber sensor network, data is acquired at a frequency of 1kHz. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed.

[0055] Spatial feature weighting module: Constructs a cross-correlation matrix of micro-vibration signal data at different acquisition frequencies using all micro-vibration signal data, and obtains the directional and propagation speed information between different monitoring points based on the corresponding values ​​in the cross-correlation matrix to form spatial feature parameters;

[0056] Time-frequency feature weighting module: Based on the frequency range of seismic waves, the analysis window length is determined by the time-frequency domain representation. Short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation based on the determined window length. The time-frequency feature parameters are determined by the time-frequency domain representation of the micro-vibration signal.

[0057] Fusion forecast module: The collected spatial and temporal characteristic parameters are encoded into qubits to obtain a qubit sequence, an earthquake prediction model based on a quantum neural network is constructed, the qubit sequence is input into the earthquake prediction model, and the earthquake prediction results are output.

[0058] The present invention has the following beneficial effects:

[0059] In this invention, firstly, by adopting a dynamic adaptive acquisition frequency adjustment mechanism, micro-vibration signals are acquired at two frequencies, 1kHz and 10kHz, under different signal fluctuation conditions. This allows for the acquisition of richer data. Furthermore, by constructing a cross-correlation matrix and using short-time Fourier transform, feature parameters are extracted from both spatial and time-frequency dimensions. This comprehensively captures the spatial propagation characteristics of the signal (such as directional information and propagation speed information between different monitoring points) as well as multiple frequency components (time-domain frequency peak, energy centroid frequency, and energy proportion of specific frequency bands, etc.), presenting a complete picture of the signal and effectively solving the one-sidedness problem of traditional methods.

[0060] Secondly, an earthquake prediction model based on quantum neural networks was constructed. Spatial and time-frequency characteristic parameters were converted into quantum bit sequences and input into the model using quantum bit encoding. The hidden layers of the quantum neural network use quantum gate operations for quantum computing, which can process information from multiple states simultaneously, uncover complex relationships between data, and effectively process high-dimensional and complex data.

[0061] Finally, by extracting the characteristic parameters of the micro-vibration signal from both spatial and temporal dimensions, the overall picture of the signal is presented. At the same time, by utilizing the quantum state superposition and entanglement characteristics of the quantum neural network model, the correlation between these multi-dimensional characteristic parameters is further analyzed in depth. The spatial propagation characteristics of the signal and the comprehensive parameters of multiple frequency components of the seismic signal are interacted with the quantum neural network to improve the accuracy of earthquake prediction. Attached Figure Description

[0062] Figure 1 This is a flowchart illustrating the steps of an earthquake prediction method and system based on fiber optic micro-vibration monitoring proposed in this invention.

[0063] Figure 2 This is a system block diagram of an earthquake prediction method and system based on fiber optic micro-vibration monitoring proposed in this invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Example 1

[0066] like Figure 1As shown, the present invention proposes an earthquake prediction method based on fiber optic micro-vibration monitoring, comprising:

[0067] S1: A dynamic adaptive acquisition frequency adjustment mechanism is used to acquire micro-vibration signals. Data acquisition is performed at a frequency of 1kHz in a distributed optical fiber sensor network. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed.

[0068] The dynamic adaptive acquisition frequency adjustment mechanism includes two acquisition modes: normal acquisition mechanism and acquisition frequency enhancement mechanism.

[0069] The specific implementation process of the dynamic adaptive acquisition frequency adjustment mechanism is as follows:

[0070] Micro-vibration monitoring is carried out using a distributed optical fiber sensor network. The optical fiber serves as the sensing medium. When a small vibration occurs underground or is disturbed by the external environment, the optical fiber will sense the corresponding vibration. The acquisition device converts these continuous micro-vibration signals into electrical signals.

[0071] Micro-vibration signal data under normal conditions were collected using a normal acquisition mechanism at a frequency of 1 kHz. Statistical analysis was performed on the micro-vibration signal data under normal conditions. The micro-vibration signal data acquired at a frequency of 1 kHz is represented as V = (v1, v2, v3, ..., v...). n1 ), where n1 is the number of signal data points acquired at a frequency of 1kHz;

[0072] Calculate the fluctuation of the micro-vibration signal at each monitoring point in the signal data acquired by the normal acquisition mechanism;

[0073] The process of acquiring the fluctuation of the micro-vibration signal at the monitoring point is as follows:

[0074] Set a time window length T (e.g., 10 seconds), and plot the changes of micro-vibration signals collected by the distributed optical fiber sensor network over time within each time window as a waveform diagram.

