Microphone array, monitoring system and method for monitoring linear sound source

Abstract

The invention belongs to the technical field of acoustic monitoring systems, and provides a microphone array for monitoring a linear sound source, a monitoring system and a method, the microphone array comprises a plurality of microphones distributed along a part of a cylindrical surface, and the plurality of microphones are distributed along the axial direction of the cylindrical surface in a variable density manner; the partial cylindrical surface microphone array is adopted, so that the space covering capability of a linear sound source is remarkably improved; the distribution of the microphones on the cylindrical surface is determined according to nonlinear factors of array height and microphone number height distribution, optimal arrangement is realized in three dimensions of radial direction, angle and height, the spatial sampling efficiency of the array is remarkably improved, sampling points are effectively increased under the condition of the same number of microphones, and the sampling efficiency is improved. In combination with integrated optimization of a dynamic reconstruction adaptive array algorithm and array arrangement, the flexibility of adapting to different monitoring scenes is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of acoustic monitoring systems, and in particular relates to a microphone array, a monitoring system and a method for monitoring a linear sound source. Background Art

[0002] Overhead power lines can generate low-frequency noise in certain circumstances, which may indicate potential equipment failure or abnormal operating conditions. Accurately locating and analyzing these noises is critical for preventive maintenance of power systems.

[0003] In the prior art, the commonly used uniformly distributed microphone array has limitations when processing linear sound sources, especially in terms of low-frequency performance and spatial resolution. This is mainly manifested in that when laying out the microphones in the array, factors such as the angle of the microphone layout, the angle variation between microphones, and the array height are not considered; in addition, the traditional array design ignores the special geometric characteristics and near-field effects of linear sound sources, making it difficult to achieve wide-band monitoring and precise positioning at the same time. Summary of the invention

[0004] In order to solve the above problems, the present invention proposes a microphone array, a monitoring system and a method for monitoring linear sound sources. A partial cylindrical microphone array is adopted to significantly improve the spatial coverage capability of linear sound sources. A variable density microphone distribution strategy is adopted to achieve optimized layout in three dimensions of radial, angle and height, which significantly improves the spatial sampling efficiency of the array and effectively increases the sampling points with the same number of microphones. Combined with the integrated optimization of the dynamic reconstruction adaptive array algorithm and the array layout, the flexibility of adapting to different monitoring scenarios is significantly improved.

[0005] In order to achieve the above objectives, in a first aspect, the present invention provides a microphone array for monitoring a linear sound source, which adopts the following technical solution:

[0006] A microphone array for monitoring a linear sound source, comprising a plurality of microphones distributed along a portion of a cylindrical surface, wherein the plurality of microphones are distributed with variable density along the axial direction of the cylinder;

[0007] When the microphones are distributed on the surface of a cylinder, the distance from the microphone to the axis of the cylinder is equal to the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, where the calculated angle is equal to the sine value of the ratio of the current microphone sequence to the number of microphones; the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, is determined by the power calculation value of the ratio of the current microphone sequence to the number of microphones multiplied by the array height, where the exponent of the power calculation is a nonlinear factor of the height distribution.

[0008] Furthermore, the distribution of the microphones in space is:

[0009] r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α

[0010] Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; z i is the position of the i-th microphone along the length direction of the axis; H is the array height; α is the nonlinear factor of the height distribution.

[0011] In order to achieve the above object, in a second aspect, the present invention further provides a monitoring system for monitoring a linear sound source, which adopts the following technical solution:

[0012] A monitoring system for monitoring linear sound sources, comprising an interconnected microphone array, a signal acquisition and preprocessing module, an adaptive beamforming processor, a multi-scale signal analysis module, and a sound source localization and feature extraction module;

[0013] The microphone array includes a plurality of microphones distributed along a portion of a cylindrical surface, and the plurality of microphones are distributed with variable density along the axial direction of the cylinder;

[0014] When the microphones are distributed on the surface of a cylinder, the distance from the microphone to the axis of the cylinder is equal to the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, where the calculated angle is equal to the sine value of the ratio of the current microphone sequence to the number of microphones; the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, is determined by the power calculation value of the ratio of the current microphone sequence to the number of microphones multiplied by the array height, where the exponent of the power calculation is a nonlinear factor of the height distribution.

[0015] Furthermore, the distribution of the microphones in space is:

[0016] r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α

[0017] Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; zi is the position of the i-th microphone along the length direction of the axis; H is the array height; α is the nonlinear factor of the height distribution.

