A distributed acoustic sensing construction event recognition method based on intrinsic physical features

Abstract

This invention discloses a method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing (DAS) technology. The method first initializes parameters and acquires the DAS signal sampling data sequence to be processed. Then, it preprocesses the data from each channel, calculates the signal-to-noise ratio (SNR) of each channel, filters effective channels, eliminates abnormal data channels, and determines whether a significant construction event exists based on the number of effective channels. Next, it extracts the pulse density feature value of the effective data and determines whether it is a hammer event. If the condition is not met, it extracts feature values ​​such as the proportion of characteristic frequency band components, second harmonic content, and duty cycle to determine whether it is a cutting machine event. If the condition is still not met, it determines whether it is a pneumatic pick event through the tailing coefficient feature value, ultimately achieving accurate identification of road construction events.

Description

Technical Field

[0001] This invention belongs to the field of distributed acoustic sensing technology, and particularly relates to a distributed acoustic sensing construction event identification method based on inherent physical characteristics. Background Technology

[0002] With the accelerated pace of infrastructure construction and urbanization, the frequency of various construction activities has increased significantly, placing higher demands on the compliance supervision and intelligent identification of construction events. Due to the complexity of the construction environment, multiple construction tools often operate simultaneously, generating intertwined vibration and other signals, making accurate identification and risk assessment of specific construction events extremely difficult. Existing monitoring methods are mainly implemented in two scenarios: first, emergency response after accidents caused by illegal construction, where the risk has already translated into actual losses, causing not only economic damage but also attracting public attention; second, illegal construction leading to sensor threshold alarms or resident complaints, where this method suffers from large monitoring blind spots, delayed response, and inability to distinguish event types, making real-time intervention difficult. This passive response status quo highlights the urgent need for intelligent and discriminative sensing of the construction process. Distributed acoustic sensing (DAS) technology can transform ordinary communication optical fibers into continuously distributed sensor arrays, achieving ultra-long-distance, high spatial resolution, and real-time continuous sensing of vibration signals along the construction line, providing an ideal technological foundation for large-scale, uninterrupted supervision of construction activities. Therefore, researching DAS methods that can accurately identify specific construction tool incidents is of great significance for improving construction safety and supervision.

[0003] Existing methods for monitoring construction vibration or events mainly include: (1) Threshold alarm method based on traditional point sensors. This method is deployed at key points and triggers an alarm when the signal strength exceeds a preset threshold. Although it can react to significant vibrations, it cannot identify the type of construction tools. At the same time, its limited deployment cannot achieve full coverage of the area and is prone to forming regulatory blind spots. (2) General vibration monitoring method based on DAS. This method uses DAS to achieve large-scale vibration detection and positioning, but it usually only stays at the level of "vibration" and cannot determine whether the monitored construction is compliant, so that regulatory personnel still need to go to the site to verify. (3) DAS detection method based on deep learning. This method uses AI algorithms to automatically extract DAS signal features for separating and classifying mixed events. It is widely used in traffic monitoring and geological exploration, but this method requires a large amount of labeled data for training, consumes a lot of computing resources, and has poor model interpretability, and cannot process large-scale data in real time. Summary of the Invention

[0004] The purpose of this invention is to provide a distributed acoustic sensing construction event identification method based on inherent physical characteristics, so as to solve the technical problems of low accuracy and poor real-time performance in construction event identification.

[0005] Technical Solution: To solve the above-mentioned technical problems, this invention proposes a distributed acoustic sensing construction event identification method based on inherent physical characteristics, comprising the following steps:

[0006] Step 1: Parameter initialization, obtain the DAS signal sampling data sequence to be processed. , l is the discrete-time index, n is the channel number, L is the number of time sampling points, and N is the number of sensor channels;

[0007] Step 2: Sample data sequence of the DAS signal to be processed Preprocessing is required;

[0008] Step 3: Calculate the signal-to-noise ratio of each channel of the preprocessed data. Remove abnormal data channels and count the number of valid channels. If the number of valid channels If no obvious construction event is found, the identification process ends; otherwise, proceed to step 4.

[0009] Step 4: Extract the pulse density feature parameters of each effective channel and calculate the average pulse density K of all effective channels;

[0010] Step 5: Determine whether the average pulse density K is lower than the threshold. If yes, it is determined to be a hammer event, the identification is completed, and the identification process ends; otherwise, proceed to step 6.

[0011] Step 6: Extract the characteristic frequency band component ratio, duty cycle, and second harmonic content of each effective channel, and calculate the average characteristic frequency band component ratio (UHF), average duty cycle (D), and average second harmonic content (HI) of all effective channels respectively.

[0012] Step 7: Based on the average pulse density K, the average proportion of characteristic frequency band components UHF, the average duty cycle D, and the average second harmonic content HI, jointly determine the event type. If the conditions are met, it is determined to be a cutting machine event, the identification is completed, and the identification process ends; otherwise, proceed to step 8.