[0075] Specifically, waveform diagrams allow for a direct observation of the signal's amplitude and frequency.

[0076] The amplitude frequency A of the waveform is continuously recorded for each time window length. The amplitude frequency A within each time window length T reflects the fluctuation of the micro-vibration signal at each monitoring point.

[0077] A fluctuation threshold γ is set, and the real-time monitored fluctuations are compared with the fluctuation threshold. The time windows of micro-vibration signals where the amplitude frequency A is greater than the fluctuation threshold γ are statistically analyzed. When the amplitude frequency A of the micro-vibration signal is greater than the fluctuation threshold γ for m consecutive time windows, the current signal fluctuation is determined to be abnormal, triggering a sampling frequency boosting mechanism. The sampling frequency is increased to 10kHz for data acquisition. The micro-vibration signal data acquired at a frequency of 10kHz is represented as S=(s1, s2, s3, ..., s...). n2 ), where n2 is the number of signal data points acquired at a frequency of 10kHz;

[0078] Specifically, the fluctuation threshold γ is based on the monitoring of the signal under normal conditions. The amplitude value corresponding to twice the standard deviation of the normal amplitude range is taken as the threshold. This dynamic adaptive acquisition frequency adjustment mechanism dynamically adjusts the data acquisition frequency of the distributed optical fiber sensor network according to the fluctuation of the micro-vibration signal. Under normal conditions, it acquires data at a lower frequency (1kHz) to reduce the amount of data. When the signal fluctuation is abnormal, it increases to a higher frequency (10kHz) to capture more details, thereby achieving high efficiency and accuracy in data acquisition.

[0079] S2: Construct a cross-correlation matrix of micro-vibration signal data between different acquisition frequencies using all micro-vibration signal data, and obtain spatial feature parameters based on the corresponding values ​​in the cross-correlation matrix to obtain the directional information and propagation speed information between different monitoring points.

[0080] Discretize the micro-vibration signal data collected at the same monitoring point using different sampling frequencies to obtain discrete signals V[δ] and S[δ] (where [δ] is the sampling point number);

[0081] This step is to convert the continuous signal into a discrete form that can be processed by a computer. When a small vibration occurs underground or when it is disturbed by the external environment, the optical fiber will sense the corresponding vibration. The acquisition device converts these continuous micro-vibration signals into electrical signals and stores them in a discrete form to obtain discrete signal sequences such as V[δ] and S[δ].

[0082] At the data processing center, computer programs calculate the cross-correlation coefficients of discrete signals collected from different monitoring points. For spatial discrete signals V[δ] and S[δ], the cross-correlation coefficient R is calculated using a formula. m VS :

[0083]

[0084] Where N is the total number of sampling points for the signal;

[0085] By changing over m consecutive time windows, the correlation between two signals under different delays can be obtained. Then, the cross-correlation coefficients between all monitoring points at different acquisition frequencies can be calculated. As the m consecutive time windows change, a series of cross-correlation values ​​are obtained. All the obtained cross-correlation values ​​are arranged in order to form a correlation coefficient matrix R.

[0086] By observing the corresponding values ​​in the cross-correlation coefficient matrix, we can analyze the distribution of the signal correlation coefficients between different monitoring points. If the correlation coefficients between monitoring points in a certain area are generally high, it indicates that the signals in that area have strong consistency and come from the same or related vibration sources. In contrast, areas with low correlation coefficients have poor signal consistency and may be affected by different factors or come from different sources.

[0087] The propagation delay τ of the signal is determined based on the peak position of the cross-correlation matrix. Using the correlation matrix, the propagation delays of the signal between multiple monitoring points are compared. For three monitoring points A, B, and C, the known propagation delays τ from a source point to these three points are respectively... A τ B τ C By comparing the magnitude and differences of these time delays, the direction of signal propagation can be preliminarily determined. For example, if τ A >τ B >τ C The signal then comes from the direction closest to A;

[0088] Then, combining the actual distance between the two points, we can use the formula... Calculate the signal propagation speed u and compare the signal propagation speed in different directions;

[0089] Specifically, if the propagation speed gradually changes in a certain direction, it means that the properties of the underground medium are different in that direction, or that the signal is affected by different factors during propagation. For example, if the propagation speed gradually decreases, it may be because the density or elastic modulus of the underground medium has changed, which hinders the signal propagation.

