Effective separation signal separation method for tunneling vibration noise

Abstract

The invention discloses an effective separation signal separation method for tunneling vibration noise. Tunneling vibration noise is generated due to the fact that mechanical vibration in underground engineering operation is transmitted to the surrounding environment. The intensity and the transmission range of the tunneling vibration noise are influenced by various factors including blasting or excavation energy, geological conditions, vibration transmission paths, underground water levels and the like. Different noise characteristics can be caused by different underground engineering operation conditions. In order to evaluate the influence of tunneling vibration noise, noise measurement and analysis are generally carried out. Therefore, in order to solve the problems, the rock breaking noise source structure of the heading machine is analyzed through a method of combining theoretical analysis, numerical simulation and field test, the signal-to-noise ratio and resolution of effective signals are improved based on advanced mathematical methods such as coherent analysis, independent component separation and CEEMD decomposition, and the noise source structure of the heading machine is obtained. The invention provides a tunneling vibration noise effective signal separation method, and aims to solve the problems in the prior art.

Description

Technical Field

[0001] The invention belongs to the technical field of noise, and in particular relates to a method for effectively separating signals of tunneling vibration noise. Background Art

[0002] The generation of tunneling vibration noise is caused by the transmission of mechanical vibrations in underground engineering operations to the surrounding environment. For example, in blasting operations, the vibrations generated by the explosion shock wave propagating into the underground rock will be transmitted to the ground and surrounding buildings, causing noise. The intensity and propagation range of tunneling vibration noise are affected by many factors, including the energy of blasting or excavation, geological conditions, vibration transmission path, groundwater level, etc. Different underground engineering operation conditions will result in different noise characteristics. In order to evaluate the impact of tunneling vibration noise, noise measurement and analysis are usually carried out. This can be achieved by using vibration sensors and noise instruments to monitor and record vibration and noise levels, and evaluate them according to relevant standards and specifications. Tunneling vibration noise is a common problem in underground engineering, and its causes and propagation characteristics are relatively complex. In order to effectively control tunneling vibration noise, it is necessary to separate the effective signals.

[0003] However, the existing signal separation methods have some problems in processing tunneling vibration noise, such as unsatisfactory separation effect, high computational complexity, and poor real-time performance.

[0004] Therefore, the present invention proposes a method for separating effective signals of tunneling vibration noise, aiming to solve the problems in the prior art. Summary of the invention

[0005] The purpose of the present invention is to provide a method for effectively separating signals of tunneling vibration noise in order to solve the above-mentioned problems.

[0006] The technical solution adopted by the present invention is as follows: a method for effectively separating signals of tunneling vibration noise, comprising a loader, a display screen, a conveyor, a hydraulic system, a tunneling cutter head, a cooling system, a tunneling crawler, an operating room, a rear support rod and a connecting arm, characterized in that the separation method comprises the following steps: S1: When the tunneling cutter head is in use, it first collects the vibration signal generated when the tunneling machine is working; S2: The tunneling cutter head transmits the signal through the signal transmission module on the conveyor, and then performs spectrum analysis on the vibration signal to determine the frequency characteristics of the effective signal; the collected vibration signal is analyzed in the time domain to extract the time domain characteristics related to the working state of the tunneling machine, which may include time series data and waveform changes; A sensor is set on the tunneling cutter head to obtain the surface wave data of the tunneling noise source corresponding to the measuring point in the detection area, and after removing the interference signal, a pure surface wave signal of the tunneling cutter head vibration is obtained; Construct a dispersion curve mode extraction network for the vibration of the tunneling cutter head, and perform phased training using surface wave simulation data with different arrangement lengths; The trained dispersion curve mode extraction network of the tunneling cutter head vibration is used to extract the multi-modal surface wave dispersion curve from the pure surface wave signal of the tunneling cutter head vibration; wherein, the pure surface wave signal of the tunneling cutter head vibration is subjected to mode separation, and the domain characteristic graph of each mode after separation is subjected to spectrum conversion to obtain the dispersion curve of each mode; A data quality optimization network for tunneling cutter head vibration is constructed, and after being trained with an interference suppression data set, interference signals are removed from the noise source surface wave data of tunneling cutter head vibration; the interference suppression tunneling cutter head vibration data set includes surface wave virtual source shot data synthesized by pure surface wave signals and surface wave interference signals of tunneling cutter head vibration, and pure surface wave signals of tunneling cutter head vibration as label data; The tunneling cutter head vibration data quality optimization network translates along the spatial axis and time axis of the tunneling noise source surface wave data in the form of a sliding window, and processes the tunneling noise source surface wave data at different time and space positions respectively; S3: A signal processing device is provided inside the operating room, and a filter is designed to filter the vibration signal according to the frequency characteristics of the effective tunneling cutter head signal; S4: A signal processing device is provided inside the operating room to perform frequency domain analysis on the filtered signal to determine the frequency domain characteristics of the effective signal; the operating room converts the filtered vibration signal into the frequency domain for analysis to extract the frequency domain characteristics related to the working state of the tunneling cutter head of the tunnel boring machine; S5: The operating room further filters the vibration signal using an adaptive filter according to the frequency domain characteristics; according to the characteristics extracted in the frequency domain, the vibration signal of the tunneling cutter head is filtered again using an adaptive filter to further remove the remaining noise and interference; S6: The operating room outputs the separated effective signal. After being processed in the above steps, it outputs the effective signal of the tunneling cutter head of the tunneling machine when it is working; then the entire tunneling vibration noise effective separation signal separation process can be ended.