[0013] Step 8: Extract the tail coefficient feature parameters of each effective channel and calculate the average tail coefficient of all effective channels. ;

[0014] Step 9: Determine the event type based on the average pulse density K and the average tail coefficient TF. Compare the two with preset thresholds. If the average tail coefficient TF is greater than the preset threshold and the average pulse density K is less than the preset threshold, the event is identified as a hammer event, the identification process is completed, and the identification process ends. Otherwise, the event is identified as a pneumatic pick event, the identification process is completed, and the identification process ends.

[0015] Furthermore, in step 1, the parameters are initialized using the following method to obtain the DAS signal sampling data sequence to be processed. Specifically, it includes the following steps:

[0016] Step 1.1: Initialize the parameters for construction event identification, specifically including the initialization of the following parameters:

[0017] The number of time sampling points L is initialized to Even numbers within the range;

[0018] The number of sensor channels N is initialized to Integers within the range;

[0019] upper limit frequency of bandpass filter Initialize to Real numbers in the Hz range;

[0020] lower cutoff frequency of bandpass filter Initialize to Real numbers in the Hz range;

[0021] The displacement corresponding to a cumulative probability of 0.5 under a standard Gaussian distribution. Initialize to ;

[0022] Lower limit for determining the number of effective channels Initialize to Integers within the range;

[0023] Initial screening signal-to-noise ratio threshold for active channels Initialize to Real numbers within the range;

[0024] Minimum signal-to-noise ratio threshold for active channels Initialize to Real numbers within the range;

[0025] Sampling frequency Initialize to Real numbers in the kHz range;

[0026] Initial value of effective pulse history index Initialize to ;

[0027] Minimum time window between adjacent pulses initialization Real numbers within the range s;

[0028] Minimum relative amplitude threshold of pulse spike Initialize to Real numbers within the range;

[0029] Upper limit of extremely slow pulse density judgment value Initialize to Real numbers in the Hz range;

[0030] Lower limit of effective excitation continuity amplitude Initialize to Real numbers within the range;

[0031] Lower limit of characteristic frequency band Initialize to Integers within the Hz range;

[0032] Lower limit of effective frequency band for harmonic detection Initialize to Integers within the Hz range;

[0033] Upper limit of effective frequency band for harmonic detection Initialize to Integers within the Hz range;

[0034] Spectral peak relative prominence Initialize to Real numbers within the range;

[0035] Harmonic Coupling Matching Frequency Tolerance Initialize to Integers within the Hz range;

[0036] Baseband determination upper limit Initialize to Real numbers in the Hz range;

[0037] Lower limit for judging ultrafast pulse density Initialize to Real numbers in the Hz range;

[0038] Lower limit for continuous duty cycle determination Initialize to Real numbers within the range;

[0039] Lower limit for determining the proportion of characteristic frequency bands Initialize to Real numbers within the range;

[0040] Lower limit for harmonic matching index determination Initialize to Real numbers within the range;

[0041] Envelope start and end relative amplitude threshold Initialize to Real numbers within the range;

[0042] Lower limit for waveform tailing coefficient determination Initialize to Real numbers within the range;

[0043] Upper limit of pulse density judgment in the fuzzy region Initialize to Real numbers in the Hz range;

[0044] Step 1.2: Obtain the DAS signal sampling data sequence to be processed. The real-time data acquired simultaneously from L sampling points by N sensors in an array is used as the data sequence to be processed. Alternatively, DAS signal data containing L sampling points collected by N sensors can be extracted from the memory as a data sequence to be processed.

[0045] Furthermore, in step 2, the DAS signal data sequence is sampled using the following method. Preprocessing includes the following steps:

[0046] Step 2.1, Design frequency range is A bandpass filter for sampling DAS signal data sequences Filtering is performed to obtain filtered data. ;

[0047] Step 2.2: For each channel, based on the filtered data... Calculate its noise floor standard deviation :

[0048]

[0049] in, Indicates taking the median of This indicates the absolute value operation, where the coefficient is... This is the displacement constant initialized in step 1.1.

[0050] Furthermore, in step 3, the signal-to-noise ratio of each channel of the preprocessed data is calculated using the following method. Remove abnormal data channels and count the number of valid channels. If the number of valid channels If no obvious construction event is found, the identification process ends; otherwise, proceed to step 4, which includes the following steps:

[0051] Step 3.1, Calculate the channel Preprocessed data signal-to-noise ratio :

[0052]

[0053] in, This is an operation to find the maximum value of a set;

[0054] Step 3.2, for For each channel, calculate the channel validity indicator variable. :

[0055]

[0056] Step 3.3: Based on the aforementioned channel validity determination indicator variable Determine whether to retain the current channel data. If so, then retain the data from that channel. Otherwise, discard the data in that channel, resulting in a valid number of channels. The remaining valid channels will be renumbered sequentially according to their original order as follows: ;

[0057] Step 3.4: Determine the number of valid channels. The value: If If no obvious construction event is found in the current DAS signal sampling data sequence to be processed, the identification process ends; If so, proceed to step 3.5;

[0058] Step 3.5: For each valid channel after renumbering The channel data is normalized by using its corresponding noise floor standard deviation to obtain normalized data. :

[0059] .