[0018] Furthermore, the adaptive beamforming processor is used to adaptively suppress interference by utilizing minimum variance distortion-free response beamforming; and the objective function aims to minimize the total power output by the beamformer.

[0019] Furthermore, in W H Under the condition of a(θ0)=1: min W W H Rw;

[0020] Where: W represents an N×1 complex column vector, representing the beamforming weight, N is the number of microphones in the microphone array, and each element ω i Corresponding to the complex weight of a microphone; W H represents the conjugate transpose of W; R represents an N×N spatial covariance matrix, whose elements R ij represents the covariance between the signals of the i-th microphone and the j-th microphone; a(θ0) represents an N×1 complex vector, which describes the phase relationship of the signal from the desired direction θ0 on each microphone, and θ0 represents the direction of the linear sound source.

[0021] Furthermore, the multi-scale signal analysis module uses continuous wavelet transform to perform time-frequency analysis on the collected acoustic signal, and the wavelet transform is:

[0022]

[0023] Among them, X(a, b) is the result of wavelet transform, which is a two-dimensional function. a is the scale parameter that controls the expansion and contraction of the wavelet; b is the translation parameter; x(t) is the acoustic signal collected by the microphone array; ψ a，b (t) is the wavelet basis function, It is a transformed version of the mother wavelet function in time and scale; is the complex conjugate of the wavelet basis function.

[0024] Furthermore, by changing the scale parameter, the wavelet transform analyzes the behavior of the signal at different frequencies. Larger scale parameter values ​​correspond to low-frequency analysis, and smaller scale parameter values ​​correspond to high-frequency analysis. The translation parameter allows the analysis of the characteristics of the signal at different time points.

[0025] Furthermore, the signal acquisition and preprocessing module uses an adaptive filter to remove environmental background noise, applies spectral subtraction technology to improve the signal-to-noise ratio, and performs frequency response correction according to the microphone response characteristics.

[0026] In order to achieve the above object, in a third aspect, the present invention further provides a monitoring method for monitoring a linear sound source, which adopts the following technical solution:

[0027] A monitoring method for monitoring linear sound sources uses the monitoring system for monitoring linear sound sources as described in the first aspect, including: installing the system below a power line or other linear sound source; collecting acoustic data through a microphone array; using an adaptive beamforming algorithm to enhance the signal; using a multi-scale signal analysis module to perform time-frequency analysis on the enhanced signal; and determining the location and characteristics of the noise source through a sound source localization and feature extraction module.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] The microphone array in the present invention includes multiple microphones distributed along a partial cylindrical surface, and the multiple microphones are distributed with variable density along the axial direction of the cylinder; the use of a partial cylindrical microphone array significantly improves the spatial coverage capability of linear sound sources; the distribution of the microphones on the cylindrical surface is determined according to the microphone angle, the angle variation amplitude array height, the number of microphones and the nonlinear factor of the height distribution, and an optimized arrangement is achieved in the radial, angular and height dimensions, which significantly improves the spatial sampling efficiency of the array. With the same number of microphones, the sampling points are effectively increased, which can be better used to collect linear sound source signals.

[0030] Based on the optimized arrangement of array microphones, the present invention combines the dynamic reconstruction adaptive array algorithm with the integrated optimization of array arrangement, which significantly improves the flexibility in adapting to different monitoring scenarios.

[0031] In the present invention, the minimum variance distortion-free response beamforming is used to adaptively suppress interference and improve the signal-to-noise ratio, especially in the complex acoustic environment of monitoring linear sound sources such as overhead power lines. In combination with the designed microphone array, interference can be suppressed to the greatest extent and the monitoring accuracy of linear sound sources can be improved. The objective function represents the total power output by the beamformer. By minimizing the total power, the noise and interference from undesired directions are actually minimized, and a set of optimal weights can be obtained, which can suppress interference to the greatest extent while maintaining the response to the target signal.

[0032] In the present invention, by changing the scale parameter, the behavior of the signal at different frequencies can be analyzed; larger scale parameter values ​​correspond to low-frequency analysis, and smaller scale parameter values ​​correspond to high-frequency analysis, so that the system can simultaneously analyze the wide-band noise generated by linear sound sources; the translation parameter allows the analysis of the characteristics of the signal at different time points. Combined with the scale parameter, the wavelet transform provides a time-frequency representation of the signal, which is particularly useful for analyzing non-stationary signals such as transient noise in power lines.