[0007] In a preferred embodiment, in step S1, the vibration signal of the excavator during operation is collected by using an acceleration sensor. Step S1 collects the vibration signal of the tunnel boring machine using the acceleration sensor to provide raw data for subsequent signal processing.

[0008] In a preferred embodiment, in the step S2, during the spectrum analysis of the vibration signal, after processing, the continuous frames are scanned by interlaced decomposition. The formula for this process is:

[0009] In the formula, S represents the clutter processing process of the vibration signal, w represents continuous frame scanning, μ represents the interlaced decomposition of the vibration signal, and λ represents the transmission subband of the vibration signal; After completing the above processing, discrete wavelet embedded technology is used to perform lossless compression and encryption of the vibration signal; the compression processing process is shown in the following formula:

[0010] In the formula, c represents the lossless compression encryption of the vibration signal, a represents the encoder, D represents the vibration signal image encoding, i represents the vibration signal decoder, j represents the number of forward wavelet transforms of the original vibration signal image, b represents the data inverse quantization processing process, and A represents the low-frequency wavelet encapsulation encryption; according to the above method, the vibration signal compression and encryption processing is completed.

[0011] In a preferred embodiment, in step S3, the vibration signal is filtered using an adaptive filter based on the previously extracted time domain features to further remove noise and interference. This step utilizes the adaptive performance of the adaptive filter to dynamically adjust the filtering parameters according to signal changes.

[0012] In a preferred embodiment, in step S3, first, the effective signal after A / D conversion is a discrete signal, and the redundant data elimination module uses a cross-correlation function to describe the similarity between random digital effective signals. Assuming that there are two digital effective signals x[i] and y[i] to be compared, the expression of the cross-correlation function is as follows:

[0013] Where: t represents time; T represents the digital effective signal transmission period; rxy[t] is the similarity value of the effective signal x[i] and y[i] at time t. The larger the similarity value, the more similar the two signals are. Then, the above cross-correlation function is normalized and the correlation coefficient is used to express the similarity. The calculation method of the correlation coefficient is as follows:

[0014] Wherein: ρxy[t] represents the correlation coefficient, and its value range is [-1, 1]; when the effective signals x[i] and y[i] are exactly the same, the maximum value 1 is taken, otherwise, the minimum value -1 is taken. According to the above process, effective signal transmission is realized to ensure the accuracy and authenticity of the effective signal transmission process.

[0015] In a preferred embodiment, in step S4, by integrating the states in the product of the likelihood function of the effective signal of the working state of the tunnel boring machine and the one-step state prediction probability, the probability distribution of the observation transfer of the effective signal of the working state of the tunnel boring machine can be obtained, thereby obtaining the mean and variance of the target observation of the effective signal of the working state of the tunnel boring machine, and calculating the Kalman gain:

[0016]

[0017] The error matrix is ​​updated as:

[0018] in ， is the Kalman gain at time K, H is the unit matrix, R is the output result of the bp neural network, and here it is used as the measurement noise covariance matrix to input the Kalman filter, and the adaptive filtering effect is achieved according to the output result; since the noise satisfies the Gaussian distribution, Q and R can be regarded as independent of each other, then Q and R can be set as diagonal matrices, and the value of Q is the diagonal matrix of R.