[0060] Furthermore, in step 4, based on the effective channels obtained in step 3, the pulse density feature parameters of each effective channel are extracted, and the average pulse density of all effective channels is calculated. :

[0061]

[0062] in, The sampling frequency initialized in step 1, The number of time sampling points, The number of effective channels, The total number of valid pulses that satisfy the constraints is calculated as follows:

[0063]

[0064] in, The valid pulse determination indicator function is defined as follows:

[0065]

[0066] in, This refers to the minimum time window for adjacent pulses initialized in step 1. For the current sampling point The previous valid pulse time index is defined by the current time. Previous valid pulse history index set Perform the calculation:

[0067]

[0068]

[0069] in, The initial value of the valid pulse history index initialized in step 1; Let be the local extremum determination function, defined as:

[0070]

[0071] in, The minimum relative amplitude threshold of the pulse spike initialized in step 1, Calculations are performed using absolute values.

[0072] Furthermore, in step 5, based on the average pulse density of all effective channels... To determine the event type, specifically:

[0073] Define the initial indicator variable for the hammer event. :

[0074]

[0075] in, The upper limit value for the extremely slow pulse density initialized in step 1 is determined; if If the event is identified as a hammer event, the identification process is completed and ends; otherwise, proceed to step 6.

[0076] Furthermore, step 6 specifically includes the following steps:

[0077] Step 6.1: Calculate the normalized data for all valid channels. average duty cycle :

[0078]

[0079] in, The number of effective channels, The indicator function for determining continuous force application is defined as follows:

[0080]

[0081] in, The lower limit of the effective excitation continuity amplitude initialized in step 1;

[0082] Step 6.2: For each valid channel Calculate the single-sided magnitude spectrum after its discrete Fourier transform. and total energy across the entire frequency band :

[0083] Calculate the discrete Fourier transform sequence :

[0084]

[0085] in, For discrete frequency point indexing, The imaginary unit, ;

[0086] Calculate the bilateral magnitude spectrum corresponding to the discrete Fourier transform sequence. :

[0087]

[0088] Extracting the positive frequency portion of the bilateral amplitude spectrum to obtain the single-sided amplitude spectrum. And force its DC component to zero:

[0089]

[0090] Calculate the single-sided amplitude spectrum Corresponding physical frequency axis sequence :

[0091]

[0092] Calculate the total energy of the single-sided amplitude spectrum within the effective frequency band. :

[0093]

[0094] Step 6.3: Calculate the average proportion of characteristic frequency bands for all effective channels. :

[0095]

[0096] in, Number of valid channels; The proportion of characteristic frequency bands in each effective channel is calculated as follows:

[0097]

[0098] in, The characteristic frequency band indication function is defined as follows:

[0099]

[0100] in, The lower limit value of the characteristic band frequency initialized in step 1;

[0101] Step 6.4: Calculate the average second harmonic content of all effective channels. :

[0102] For each valid channel Calculate the local total energy within the effective frequency band of harmonic detection. :

[0103]

[0104] in, The set of frequency indexes within the effective frequency band for harmonic detection is defined as:

[0105]

[0106] in, and These are the lower and upper limits of the effective frequency band for harmonic detection initialized in step 1, respectively.

[0107] For sets Calculate the peak prominence of any local maximum frequency point within the range. :

[0108]

[0109] in, These are the amplitudes of the local minimum points adjacent to the local maximum frequency point on the left and right sides, respectively.

[0110] In the set Internal search for valid spectral peaks, define a valid spectral peak index set. for:

[0111]

[0112] in, The relative prominence of the spectral peaks initialized in step 1;

[0113] like If the frequency is not empty, extract the frequency corresponding to the effective spectral peak with the largest amplitude as the suspected fundamental frequency. Its corresponding frequency index is denoted as And define candidate indicator variables for second harmonics. :

[0114]

[0115] in, The harmonic coupling matching frequency tolerance initialized in step 1; It is an existential quantifier;

[0116] Calculate the second harmonic content of each effective channel. :

[0117]

[0118] in, This is the upper limit value for determining the base frequency initialized in step 1; Number of valid channels; and They represent sets respectively The amplitudes of the first and second largest effective spectral peaks in the intermediate amplitude range, if there is only one element in this set, then ; It is an empty set;