[0090] The directional and propagation speed information between different monitoring points is statistically analyzed. This data constitutes spatial characteristic parameters. For monitoring point i, the set of spatial characteristic parameters is represented as K = [u...]. i1 , τ i1 u i2 , τ i2 ....];

[0091] These spatial characteristic parameters can help researchers understand the propagation direction and speed of underground vibrations, as well as the correlation between vibrations at different locations, thereby inferring changes in underground geological structures or potential signs of seismic activity.

[0092] S3: The analysis window length is determined based on the frequency range of the seismic waves using time-frequency domain representation. The time-frequency micro-vibration signal is converted into a time-frequency domain representation based on the determined window length using short-time Fourier transform. The time-frequency characteristic parameters are then determined using the time-frequency domain representation of the micro-vibration signal.

[0093] The process of determining the window function length M is as follows:

[0094] For low-frequency components (e.g., 1Hz) in the seismic wave frequency range, a larger analysis window length is needed to improve frequency resolution in order to accurately analyze these low-frequency components. For a 1Hz signal, according to the frequency resolution formula above, if the sampling frequency f... s =1000Hz. In order to distinguish at least a 1Hz frequency difference, the window length must be at least 1000 sampling points. If the sampling interval is 1ms, then the corresponding window length is 1s.

[0095] When the seismic wave frequency range includes high-frequency components (such as 10Hz), a shorter analysis window length is needed to ensure time resolution in order to capture rapid changes in the high-frequency signal. For example, for a 10Hz signal, if the sampling frequency f... s =1000Hz, considering the time resolution requirements, the window length is set to 100 sampling points, corresponding to 0.1s;

[0096] The window function length in the short-time Fourier transform is determined based on the window length M. The window function length is equal to the analysis window length because the window function is designed to weight the signal within the selected window to ensure that each signal sample within the window is appropriately weighted. For example, if the determined analysis window length is 500 sampling points, then the window function length is also set to 500 sampling points. Let the determined window function length be M.

[0097] The short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation, specifically a discretized time-domain signal. A continuous micro-vibration signal within the time domain is identified, and it is sampled to obtain a discrete time-domain signal G[δ], where [δ] is the sampling point number. The sampling process satisfies the Nyquist sampling theorem, i.e., the sampling frequency f... s It must be greater than or equal to the highest frequency f of the signal. max Twice as much;

[0098] The short-time Fourier transform of the discrete-time signal G[δ] is expressed as follows:

[0099]

[0100] Where t is the time index. Here, M is the frequency index, M is the window function length, and w[tm] is the window function. These are the parameters for the Fourier transform process;

[0101] As the time index t changes, the above calculations are performed on signal segments at different locations to obtain a series of results. These values ​​form a two-dimensional array, which represents the micro-vibration signal in the time-frequency domain. One dimension corresponds to time t, and the other dimension corresponds to frequency.

[0102] The process of determining time-frequency characteristic parameters through the time-frequency domain representation of micro-vibration signals is as follows:

[0103] Time-frequency characteristic parameters include the peak time-domain frequency determined by time-frequency domain representation. The energy centroid frequency and the energy percentage E in a specific frequency band;

[0104] Time-domain frequency peak Acquisition process:

[0105] Obtained from a series After constructing a two-dimensional array of values ​​(time-frequency domain representation), at each time point in the time-frequency domain representation, the search column is used to... The value is obtained by comparing the magnitudes of all frequency indices f, based on the highest energy value. The largest The corresponding actual frequency value is the peak frequency in the time domain. It reflects the most dominant vibration frequency in the signal at that moment;

[0106] The process of obtaining the energy center of gravity frequency F:

[0107] Based on formula The formula is obtained by finding the sum of the products of frequency and the square of energy in the numerator and the sum of the squares of energy in the denominator. The energy centroid frequency is obtained by the ratio of the two, which reflects the concentration of signal energy in the frequency domain.

[0108] The process of obtaining the energy percentage E in a specific frequency band:

[0109] In a two-dimensional array with time-frequency domain representation Then, for a given frequency band Based on the formula: The numerator is the sum of energy within a specific frequency band, and the denominator is the sum of energy of the entire signal at that moment. The ratio of the two is used to obtain the energy proportion of a specific frequency band, which is used to analyze the energy distribution characteristics of the signal within a specific frequency range and helps to identify frequency components related to earthquakes.