[0019] In a preferred embodiment, in step S5, the maximum and minimum values ​​of the signal are combined into an envelope by cubic spline interpolation, a new mean signal is obtained according to the envelope, and the original signal is subtracted from the mean signal to obtain a new signal; it is determined whether the new signal meets the conditions of the above-mentioned IMF component, if so, the new signal is the first IMF component, if not, the new signal is used as the original signal to repeat the above operation until the conditions are met; after the conditions are met, the original signal minus the IMF component is subtracted as the residual signal, the IMF component is continued to be extracted from the residual signal, and after a preset number of iterations and a preset number of IMF layers, the original signal is decomposed into multiple IMF components and a residual residual part.

[0020] In a preferred embodiment, in step S5, a semi-soft threshold method is used to denoise the noisy part after CEEMD decomposition; the formula is:

[0021] Where, W' j,k is the processed coefficient, w is the decomposed wavelet coefficient, and λ is the determined threshold coefficient; a fixed noise is added by generating a simulated vibration signal, which includes the high-frequency noise, acoustic signal noise and environmental noise existing in the hardware system; then the same signal is processed using different wavelet bases and different wavelet base coefficients, and the most suitable wavelet basis function and wavelet basis coefficient are selected based on the signal-to-noise ratio value between the filtered signal and the original signal.

[0022] In a preferred embodiment, in step S5, an adaptive filter is used to filter the vibration signal according to the time domain and frequency domain characteristics to achieve separation of the effective signal of excavation vibration; after time domain and frequency domain analysis, the vibration signal is processed using an adaptive filter; the adaptive filter can adaptively adjust the filtering parameters according to the changes in the signal.

[0023] In a preferred embodiment, in step S4, the filtered signal is subjected to frequency domain analysis to determine the frequency domain characteristics of the effective signal, including frequency and amplitude; when the filtered vibration signal is subjected to frequency domain analysis, frequency domain characteristics related to the working state of the tunnel boring machine are extracted; these characteristics may include the frequency and amplitude of the effective signal; by analyzing these characteristics, the working state and performance status of the tunnel boring machine can be further understood.

[0024] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. In the present invention, the rock-breaking noise source structure of the tunnel boring machine is analyzed by combining theoretical analysis, numerical simulation and field test, and the effective signal-to-noise ratio and resolution are improved based on advanced mathematical methods such as coherence analysis, independent component separation, and CEEMD decomposition. The propagation law of the seismic wave field of the rock-breaking noise source is studied, and a database of seismic wave field characteristics of the noise source of unfavorable geological bodies is established. A new technology for advanced detection and imaging of seismic waves of rock-breaking noise sources based on machine learning is realized, and on this basis, data processing and interpretation software with independent intellectual property rights is developed. The expected results can effectively improve the efficiency, accuracy, timeliness and intelligence level of advanced detection work, which is of great significance for ensuring the safe and rapid excavation of coal mine tunnels.

[0025] 2. In the present invention, independent analysis technology is introduced, combined with CEEMD decomposition and wavelet transform, to establish an effective signal recognition method, so as to realize the extraction of effective signals of rock-breaking noise of tunnel boring machines in tunnel environments. Based on numerical simulation technology, a geological geophysical model and a sample library of characteristic parameters of seismic wave fields of noise sources are established, and a machine learning method is used to realize the imaging of undesirable geological bodies in seismic wave fields of complex noise sources. The distributed storage algorithm based on Hadoop / Spark architecture realizes the transformation from "separation of computing and storage" to "migration of computing to storage", which can effectively improve the efficiency of data storage, access and processing, and provide guarantee for advanced geological exploration at the data level.