[0119] The average value of the second harmonic content The calculation is as follows:

[0120]

[0121] Furthermore, in step 7, based on the average pulse density... Average percentage of characteristic frequency band components Average duty cycle Average value of second harmonic content The joint determination of the event type includes the following steps:

[0122] Define the initial indicator variable for the cutting machine event. :

[0123]

[0124] in, The lower limit value for determining the extremely fast pulse density initialized in step 1. The lower limit value for determining the continuous duty cycle initialized in step 1. The lower limit value for determining the proportion of characteristic frequency bands initialized in step 1. The lower limit value is determined for the harmonic matching index initialized in step 1; if If the event is identified as a cutting machine event, the identification process is completed and ends; otherwise, proceed to step 8.

[0125] Furthermore, in step 8, the average tailing coefficient of all valid channels is calculated. Specifically, it includes the following steps:

[0126] Step 8.1: For each valid channel If the channel satisfies the constraint conditions, the total number of valid pulses Then calculate the global maximum amplitude of the channel. and its corresponding time index :

[0127]

[0128]

[0129] in, To return the index function of the argument that maximizes the objective function;

[0130] like There is no need to calculate the global maximum amplitude and time index of the channel; simply execute step 8.3 for the channel.

[0131] Step 8.2: Define the candidate index set for truncating the edge to the left of the point with the largest amplitude value. Truncating the candidate index set with the right edge :

[0132]

[0133]

[0134] Calculate the starting index of the edge truncation respectively Termination index truncated with the right edge :

[0135]

[0136]

[0137] in, The relative amplitude thresholds for the start and end of the envelope initialized in step 1; This is an operation to find the minimum value of a set;

[0138] Step 8.3: Define the waveform tailing coefficient for each valid channel. :

[0139]

[0140] Step 8.4: Average the waveform tailing coefficient. The calculation is as follows:

[0141]

[0142] Furthermore, in step 9, based on the average pulse density... and the average tailing coefficient The joint determination of event types includes the following steps: defining indicator variables for the hammer and jackhammer events. :

[0143]

[0144] in, The lower limit value is determined for the waveform tailing coefficient initialized in step 1. The upper limit value for determining the pulse density of the ambiguous region initialized in step 1; if If the event is identified as a hammer event, the identification process is completed and ends; otherwise, it is identified as a jackhammer event, the identification process is completed and ends.

[0145] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0146] 1. This invention utilizes the abnormal noise level characteristics when the DAS channel coupling is abnormal or continuous abnormal shaking occurs. Through signal-to-noise ratio analysis, abnormal DAS channel signals are effectively identified and eliminated, while valid signals are retained, as shown in step 3, thereby improving the robustness of construction event identification.

[0147] 2. This invention extracts multiple feature values ​​from the data, including pulse density, tail coefficient, second harmonic content, characteristic frequency band component ratio, and duty cycle, as shown in steps 4, 6, and 8, thereby improving the accuracy and anti-interference characteristics of construction event identification, and thus enabling long-distance construction event identification.

[0148] 3. This invention adopts a hierarchical and progressive feature decision architecture, which makes layered discriminations in the order of pulse density to duty cycle, proportion of characteristic frequency band components, second harmonic content to tail coefficient, as shown in steps 5, 7 and 9. This enables the rapid screening of typical events first, and then the fine differentiation of fuzzy samples. It eliminates the need for full feature calculation, which significantly reduces the amount of computation and latency to a certain extent, and can meet the real-time monitoring needs of construction events.

[0149] 4. This invention utilizes the all-weather, wide-range, and high-resolution sensing characteristics of the DAS system, combined with the unique physical characteristics of construction events, to achieve automatic identification of multiple types of construction events. It has low computational load, high identification accuracy, and can realize long-distance real-time monitoring. It is effectively adapted to the actual application scenarios of engineering sites and improves the intelligence and efficiency of road construction event management. Attached Figure Description

[0150] Figure 1 This is a schematic flowchart of the method of the present invention;

[0151] Figure 2 This is a partial channel time-domain waveform diagram of the DAS signal sampling data sequence in Example 1;

[0152] Figure 3 This is a local channel time-domain waveform diagram of the preprocessed signal in Example 1;

[0153] Figure 4 The signal-to-noise ratio diagram for each channel in Example 1 is shown.

[0154] Figure 5 This is a pulse density diagram of each effective channel in Example 1;

[0155] Figure 6 This is a diagram showing the proportion of characteristic frequency band components in each effective channel of Example 1;

[0156] Figure 7 This is a duty cycle diagram for each effective channel in Example 1;

[0157] Figure 8 This is a diagram showing the second harmonic content in each effective channel of Example 1;

[0158] Figure 9 This is a partial channel time-domain waveform diagram of the DAS signal sampling data sequence in Example 2;

[0159] Figure 10 This is a partial channel time-domain waveform diagram of the preprocessed signal in Example 2;

[0160] Figure 11 The signal-to-noise ratio diagrams for each channel in Example 2 are shown below.