[0033] The present invention uses a multi-scale signal analysis module to effectively process the complex sound field generated by linear sound sources at different distances and directions; it can separate and identify different types of power line noise (such as corona discharge, mechanical vibration, etc.); it can improve the system's noise source positioning and feature extraction capabilities in complex environments; by combining this time-frequency analysis method with a spatial sampling strategy, comprehensive acoustic signal analysis in three dimensions of time, frequency and space is achieved, greatly improving the performance and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings in the specification that constitute a part of this embodiment are used to provide a further understanding of this embodiment. The schematic embodiments of this embodiment and their descriptions are used to explain this embodiment and do not constitute improper limitations on this embodiment.

[0035] Figure 1 It is a structural schematic diagram of the microphone array and monitoring system of the present invention;

[0036] Figure 2 It is a schematic diagram of the variable density distribution of the microphone on a part of the cylindrical surface according to the present invention;

[0037] Figure 3 It is a schematic diagram of the combination of the multi-scale spatial sampling strategy and the variable density distribution of the present invention;

[0038] Figure 4 A schematic diagram of the configuration change of the dynamically reconfigurable adaptive array of the present invention in different working modes;

[0039] Figure 5 This is a diagram showing the effect of combining the near-field acoustic model of a linear sound source with an array arrangement according to the present invention. DETAILED DESCRIPTION

[0040] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0041] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0042] Embodiment 1:

[0043] like Figure 1As shown, this embodiment provides a microphone array for monitoring linear sound sources, and the microphone array includes multiple microphones distributed along a partial cylindrical surface, and the multiple microphones are distributed with variable density along the axial direction of the cylinder; the use of a partial cylindrical microphone array significantly improves the spatial coverage capability of linear sound sources; the distribution of the microphones on the cylindrical surface is determined according to the nonlinear factor of the array height and the height distribution of the number of microphones, and an optimized layout is achieved in the radial, angular and height dimensions, which significantly improves the spatial sampling efficiency of the array. With the same number of microphones, the sampling points are effectively increased, and the flexibility of adapting to different monitoring scenarios is significantly improved by combining the dynamic reconstruction adaptive array algorithm with the integrated optimization of the array layout.

[0044] Optionally, when multiple microphones are distributed with variable density along the axial direction of the cylinder, a low-frequency arrangement area, a medium-frequency arrangement area and a high-frequency arrangement area are set along the axial direction; the density of the microphones in the low-frequency arrangement area should be lower than the density of the microphones in the medium-frequency arrangement area; similarly, the density of the microphones in the medium-frequency arrangement area should be lower than the density of the microphones in the high-frequency arrangement area.

[0045] The distribution of the microphones in space is:

[0046] r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α

[0047] Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; z i is the position of the i-th microphone along the length direction of the axis; H is the array height; α is the nonlinear factor of the height distribution, which can be determined based on experimental data through comparison and screening. For example, among multiple sets of experimental data obtained with multiple different preset nonlinear factor values, the value corresponding to the best signal acquisition effect is selected as the final nonlinear factor.

[0048] Embodiment 2:

[0049] This embodiment provides a monitoring system for monitoring linear sound sources, including a microphone array, a signal acquisition and preprocessing module, an adaptive beamforming processor, a multi-scale signal analysis module, and a sound source localization and feature extraction module. The monitoring system can be installed under a power line to perform all-round monitoring of line noise.

[0050] like Figure 2As shown, the microphones are distributed with variable density on a portion of the cylindrical surface, and a mathematical function is used to optimize the microphone positions to achieve the best sampling of linear sound sources. Figure 2 The figure shows the variable density distribution of microphones in the radial (r), angular (θ) and height (z) dimensions. This distribution enables the system to maximize the effective acoustic aperture while maintaining a compact structure.

[0051] Optionally, according to the monitoring requirements, select an appropriate cylinder radius R (usually 0.1m-2m) and angle range (about 30 degrees). The function determines the exact position of each microphone. The configuration can include 32-64 microphones, which can be adjusted according to actual needs. Install the array below the power line, ensuring that the center of the array is parallel to the line to be monitored.

[0052] like Figure 3 As shown, by implementing sampling intervals of different sizes in a single physical structure, the system is able to simultaneously optimize low-frequency, mid-frequency, and high-frequency performance. Figure 3 The figure shows how different sampling densities are achieved in different parts of the array, with larger spacing in the low-frequency part, medium spacing in the mid-frequency part, and smaller spacing in the high-frequency part. The purpose of this partitioning design is to achieve optimal spatial sampling in different frequency bands, thereby significantly expanding the effective operating frequency band of the system. The various regions work together, with the low-frequency area responsible for capturing large-scale structures and long-distance propagation signals, the mid-frequency area balancing spatial resolution and coverage, and the high-frequency area providing fine spatial details and near-field information. This strategy enables the system to process broadband signals simultaneously, improving the flexibility and adaptability of the system.