[0026] 3. In the present invention, the advance earthquake prediction technology of tunneling noise source uses tunneling noise as the earthquake source for real-time collection, does not require shutdown for observation, and does not affect normal tunneling construction. The present invention can effectively improve the efficiency of mine tunneling work and reduce construction costs; the present invention can effectively improve the accuracy and timeliness of geological disaster prediction, reduce the occurrence of safety accidents caused by hidden geological factors, reduce disaster reduction costs, and generate significant social and economic value. It can effectively prevent and reduce the occurrence of geological disaster-related production safety accidents, improve the happiness index of staff, reduce possible social instability factors, and generate potential good social benefits. After the local pilot demonstration, the present invention has the potential for large-scale application and promotion, and can be promoted nationwide. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the structure of the tunneling equipment of the present invention; Figure 2 It is a schematic diagram of the structure of the tunneling equipment in the present invention from a rear side perspective; Figure 3 This is a system block diagram of the signal acquisition module in the present invention; Figure 4 This is a flow chart for identifying the excavation vibration noise source in the present invention; Figure 5 This is a flow chart of tunneling vibration noise signal processing in the present invention.

[0028] Markings in the figure: 1-loader, 2-display screen, 3-conveyor, 4-hydraulic system, 5-excavation cutter head, 6-cooling system, 7-excavation crawler, 8-operating room, 9-rear support rod, 10-connecting arm. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0030] Reference Figure 1-5 , A method for effectively separating signals of tunneling vibration noise includes a loader 1, a display screen 2, a conveyor 3, a hydraulic system 4, a tunneling cutter head 5, a cooling system 6, a tunneling crawler 7, an operating room 8, a rear support rod 9 and a connecting arm 10. The separation method includes the following steps: S1: When the tunneling cutter head 5 is in use, it first collects the vibration signal generated when the tunneling machine is working; S2: The tunneling cutter head 5 transmits the signal through the signal transmission module on the conveyor 3, and then performs spectrum analysis on the vibration signal to determine the frequency characteristics of the effective signal; analyzes the collected vibration signal in the time domain to extract the time domain characteristics related to the working state of the tunneling machine, which may include time series data and waveform changes; A sensor is arranged on the tunneling cutter head 5 to obtain the surface wave data of the tunneling noise source corresponding to the measuring point in the detection area, and after removing the interference signal, a pure surface wave signal of the vibration of the tunneling cutter head 5 is obtained; Construct a dispersion curve mode extraction network for the vibration of the tunneling cutter head 5, and perform phased training using surface wave simulation data with different arrangement lengths; The trained dispersion curve mode extraction network of the vibration of the tunneling cutter head 5 is used to extract the multi-modal surface wave dispersion curve from the pure surface wave signal of the vibration of the tunneling cutter head 5; wherein, the pure surface wave signal of the vibration of the tunneling cutter head 5 is subjected to mode separation, and the domain characteristic graph of each mode after separation is subjected to spectrum conversion to obtain the dispersion curve of each mode; A data quality optimization network for the vibration of the tunneling cutter head 5 is constructed, and after being trained with an interference suppression data set, interference signals are removed from the noise source surface wave data of the vibration of the tunneling cutter head 5; the interference suppression data set for the vibration of the tunneling cutter head 5 includes surface wave virtual source shot data synthesized by a pure surface wave signal and a surface wave interference signal of the vibration of the tunneling cutter head 5, and a pure surface wave signal of the vibration of the tunneling cutter head 5 as label data; The vibration data quality optimization network of the tunneling cutter head 5 translates along the spatial axis and time axis of the tunneling noise source surface wave data in the form of a sliding window, and processes the tunneling noise source surface wave data at different time and space positions respectively; S3: A signal processing device is provided inside the operating room 8, and a filter is designed to filter the vibration signal according to the frequency characteristics of the effective signal of the tunneling cutter head 5; S4: A signal processing device is provided inside the operating room 8 to perform frequency domain analysis on the filtered signal to determine the frequency domain characteristics of the effective signal; the operating room 8 converts the filtered vibration signal into the frequency domain for analysis to extract the frequency domain characteristics related to the working state of the tunneling cutter head 5 of the tunneling machine; S5: The operating room 8 further filters the vibration signal using an adaptive filter according to the frequency domain characteristics; according to the characteristics extracted in the frequency domain, the vibration signal of the tunneling cutter head 5 is filtered again using an adaptive filter to further remove the remaining noise and interference; S6: The operating room 8 outputs the effective signal after separation. After being processed in the above steps, it outputs the effective signal of the tunneling cutter head 5 of the tunneling machine when it is working; then the entire tunneling vibration noise effective separation signal separation process can be ended.