[0110] Specifically, the time-frequency characteristic parameters for monitoring point i are expressed as follows:

[0111]

[0112] S4: Encode the collected spatial and temporal characteristic parameters into qubits to obtain a qubit sequence, construct an earthquake prediction model based on a quantum neural network, input the qubit sequence into the earthquake prediction model, and output the earthquake prediction result;

[0113] The process of obtaining a sequence of qubits is as follows:

[0114] Before encoding the qubits, the spatial characteristic parameters and time-frequency characteristic parameters are normalized and preprocessed to obtain characteristic parameters that combine the spatial characteristic parameters and time-frequency characteristic parameters.

[0115] The precision of the qubit encoding is determined based on the preprocessed characteristic parameters;

[0116] The principle of quantum superposition is used for encoding, mapping each feature parameter to different states of the qubit. For feature parameters with values ​​in the range [0,1], they are represented as the superposition of the probabilities of the qubit being in the |0> state and the |1> state. Each parameter is encoded in turn to form a sequence of qubits.

[0117] Constructing an earthquake prediction model based on quantum neural networks;

[0118] The number of qubits is determined based on the length of the encoded qubit sequence and the structural requirements of the quantum neural network;

[0119] Specifically, if the input layer of a quantum neural network needs to receive all encoded feature parameter information, then the number of qubits must be at least equal to the length of the encoded qubit sequence. Considering the complexity and resource consumption of quantum computing, the number of qubits is optimized through experiments and theoretical analysis. The number of qubits is gradually increased, the training effect and computational efficiency of the model are observed, and the number of qubits that minimizes computational resource consumption while ensuring the accuracy of the model is selected.

[0120] The process of constructing an earthquake prediction model is as follows:

[0121] Earthquake prediction models consist of an input layer, a hidden layer, and an output layer.

[0122] The number of qubits in the input layer is the same as the total number of qubits in the encoded feature parameters, and is used to receive the encoded feature information. The hidden layer uses quantum gate operations to perform quantum computing.

[0123] Specifically, the hidden layer of the earthquake prediction module performs quantum computing through quantum gate operations. These quantum gates transform and entangle the state of the input qubits to achieve feature extraction and information processing. Quantum gates can transform the state of the input qubits. Through a series of combinations and operations of quantum gates, the hidden layer performs layer-by-layer transformation and feature extraction on the input qubit sequence. When processing earthquake-related feature parameters, the hidden layer may extract feature patterns related to the probability of earthquake occurrence and magnitude. These feature patterns are obtained directly through quantum computing on the original spatial and time-frequency feature parameters.

[0124] The qubits in the output layer correspond to the earthquake prediction results. One qubit is used to represent whether an earthquake has occurred (0 means no earthquake, 1 means earthquake). The resulting qubit state is mapped to the probability of an earthquake occurring.

[0125] For example:

[0126] The input layer result of the earthquake prediction model is: the final state of the qubit is a superposition of |ψ>=α|0>+β|1>, where |α| 2 |β| represents the probability that an earthquake will not occur. 2 Let |β| represent the probability of an earthquake occurring. 2 This represents the probability of an earthquake occurring. For example, a quantum bit is in a state of 0.2|0>+0.8|1>, 0.8 2 =0.64, then the model outputs a probability of 64% that an earthquake will occur;

[0127] Furthermore, when the model outputs a probability below 30%, it is considered a low-risk level; a probability between 30% and 70% is considered a medium-risk level; and a probability above 70% is considered a high-risk level. If the model outputs an earthquake occurrence probability of 64%, the corresponding earthquake prediction level is "medium-risk level," indicating that relevant departments and personnel need to activate emergency plans.

[0128] Example 2

[0129] like Figure 2 As shown, the present invention proposes an earthquake prediction system based on fiber optic micro-vibration monitoring, comprising:

[0130] Dynamic signal acquisition module: It adopts a dynamic adaptive acquisition frequency adjustment mechanism to acquire micro-vibration signals. In the distributed optical fiber sensor network, data is acquired at a frequency of 1kHz. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed.

[0131] Spatial feature weighting module: Constructs a cross-correlation matrix of micro-vibration signal data at different acquisition frequencies using all micro-vibration signal data, and obtains the directional and propagation speed information between different monitoring points based on the corresponding values ​​in the cross-correlation matrix to form spatial feature parameters;

[0132] Time-frequency feature weighting module: Based on the frequency range of seismic waves, the analysis window length is determined by the time-frequency domain representation. Short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation based on the determined window length. The time-frequency feature parameters are determined by the time-frequency domain representation of the micro-vibration signal.