[0161] Figure 12 This is a pulse density diagram of each effective channel in Example 2;

[0162] Figure 13 This is a diagram showing the proportion of characteristic frequency band components in each effective channel of Example 2;

[0163] Figure 14 This is a duty cycle diagram for each effective channel in Example 2;

[0164] Figure 15 This is a diagram showing the second harmonic content in each effective channel of Example 2;

[0165] Figure 16 This is a diagram of the trailing coefficients in each effective channel of Example 2. Detailed Implementation

[0166] To better understand the purpose, structure, and function of this invention, the following detailed description, in conjunction with the accompanying drawings, provides an explanation of a distributed acoustic sensing identification method based on inherent physical characteristics proposed in this invention.

[0167] Implementation Example 1

[0168] On-site verification of DAS data collection and construction event identification was conducted at Nanjing Jinling Institute of Technology.

[0169] The following section of DAS data, consisting of 485 DAS channels and lasting 60 seconds, is selected for construction event identification:

[0170] Based on step 1, set the sampling frequency. kHz, upper limit frequency of bandpass filter Hz, lower cutoff frequency of bandpass filter Hz, under a standard Gaussian distribution, the cumulative probability is The displacement corresponding to time Robust estimation constant for background noise Lower limit for determining the number of effective channels Initial screening signal-to-noise ratio threshold for active channels Minimum signal-to-noise ratio threshold for active channels Lower limit of effective excitation continuity amplitude Minimum relative amplitude threshold of pulse spike Minimum time window of adjacent pulses s, relative amplitude threshold of envelope start and end Lower limit of characteristic frequency band Hz, lower limit of effective frequency band for harmonic detection Hz, upper limit of effective frequency band for harmonic detection Hz, relative prominence of spectral peaks Harmonic coupling matching frequency tolerance Hz, upper limit for determining extremely slow pulse density Lower limit for determining extremely fast pulse density Hz, upper limit of base frequency determination Hz, upper limit for determining pulse density in the fuzzy region Hz, lower limit for continuous duty cycle determination Lower limit for determining the proportion of characteristic frequency bands Lower limit for harmonic matching index determination Lower limit for waveform tailing coefficient determination Read the local channel time-domain waveform of the DAS signal sampling data sequence, such as Figure 2 As shown;

[0171] Based on step 2, the designed frequency range is: A bandpass filter for sampling DAS signal data sequences Perform filtering and process the filtered data. Calculate the standard deviation of the noise floor The preprocessed data results are as follows Figure 3 As shown;

[0172] Based on step 3, calculate the signal-to-noise ratio of each channel. The result is as follows Figure 4 As shown, remove elements with a signal-to-noise ratio greater than the upper threshold. and less than the lower threshold After obtaining the data, the number of effective channels is determined. The normalized result of the noise floor was calculated. ;

[0173] Based on step 4, calculate the average pulse density of all effective channels. , and thus The result is as follows Figure 5 As shown;

[0174] Based on step 5, since the average pulse density is greater than the upper limit for judging extremely slow pulse density, proceed to step 6;

[0175] Based on step 6, calculate the eigenvalues ​​of each effective channel and take the average value, as well as the average proportion of the characteristic frequency band components. Average duty cycle Average value of second harmonic content The result is as follows Figure 6 , Figure 7 and Figure 8 As shown;

[0176] Based on step 7, the feature values ​​and thresholds are compared and a joint judgment is made, since the average proportion of feature frequency band components is considered. The lower limit for determining the proportion of frequency bands above the characteristic frequency band Therefore, the construction event in this implementation example is determined to be a cutting machine event, the identification is completed, and the identification process ends.

[0177] This data was collected 4 m around the cutting machine during operation, and this method successfully identified the type of construction event.

[0178] Implementation Example 2:

[0179] On-site verification of DAS data collection and construction event identification was conducted at Nanjing Jinling Institute of Technology.

[0180] The following section of DAS data, consisting of 485 DAS channels and lasting 60 seconds, is selected for construction event identification:

[0181] Based on step 1, set the sampling frequency. kHz, upper limit frequency of bandpass filter Hz, lower cutoff frequency of bandpass filter Hz, under a standard Gaussian distribution, the cumulative probability is The displacement corresponding to time Robust estimation constant for background noise Lower limit for determining the number of effective channels Initial screening signal-to-noise ratio threshold for active channels Minimum signal-to-noise ratio threshold for active channels Lower limit of effective excitation continuity amplitude Minimum relative amplitude threshold of pulse spike Minimum time window of adjacent pulses s, relative amplitude threshold of envelope start and end Lower limit of characteristic frequency band Hz, lower limit of effective frequency band for harmonic detection Hz, upper limit of effective frequency band for harmonic detection Hz, relative prominence of spectral peaks Harmonic coupling matching frequency tolerance Hz, upper limit for determining extremely slow pulse density Lower limit for determining extremely fast pulse density Hz, upper limit of base frequency determination Hz, upper limit for determining pulse density in the fuzzy region Hz, lower limit for continuous duty cycle determination Lower limit for determining the proportion of characteristic frequency bands Lower limit for harmonic matching index determination Lower limit for waveform tailing coefficient determination Read the local channel time-domain waveform of the DAS signal sampling data sequence, such as Figure 9 As shown;

[0182] Based on step 2, the designed frequency range is: A bandpass filter for sampling DAS signal data sequences Perform filtering and process the filtered data. Calculate the standard deviation of the noise floor The preprocessed data results are as follows Figure 10 As shown;

[0183] Based on step 3, calculate the signal-to-noise ratio of each channel. The result is as follows Figure 11 As shown, remove elements with a signal-to-noise ratio greater than the upper threshold. and less than the lower threshold After obtaining the data, the number of effective channels is determined. The normalized result of the noise floor was calculated. ;

[0184] Based on step 4, calculate the average pulse density of all effective channels. , and thus The result is as follows Figure 12 As shown;

[0185] Based on step 5, since the average pulse density is greater than the upper limit for judging extremely slow pulse density, proceed to step 6;

[0186] Based on step 6, calculate the eigenvalues ​​of each effective channel and take the average value, as well as the average proportion of the characteristic frequency band components. Average duty cycle =0.025, average value of second harmonic content The result is as follows Figure 13 , Figure 14 and Figure 15 As shown;

[0187] Based on step 7, the feature value and the threshold are compared and a joint judgment is made. If the condition is not met, step 8 is executed.

[0188] Based on step 8, calculate the average trailing coefficient of each effective channel. , and thus The result is as follows Figure 16 As shown;

[0189] Based on step 9, according to the average pulse density and the average tailing coefficient The event is compared with a threshold and a joint judgment is made. Since the condition is not met, the construction event in this implementation example is determined to be a pneumatic pick event, the identification is completed, and the identification process ends.

[0190] This data was collected from an area 8 meters around the jackhammer operation site, and this method successfully identified the type of jackhammer incident.

[0191] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A method for identifying construction events using distributed acoustic sensing based on inherent physical characteristics, characterized in that, Includes the following steps: Step 1: Parameter initialization, obtain the DAS signal sampling data sequence to be processed. , l is the discrete-time index, n is the channel number, L is the number of time sampling points, and N is the number of sensor channels; Step 2: Sample data sequence of the DAS signal to be processed Preprocessing is required; Step 3: Calculate the signal-to-noise ratio of each channel of the preprocessed data. Remove abnormal data channels and count the number of valid channels. If the number of valid channels If no obvious construction event is found, the identification process ends; otherwise, proceed to step 4. Step 4: Extract the pulse density feature parameters of each effective channel and calculate the average pulse density K of all effective channels; Step 5: Determine whether the average pulse density K is lower than the threshold. If so, it is determined to be a hammer event, the identification is completed, and the identification process ends. Otherwise, proceed to step 6; Step 6: Extract the characteristic frequency band component ratio, duty cycle, and second harmonic content of each effective channel, and calculate the average characteristic frequency band component ratio (UHF), average duty cycle (D), and average second harmonic content (HI) of all effective channels respectively. Step 7: Based on the average pulse density K, the average proportion of characteristic frequency band components UHF, the average duty cycle D, and the average second harmonic content HI, jointly determine the event type. If the conditions are met, it is determined to be a cutting machine event, the identification is completed, and the identification process ends; otherwise, proceed to step 8. Step 8: Extract the tail coefficient feature parameters of each effective channel and calculate the average tail coefficient of all effective channels. ; Step 9: Determine the event type by jointly using the average pulse density K and the average tail coefficient TF. Compare the two with preset thresholds. If the average tail coefficient TF is greater than the preset threshold and the average pulse density K is less than the preset threshold, then it is determined to be a hammer event, the identification is completed, and the identification process ends. Otherwise, it is determined to be a pneumatic pick incident, the identification is completed, and the identification process ends.

2. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 1, is characterized in that... In step 1, the parameters are initialized using the following method to obtain the DAS signal sampling data sequence to be processed. Specifically, it includes the following steps: Step 1.1: Initialize the parameters for construction event identification, specifically including the initialization of the following parameters: The number of time sampling points L is initialized to Even numbers within the range; The number of sensor channels N is initialized to Integers within the range; upper limit frequency of bandpass filter Initialize to Real numbers in the Hz range; lower cutoff frequency of bandpass filter Initialize to Real numbers in the Hz range; The displacement corresponding to a cumulative probability of 0.5 under a standard Gaussian distribution. Initialize to ; Lower limit for determining the number of effective channels Initialize to Integers within the range; Initial screening signal-to-noise ratio threshold for active channels Initialize to Real numbers within the range; Minimum signal-to-noise ratio threshold for active channels Initialize to Real numbers within the range; Sampling frequency Initialize to Real numbers in the kHz range; Initial value of effective pulse history index Initialize to ; Minimum time window between adjacent pulses initialization Real numbers within the range s; Minimum relative amplitude threshold of pulse spike Initialize to Real numbers within the range; Upper limit of extremely slow pulse density judgment value Initialize to Real numbers in the Hz range; Lower limit of effective excitation continuity amplitude Initialize to Real numbers within the range; Lower limit of characteristic frequency band Initialize to Integers within the Hz range; Lower limit of effective frequency band for harmonic detection Initialize to Integers within the Hz range; Upper limit of effective frequency band for harmonic detection Initialize to Integers within the Hz range; Spectral peak relative prominence Initialize to Real numbers within the range; Harmonic Coupling Matching Frequency Tolerance Initialize to Integers within the Hz range; Baseband determination upper limit Initialize to Real numbers in the Hz range; Lower limit for judging ultrafast pulse density Initialize to Real numbers in the Hz range; Lower limit for continuous duty cycle determination Initialize to Real numbers within the range; Lower limit for determining the proportion of characteristic frequency bands Initialize to Real numbers within the range; Lower limit for harmonic matching index determination Initialize to Real numbers within the range; Envelope start and end relative amplitude threshold Initialize to Real numbers within the range; Lower limit for waveform tailing coefficient determination Initialize to Real numbers within the range; Upper limit of pulse density judgment in the fuzzy region Initialize to Real numbers in the Hz range; Step 1.2: Obtain the DAS signal sampling data sequence to be processed. The real-time data acquired simultaneously from L sampling points by N sensors in an array is used as the data sequence to be processed. Alternatively, DAS signal data containing L sampling points collected by N sensors can be extracted from the memory as a data sequence to be processed.

3. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 2, is characterized in that... In step 2, the DAS signal data sequence is sampled using the following method. Preprocessing includes the following steps: Step 2.1, Design frequency range is A bandpass filter for sampling DAS signal data sequences Filtering is performed to obtain filtered data. ; Step 2.2: For each channel, based on the filtered data... Calculate its noise floor standard deviation : ； in, Indicates taking the median of This indicates the absolute value operation, where the coefficient is... This is the displacement constant initialized in step 1.

1.

4. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 3, is characterized in that... In step 3, the signal-to-noise ratio of each channel of the preprocessed data is calculated using the following method. Remove abnormal data channels and count the number of valid channels. If the number of valid channels If no obvious construction event is found, the identification process ends; otherwise, proceed to step 4, which includes the following steps: Step 3.1, Calculate the channel Preprocessed data signal-to-noise ratio : ； in, This is an operation to find the maximum value of a set; Step 3.2, for For each channel, calculate the channel validity indicator variable. : ； Step 3.3: Based on the aforementioned channel validity determination indicator variable Determine whether to retain the current channel data. If so, then retain the data from that channel. Otherwise, discard the data in that channel, resulting in a valid number of channels. The remaining valid channels will be renumbered sequentially according to their original order as follows: ; Step 3.4: Determine the number of valid channels. The value: If If no obvious construction event is found in the current DAS signal sampling data sequence to be processed, the identification process ends; If so, proceed to step 3.5; Step 3.5: For each valid channel after renumbering The channel data is normalized by using its corresponding noise floor standard deviation to obtain normalized data. : 。 5. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 4, is characterized in that... In step 4, based on the effective channels obtained in step 3, the pulse density feature parameters of each effective channel are extracted, and the average pulse density of all effective channels is calculated. : ； in, The sampling frequency initialized in step 1, The number of time sampling points, The number of effective channels, The total number of valid pulses that satisfy the constraints is calculated as follows: ； in, The valid pulse determination indicator function is defined as follows: ； in, This refers to the minimum time window for adjacent pulses initialized in step 1. For the current sampling point The previous valid pulse time index is defined by the current time. Previous valid pulse history index set Perform the calculation: ； ； in, The initial value of the valid pulse history index initialized in step 1; Let be the local extremum determination function, defined as: ； in, The minimum relative amplitude threshold of the pulse spike initialized in step 1, Calculations are performed using absolute values.

6. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 5, is characterized in that... In step 5, based on the average pulse density of all effective channels... To determine the event type, specifically: Define the initial indicator variable for the hammer event. : ； in, The upper limit value for the extremely slow pulse density initialized in step 1 is determined; if If the event is identified as a hammer event, the identification process is completed and ends; otherwise, proceed to step 6.

7. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 6, is characterized in that... Step 6 specifically includes the following steps: Step 6.1: Calculate the normalized data for all valid channels. average duty cycle : ； in, The number of effective channels, The indicator function for determining continuous force application is defined as follows: ； in, The lower limit of the effective excitation continuity amplitude initialized in step 1; Step 6.2: For each valid channel Calculate the single-sided magnitude spectrum after its discrete Fourier transform. and total energy across the entire frequency band : Calculate the discrete Fourier transform sequence : ； in, For discrete frequency point indexing, The imaginary unit, ; Calculate the bilateral magnitude spectrum corresponding to the discrete Fourier transform sequence. : ； Extracting the positive frequency portion of the bilateral amplitude spectrum to obtain the single-sided amplitude spectrum. And force its DC component to zero: ； Calculate the single-sided amplitude spectrum Corresponding physical frequency axis sequence : ； Calculate the total energy of the single-sided amplitude spectrum within the effective frequency band. : ； Step 6.3: Calculate the average proportion of characteristic frequency bands for all effective channels. : ； in, Number of valid channels; The proportion of characteristic frequency bands in each effective channel is calculated as follows: ； in, The characteristic frequency band indication function is defined as follows: ； in, The lower limit value of the characteristic band frequency initialized in step 1; Step 6.4: Calculate the average second harmonic content of all effective channels. : For each valid channel Calculate the local total energy within the effective frequency band of harmonic detection. : ； in, The set of frequency indexes within the effective frequency band for harmonic detection is defined as: ； in, and These are the lower and upper limits of the effective frequency band for harmonic detection initialized in step 1, respectively. For sets Calculate the peak prominence of any local maximum frequency point within the range. : ； in, These are the amplitudes of the local minimum points adjacent to the local maximum frequency point on the left and right sides, respectively. In the set Internal search for valid spectral peaks, define a valid spectral peak index set. for: ； in, The relative prominence of the spectral peaks initialized in step 1; like If the frequency is not empty, extract the frequency corresponding to the effective spectral peak with the largest amplitude as the suspected fundamental frequency. Its corresponding frequency index is denoted as And define candidate indicator variables for second harmonics. : ； in, The harmonic coupling matching frequency tolerance initialized in step 1; It is an existential quantifier; Calculate the second harmonic content of each effective channel. : ； in, This is the upper limit value for determining the base frequency initialized in step 1; Number of valid channels; and They represent sets respectively The amplitudes of the first and second largest effective spectral peaks in the intermediate amplitude range, if there is only one element in this set, then ; It is an empty set; The average value of the second harmonic content The calculation is as follows: 。 8. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 7, is characterized in that... In step 7, based on the average pulse density... Average percentage of characteristic frequency band components Average duty cycle Average value of second harmonic content The joint determination of the event type includes the following steps: Define the initial indicator variable for the cutting machine event. : ； in, The lower limit value for determining the extremely fast pulse density initialized in step 1. The lower limit value for determining the continuous duty cycle initialized in step 1. The lower limit value for determining the proportion of characteristic frequency bands initialized in step 1. The lower limit value is determined for the harmonic matching index initialized in step 1; if If the event is identified as a cutting machine event, the identification process is completed and ends; otherwise, proceed to step 8.

9. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 8, is characterized in that... In step 8, the average tailing coefficient of all valid channels is calculated. Specifically, it includes the following steps: Step 8.1: For each valid channel If the channel satisfies the constraint conditions, the total number of valid pulses Then calculate the global maximum amplitude of the channel. and its corresponding time index : ； ； in, To return the index function of the argument that maximizes the objective function; like There is no need to calculate the global maximum amplitude and time index of the channel; simply execute step 8.3 for the channel. Step 8.2: Define the candidate index set for truncating the edge to the left of the point with the largest amplitude value. Truncating the candidate index set with the right edge : ； ； Calculate the starting index of the edge truncation respectively Termination index truncated with the right edge : ； ； in, The relative amplitude thresholds for the start and end of the envelope initialized in step 1; This is an operation to find the minimum value of a set; Step 8.3: Define the waveform tailing coefficient for each valid channel. : ； Step 8.4: Average the waveform tailing coefficient. The calculation is as follows: 。 10. The method for identifying construction events based on inherent physical characteristics using distributed acoustic sensing, as described in claim 9, is characterized in that... In step 9, based on the average pulse density... and the average tailing coefficient The joint determination of event types includes the following steps: defining indicator variables for the hammer and jackhammer events. : ； in, The lower limit value is determined for the waveform tailing coefficient initialized in step 1. The upper limit value for determining the pulse density of the ambiguous region initialized in step 1; if If the event is identified as a hammer event, the identification process is completed and ends; otherwise, it is identified as a jackhammer event, the identification process is completed and ends.

Images