[0031] In step S1, the vibration signal of the excavator when working is collected by using an acceleration sensor. In step S1, the vibration signal of the excavator is collected by using an acceleration sensor to provide raw data for subsequent signal processing.

[0032] In step S2, during the spectrum analysis of the vibration signal, the continuous frames are scanned by interlaced decomposition after processing. The formula for this process is:

[0033] In the formula, S represents the clutter processing process of the vibration signal, w represents continuous frame scanning, μ represents the interlaced decomposition of the vibration signal, and λ represents the transmission subband of the vibration signal; After completing the above processing, discrete wavelet embedded technology is used to perform lossless compression and encryption of vibration signals. The compression process is shown in the following formula:

[0034] In the formula, c represents the lossless compression encryption of the vibration signal, a represents the encoder, D represents the vibration signal image encoding, i represents the vibration signal decoder, j represents the number of forward wavelet transforms of the original vibration signal image, b represents the data inverse quantization process, and A represents the low-frequency wavelet encapsulation encryption. According to the above method, the vibration signal compression and encryption process is completed.

[0035] In step S3, the vibration signal is filtered using an adaptive filter based on the previously extracted time domain features to further remove noise and interference. This step utilizes the adaptive performance of the adaptive filter to dynamically adjust the filter parameters according to signal changes to improve the filtering effect.

[0036] In step S3, first, the valid signal after A / D conversion is a discrete signal. The redundant data elimination module uses a cross-correlation function to describe the similarity between random digital valid signals. Assuming that there are two digital valid signals x[i] and y[i] to be compared, the expression of the cross-correlation function is as follows:

[0037] Where: t represents time; T represents the digital effective signal transmission period; rxy[t] is the similarity value of the effective signal x[i] and y[i] at time t. The larger the similarity value, the more similar the two signals are. Then, the above cross-correlation function is normalized and the correlation coefficient is used to express the similarity. The calculation method of the correlation coefficient is as follows:

[0038] Wherein: ρxy[t] represents the correlation coefficient, and its value range is [-1, 1]; when the effective signals x[i] and y[i] are exactly the same, the maximum value 1 is taken, otherwise, the minimum value -1 is taken. According to the above process, effective signal transmission is realized to ensure the accuracy and authenticity of the effective signal transmission process.

[0039] In step S4, by integrating the states in the product of the likelihood function of the effective signal of the working state of the tunnel boring machine and the one-step state prediction probability, the probability distribution of the observation transfer of the effective signal of the working state of the tunnel boring machine can be obtained, thereby obtaining the mean and variance of the target observation of the effective signal of the working state of the tunnel boring machine, and the Kalman gain can be calculated:

[0040]

[0041] The error matrix is ​​updated as:

[0042] in ， is the Kalman gain at time K, H is the unit matrix, and R is the output result of the BP neural network, which is used as the measurement noise covariance matrix to input the Kalman filter, and the adaptive filtering effect is achieved according to the output result. Since the noise satisfies the Gaussian distribution, Q and R can be regarded as independent of each other, so Q and R can be set as diagonal matrices, and the value of Q is the diagonal matrix of R.

[0043] In step S5, the maximum and minimum values ​​of the signal are combined into an envelope by cubic spline interpolation, a new mean signal is obtained according to the envelope, and the original signal is subtracted from the mean signal to obtain a new signal. It is determined whether the new signal meets the above-mentioned IMF component conditions. If so, the new signal is the first IMF component. If not, the new signal is used as the original signal to repeat the above operation until the conditions are met. After the conditions are met, the original signal minus the IMF component is used as the residual signal, and the IMF component is further extracted from the residual signal. After a preset number of iterations and a preset number of IMF layers, the original signal is decomposed into multiple IMF components and a residual residual part.

[0044] In step S5, a semi-soft threshold method is used to denoise the noisy part after CEEMD decomposition. The formula is: Where, W' j,k is the processed coefficient, w is the decomposed wavelet coefficient, and λ is the determined threshold coefficient; fixed noise is added by generating a simulated vibration signal, which includes high-frequency noise, acoustic signal noise and environmental noise in the hardware system. Then different wavelet bases and different wavelet base coefficients are used to process the same signal, and the signal-to-noise ratio between the filtered signal and the original signal is used to select the most suitable wavelet basis function and wavelet base coefficient.