[0133] Fusion forecast module: The collected spatial and temporal characteristic parameters are encoded into qubits to obtain a qubit sequence, an earthquake prediction model based on a quantum neural network is constructed, the qubit sequence is input into the earthquake prediction model, and the earthquake prediction results are output.

[0134] In the application, several formulas are calculated by removing dimensions and taking their numerical values. The formulas are established by collecting a large amount of data and simulating the most recent real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so they will not be elaborated here.

[0135] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0136] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An earthquake prediction method based on fiber optic micro-vibration monitoring, characterized in that, include: S1: A dynamic adaptive acquisition frequency adjustment mechanism is used to acquire micro-vibration signals. Data acquisition is performed at a frequency of 1kHz in a distributed optical fiber sensor network. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed. S2: Construct a cross-correlation matrix of micro-vibration signal data between different acquisition frequencies using all micro-vibration signal data, and obtain spatial feature parameters based on the corresponding values ​​in the cross-correlation matrix to obtain the directional information and propagation speed information between different monitoring points. S3: The analysis window length is determined based on the frequency range of the seismic waves using time-frequency domain representation. The time-frequency micro-vibration signal is converted into a time-frequency domain representation based on the determined window length using short-time Fourier transform. The time-frequency characteristic parameters are then determined using the time-frequency domain representation of the micro-vibration signal. S4: Encode the collected spatial and temporal characteristic parameters into qubits to obtain a qubit sequence, construct an earthquake prediction model based on a quantum neural network, input the qubit sequence into the earthquake prediction model, and output the earthquake prediction results.

2. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The dynamic adaptive acquisition frequency adjustment mechanism includes a normal acquisition mechanism and an acquisition frequency boosting mechanism.

3. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 2, characterized in that, The specific implementation process of the dynamic adaptive acquisition frequency adjustment mechanism is as follows: Micro-vibration monitoring is performed using a distributed optical fiber sensor network, and the acquisition device converts continuous micro-vibration signals into electrical signals. Micro-vibration signal data under normal conditions are collected using a normal acquisition mechanism at a frequency of 1 kHz, represented as V = (v1, v2, v3, ..., v...). n1 ), where n1 is the number of signal data points acquired at a frequency of 1kHz; Calculate the fluctuation of the micro-vibration signal at each monitoring point in the signal data acquired by the normal acquisition mechanism; A fluctuation threshold γ is set based on the fluctuation situation. The fluctuation situation monitored in real time is compared with the fluctuation threshold. The time window of the micro vibration signal when the amplitude frequency A is greater than the fluctuation threshold γ is counted. When the amplitude frequency A of the micro vibration signal is greater than the fluctuation threshold γ for m consecutive time windows, the current signal fluctuation is determined to be abnormal, and the acquisition frequency enhancement mechanism is triggered. The acquisition frequency is increased to 10kHz for data acquisition. The micro-vibration signal data acquired at a frequency of 10kHz is represented as S=(s1, s2, s3, ..., s n2 ), where n2 is the number of signal data points acquired at a frequency of 10kHz.

4. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 3, characterized in that, The process for determining the fluctuation situation is as follows: Set a time window length T, and plot the waveform of the micro-vibration signal collected by the distributed optical fiber sensor network over time within each time window. The amplitude frequency A of the waveform is continuously recorded for each time window length. The amplitude frequency A within each time window length T reflects the fluctuation of the micro-vibration signal at each monitoring point.

5. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The process of constructing the cross-correlation matrix is ​​as follows: Discretize the micro-vibration signal data collected at the same monitoring point using different sampling frequencies to obtain discrete signals V[δ] and S[δ], where [δ] is the sampling point number; The cross-correlation coefficients of discrete signals collected from different monitoring points are calculated using a computer program. For spatial discrete signals V[δ] and S[δ], their cross-correlation coefficients R are calculated using a formula. m VS : Where N is the total number of sampling points for the signal; As m consecutive time windows change, a series of cross-correlation values ​​are obtained. All the obtained cross-correlation values ​​are arranged in order to form a correlation coefficient matrix R.

6. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The process of obtaining the spatial feature parameters is as follows: Observe the corresponding values ​​in the cross-correlation matrix; The propagation delay τ of the signal is determined by the peak position of the cross-correlation coefficient matrix; The direction of signal propagation is obtained by comparing the propagation delay of signals between multiple monitoring points; By combining the actual distance between two points, using the formula Calculate the signal propagation speed u and compare the signal propagation speed in different directions; The directional and propagation speed information between different monitoring points is statistically analyzed. This data constitutes spatial characteristic parameters. For monitoring point i, the set of spatial characteristic parameters is represented as K = [u...]. i1 , τ i1 u i2 , τ i2 ....].

7. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The process of obtaining the time-frequency characteristic parameters is as follows: Determine the window function length M; The window function length in the short-time Fourier transform is determined based on the window length M; The short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation, specifically a discretized time-domain signal. A continuous micro-vibration signal within a time domain is determined, and it is sampled to obtain a time-domain discrete signal G[δ], where [δ] is the sampling point number. The short-time Fourier transform of the discrete-time signal G[δ] is expressed as follows: Where t is the time index, f is the frequency index, M is the window function length, and w[tm] is the window function. These are the parameters for the Fourier transform process; As the time index t changes, the above calculations are performed on signal segments at different locations to obtain a series of results. These values ​​form a two-dimensional array, which is the time-frequency domain representation of the micro-vibration signal. One dimension corresponds to time t, and the other dimension corresponds to frequency f. The time-frequency characteristic parameters are determined by representing the micro-vibration signal in the time-frequency domain. These parameters include the peak frequency in the time domain. The energy centroid frequency and the energy percentage E in a specific frequency band; The time-frequency characteristic parameters of monitoring point i are expressed as follows:

8. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The process of obtaining the quantum bit sequence is as follows: Normalization preprocessing is performed on spatial feature parameters and time-frequency feature parameters; The precision of the qubit encoding is determined based on the preprocessed characteristic parameters; The principle of quantum superposition is used for encoding, and each feature parameter is mapped to a different state of the qubit. For feature parameters with values ​​in the range [0,1], they are represented as the superposition of the probabilities of the qubit being in the |0> state and the |1> state. Each parameter is encoded in turn to form a qubit sequence.

9. The earthquake prediction method based on fiber optic micro-vibration monitoring according to claim 1, characterized in that, The process of constructing the earthquake prediction model is as follows: The number of qubits is determined based on the length of the encoded qubit sequence and the structural requirements of the quantum neural network; Earthquake prediction models consist of an input layer, a hidden layer, and an output layer. The number of qubits in the input layer is consistent with the total number of qubits in the encoded feature parameters; The hidden layer uses quantum gate operations to perform quantum computing; The qubits in the output layer correspond to the earthquake prediction results. One qubit is used to represent whether an earthquake has occurred, and the resulting qubit state is mapped to the probability of an earthquake occurring. The input layer result of the earthquake prediction model is: the qubit is in a superposition state of |Ψ>=α|0>+β|1>, where α represents the probability amplitude of the probability that the earthquake will not occur, β represents the probability amplitude of the probability that the earthquake will occur, and |α| 2 |β| represents the probability that an earthquake will not occur. 2 This indicates the probability of an earthquake occurring.

10. An earthquake prediction system based on fiber optic micro-vibration monitoring, using the method described in any one of claims 1 to 9, characterized in that, include: Dynamic signal acquisition module: It adopts a dynamic adaptive acquisition frequency adjustment mechanism to acquire micro-vibration signals. In the distributed optical fiber sensor network, data is acquired at a frequency of 1kHz. At the same time, the data fluctuation is acquired. A fluctuation threshold is set. When the fluctuation of the micro-vibration signal exceeds the fluctuation threshold for m consecutive time windows, the acquisition frequency is increased to 10kHz for data acquisition, and all micro-vibration signal data are statistically analyzed. Spatial feature weighting module: Constructs a cross-correlation matrix of micro-vibration signal data at different acquisition frequencies using all micro-vibration signal data, and obtains the directional and propagation speed information between different monitoring points based on the corresponding values ​​in the cross-correlation matrix to form spatial feature parameters; Time-frequency feature weighting module: Based on the frequency range of seismic waves, the analysis window length is determined by the time-frequency domain representation. Short-time Fourier transform is used to convert the time-domain micro-vibration signal into a time-frequency domain representation based on the determined window length. The time-frequency feature parameters are determined by the time-frequency domain representation of the micro-vibration signal. Fusion forecast module: The collected spatial and temporal characteristic parameters are encoded into qubits to obtain a qubit sequence, an earthquake prediction model based on a quantum neural network is constructed, the qubit sequence is input into the earthquake prediction model, and the earthquake prediction results are output.

Images