[0045] In step S5, an adaptive filter is used to filter the vibration signal according to the time domain and frequency domain characteristics to achieve separation of effective signals.

[0046] After time domain and frequency domain analysis, the vibration signal is processed using an adaptive filter. The adaptive filter can adaptively adjust the filter parameters according to the changes in the signal to achieve the best filtering effect. Through the application of the adaptive filter, noise and interference can be better removed and the signal quality can be improved.

[0047] In step S4, the filtered signal is subjected to frequency domain analysis to determine the frequency domain characteristics of the effective signal, including frequency, amplitude, etc. When the filtered vibration signal is subjected to frequency domain analysis, frequency domain characteristics related to the working state of the tunnel boring machine are extracted. These characteristics may include the frequency and amplitude of the effective signal. By analyzing these characteristics, the working state and performance of the tunnel boring machine can be further understood.

[0048] In the present invention, the rock-breaking noise source structure of the tunnel boring machine is analyzed by combining theoretical analysis, numerical simulation and field test, and the effective signal-to-noise ratio and resolution are improved based on advanced mathematical methods such as coherence analysis, independent component separation, and CEEMD decomposition. The propagation law of the seismic wave field of the rock-breaking noise source is studied, and a database of seismic wave field characteristics of the noise source of unfavorable geological bodies is established. A new technology of advanced detection and imaging of seismic waves of rock-breaking noise sources based on machine learning is realized, and on this basis, data processing and interpretation software with independent intellectual property rights is developed. The expected results can effectively improve the efficiency, accuracy, timeliness and intelligence level of advanced detection work, which is of great significance for ensuring the safe and rapid excavation of coal mine tunnels.

[0049] In the present invention, an independent analysis technology is introduced, combined with CEEMD decomposition and wavelet transform, to establish an effective signal recognition method, and to realize the extraction of effective signals of rock-breaking noise of tunnel boring machines in tunnel environments. Based on numerical simulation technology, a geological geophysical model and a sample library of characteristic parameters of seismic wave fields of noise sources are established, and a machine learning method is used to realize the imaging of undesirable geological bodies in seismic wave fields of complex noise sources. The distributed storage algorithm based on Hadoop / Spark architecture realizes the transformation from "separation of computing and storage" to "migration of computing to storage", which can effectively improve the efficiency of data storage, access, and processing, and provide a guarantee for advanced geological exploration at the data level.

[0050] In the present invention, the time-frequency analysis of the measured data of the rock-breaking noise of the tunnel boring machine is analyzed, and the source characteristics of the rock-breaking noise of the tunnel boring machine are revealed by combining theoretical analysis and investigation. The independent component analysis technology is introduced, combined with conventional data extraction methods such as CEEMD decomposition and wavelet transform, to identify the effective signals of tunneling vibration noise in complex environments. A geological geophysical model for advanced detection of unfavorable geological bodies such as rock layer interfaces, broken zones, caves, and water-rich rock layers is established, and numerical simulation is carried out using the time-domain finite difference method. The relationship between the characteristic parameters of the seismic wave field of the vibration noise source and the geological geophysical model is explained. According to the simulation results, a characteristic parameter-unfavorable geological body model sample library is established to realize the unfavorable geological body imaging technology based on deep learning. A massive data management library is established using the Hadoop / Spark architecture, and supporting client software is developed based on the project research content to build a set of vibration noise source seismic wave advance detection data processing and interpretation platform.

[0051] In the present invention, the advance earthquake prediction technology of tunneling noise source uses tunneling noise as the earthquake source for real-time collection, does not require shutdown observation, and does not affect normal tunneling construction. The present invention can effectively improve the efficiency of mine tunneling work and reduce construction costs; the present invention can effectively improve the accuracy and timeliness of geological disaster prediction, reduce the occurrence of safety accidents caused by hidden geological factors, reduce disaster reduction costs, and generate significant social and economic value. It can effectively prevent and reduce the occurrence of geological disaster-related production safety accidents, improve the happiness index of staff, reduce possible social instability factors, and generate potential good social benefits. After the local pilot demonstration, the present invention has the potential for large-scale application and promotion, and can be promoted nationwide.

[0052] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0053] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for effectively separating signals of tunneling vibration noise, comprising a loader (1), a display screen (2), a conveyor (3), a hydraulic system (4), a tunneling cutter head (5), a cooling system (6), a tunneling crawler (7), an operating room (8), a rear support rod (9) and a connecting arm (10), characterized in that: The separation method comprises the following steps: S1: When the tunneling cutter head (5) is in use, it first collects the vibration signal generated when the tunneling machine is working; S2: The tunneling cutter head (5) transmits the signal through the signal transmission module on the conveyor (3), and then performs spectrum analysis on the vibration signal to determine the frequency characteristics of the effective signal; the collected vibration signal is analyzed in the time domain to extract the time domain characteristics related to the working state of the tunneling machine, and these characteristics may include time series data and waveform changes; A sensor is arranged on the tunneling cutter head (5) to obtain surface wave data of the tunneling noise source corresponding to the measuring point in the detection area, and after removing the interference signal, a pure surface wave signal of the vibration of the tunneling cutter head (5) is obtained; Constructing a dispersion curve mode extraction network for the vibration of the tunneling cutter head (5), and performing phased training using surface wave simulation data of different arrangement lengths; A multi-modal surface wave dispersion curve is extracted from a pure surface wave signal of the vibration of the tunneling cutter head (5) using a trained dispersion curve mode extraction network; wherein the pure surface wave signal of the vibration of the tunneling cutter head (5) is subjected to mode separation, and a domain feature map of each mode after separation is subjected to spectrum conversion to obtain a dispersion curve of each mode; A data quality optimization network for the vibration of the tunneling cutter head (5) is constructed, and after being trained with an interference suppression data set, interference signals are removed from the noise source surface wave data of the vibration of the tunneling cutter head (5); the interference suppression data set for the vibration of the tunneling cutter head (5) includes surface wave virtual source shot data synthesized by a pure surface wave signal and a surface wave interference signal of the vibration of the tunneling cutter head (5), and a pure surface wave signal of the vibration of the tunneling cutter head (5) as label data; The tunneling cutter head (5) vibration data quality optimization network translates along the spatial axis and time axis of the tunneling noise source surface wave data in the form of a sliding window, and processes the tunneling noise source surface wave data at different time and space positions respectively; S3: A signal processing device is provided inside the operating room (8), and a filter is designed to filter the vibration signal according to the frequency characteristics of the effective tunneling cutter head (5) signal; S4: A signal processing device is provided inside the operating room (8) to perform frequency domain analysis on the filtered signal to determine the frequency domain characteristics of the effective signal; the operating room (8) converts the filtered vibration signal into the frequency domain for analysis to extract the frequency domain characteristics related to the working state of the tunnel boring machine cutting head (5); S5: The operating room (8) further filters the vibration signal using an adaptive filter according to the frequency domain characteristics; based on the characteristics extracted in the frequency domain, the vibration signal of the tunneling cutter head (5) is filtered again using an adaptive filter to further remove the remaining noise and interference; S6: The operating room (8) outputs the separated effective signal, and after being processed in the above steps, outputs the effective signal of the tunneling cutter head (5) of the tunneling machine when it is working; then the entire tunneling vibration noise effective separation signal separation process is terminated.

2. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In the step S1, the vibration signal of the excavator when working is collected by using the acceleration sensor of the tunneling cutter head (5). Step S1 collects the vibration signal of the tunneling machine by using the acceleration sensor to provide raw data for subsequent signal processing.

3. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In the step S2, during the spectrum analysis of the vibration signal of the tunneling cutter head (5), after processing, continuous frames are scanned in an interlaced decomposition manner. The formula for this process is: In the formula, S represents the clutter processing process of the vibration signal, w represents continuous frame scanning, μ represents the interlaced decomposition of the vibration signal, and λ represents the transmission subband of the vibration signal; After completing the above processing, discrete wavelet embedded technology is used to perform lossless compression and encryption of the vibration signal; the compression processing process is shown in the following formula: In the formula, c represents the lossless compression encryption of the vibration signal, a represents the encoder, D represents the vibration signal image encoding, i represents the vibration signal decoder, j represents the number of forward wavelet transforms of the original vibration signal image, b represents the data inverse quantization processing process, and A represents the low-frequency wavelet encapsulation encryption; according to the above method, the vibration signal compression and encryption processing is completed.

4. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In step S3, the vibration signal is filtered using an adaptive filter based on the previously extracted time domain features to further remove noise and interference. This step utilizes the adaptive performance of the adaptive filter to dynamically adjust the filtering parameters according to signal changes.

5. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In the step S3, first, the effective tunneling cutter head (5) signal after A / D analog-to-digital conversion is a discrete signal, and the redundant data elimination module uses a cross-correlation function to describe the similarity between the effective signals of the random digital tunneling cutter head (5). Assuming that there are two effective digital signals x[i] and y[i] to be compared, the expression of the cross-correlation function is as follows: Where: t represents time; T represents the digital effective signal transmission period; rxy[t] is the similarity value of the effective signal x[i] and y[i] at time t. The larger the similarity value, the more similar the two signals are. Then, the above cross-correlation function is normalized and the correlation coefficient is used to express the similarity. The calculation method of the correlation coefficient is as follows: Wherein: ρxy[t] represents the correlation coefficient, and its value range is [-1, 1]; when the effective signals x[i] and y[i] are exactly the same, the maximum value 1 is taken, otherwise, the minimum value -1 is taken. According to the above process, effective signal transmission is realized to ensure the accuracy and authenticity of the effective signal transmission process.

6. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In step S4, by integrating the state in the product of the likelihood function of the effective signal of the working state of the tunnel boring machine cutter head (5) and the one-step state prediction probability, the probability distribution of the observation transfer of the effective signal of the working state of the tunnel boring machine can be obtained, thereby obtaining the mean and variance of the target observation of the effective signal of the working state of the tunnel boring machine, and calculating the Kalman gain: The error matrix is ​​updated as: in ， is the Kalman gain at time K, H is the unit matrix, R is the output result of the bp neural network, and here it is used as the measurement noise covariance matrix to input the Kalman filter, and the adaptive filtering effect is achieved according to the output result; since the noise satisfies the Gaussian distribution, Q and R can be regarded as independent of each other, then Q and R can be set as diagonal matrices, and the value of Q is the diagonal matrix of R.

7. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In the step S5, the maximum and minimum values ​​of the signal are combined into an envelope by the cubic spline interpolation method, a new mean signal is obtained according to the envelope, and the original signal is subtracted from the mean signal to obtain a new signal; it is determined whether the new signal meets the above-mentioned IMF component conditions, if so, the new signal is the first IMF component, if not, the new signal is used as the original signal to repeat the above operation until the conditions are met; after the conditions are met, the original signal minus the IMF component is used as the residual signal, the IMF component is continuously extracted from the residual signal, and after a preset number of iterations and a preset number of IMF layers, the original signal is decomposed into multiple IMF components and a residual residual part.

8. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In step S5, a semi-soft threshold method is used to denoise the noisy part after CEEMD decomposition; the formula is: Where, W' j,k is the processed coefficient, w is the decomposed wavelet coefficient, and λ is the determined threshold coefficient; a fixed noise is added by generating a simulated vibration signal, which includes the high-frequency noise, acoustic signal noise and environmental noise existing in the hardware system; then the same signal is processed using different wavelet bases and different wavelet base coefficients, and the most suitable wavelet basis function and wavelet basis coefficient are selected based on the signal-to-noise ratio value between the filtered signal and the original signal.

9. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In the step S5, the vibration signal is filtered and processed using an adaptive filter according to the time domain and frequency domain characteristics to achieve separation of effective excavation vibration signals; after time domain and frequency domain analysis, the vibration signal is processed using an adaptive filter; Adaptive filters can adaptively adjust filtering parameters according to signal changes.

10. The method for effectively separating tunneling vibration noise signals according to claim 1, characterized in that: In step S4, the filtered signal is subjected to frequency domain analysis to determine the frequency domain characteristics of the effective signal, including frequency and amplitude; when the filtered vibration signal is subjected to frequency domain analysis, frequency domain characteristics related to the working state of the tunnel boring machine are extracted; these characteristics may include the frequency and amplitude of the effective signal; by analyzing these characteristics, the working state and performance status of the tunnel boring machine can be further understood.

Images