[0053] like Figure 4 As shown, the dynamic reconfiguration adaptive array algorithm allows the system to dynamically adjust its effective configuration according to real-time monitoring requirements. Figure 4 The figure shows the configuration changes of the same array in different working modes, which reflects the high flexibility and adaptability of the system. This design enables the system to effectively cope with different noise scenarios and monitoring tasks.

[0054] like Figure 5 The figure shows the effect of combining the near-field acoustic model of a linear sound source with a variable density array arrangement. This model takes into account the special geometric characteristics and near-field effects of the linear sound source, and perfectly combines it with the variable density distribution array. Figure 5 The optimization of the acoustic field distribution and array response is shown in Figure 2, demonstrating the superior performance of the system in near-field source localization and feature extraction.

[0055] The microphone array includes N microphones distributed with variable density along a partial cylindrical surface; optionally, the partial cylindrical surface is a partial cylindrical surface with a range of about 30 degrees. Optionally, the spatial distribution of the microphones can be defined by the following function:

[0056] r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α

[0057] Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; z i is the position of the i-th microphone along the length direction of the axis; H is the array height, and α is the nonlinear factor of the height distribution.

[0058] In the signal acquisition and preprocessing module, each microphone collects acoustic signals, and the sampling rate is usually set to 44.1kHz or higher to capture high-frequency details. After passing through the preamplifier and analog-to-digital converter, the signal enters the preprocessing module for the following processing: noise suppression, using an adaptive filter to remove environmental background noise; signal enhancement, applying spectral subtraction technology to improve the signal-to-noise ratio; frequency equalization, frequency response correction based on the microphone response characteristics; time synchronization, ensuring accurate time alignment of all microphone signals. Signal acquisition and preprocessing can also be achieved through conventional technologies without affecting the implementation of the overall solution.

[0059] The adaptive beamforming processor uses the following optimization algorithm:

[0060] Optionally, the minimum variance distortionless response (MVDR) beamforming can be used to adaptively suppress interference and improve the signal-to-noise ratio, especially in complex acoustic environments such as overhead power line linear sound source monitoring. Combined with the designed microphone array, it can maximally suppress interference and improve the monitoring accuracy of linear sound sources. H Under the condition of a(θ0)=1: min W W H Rw;min W W H Rw describes a constrained optimization problem.

[0061] Where: W represents an N×1 complex column vector, representing the beamforming weight, N is the number of microphones in the microphone array, and each element ω i Corresponding to the complex weight of a microphone; W H represents the conjugate transpose (Hermitian transpose) of W; R represents an N×N spatial covariance matrix, which describes the spatial correlation of the received signal, and its element Rij represents the covariance between the signals of the i-th microphone and the j-th microphone; a(θ0) represents an N×1 complex vector, called a steering vector, which describes the phase relationship of the signal from the desired direction θ0 on each microphone, θ0 represents the direction of arrival of the desired signal, usually the direction of a linear sound source.

[0062] min W W H The physical meaning of the optimization problem described by rw is: the objective function W H Rw represents the total power output by the beamformer. By minimizing this function, we are actually minimizing the noise and interference from undesired directions. Constraint W H a(θ0)=1 ensures that the signal from the desired direction θ0 remains unchanged (unit gain), which is a distortion-free constraint. By solving this optimization problem, we can obtain an optimal set of weights w that can maximally suppress interference while maintaining the response to the target signal.

[0063] This process takes into account the variable density distribution characteristics of the array, and achieves the enhancement of the target direction signal and interference suppression through iterative optimization. The update rate of the algorithm can be dynamically adjusted according to the speed of environmental changes, usually between 10-100 times / second.

[0064] In an embodiment, the minimum variance distortionless response beamforming algorithm is optimized to adapt to a partial cylindrical array with variable density distribution, and the calculation method of a(θ0) is adjusted to take into account the geometric characteristics of the non-uniform array, thereby achieving more precise beam control and more effective noise suppression.

[0065] The multi-scale signal analysis module adopts continuous wavelet transform (CWT):

[0066]

[0067] Wherein, X(a, b) is the result of wavelet transform, which is a two-dimensional function. a is a scale parameter that controls the expansion (frequency) of the wavelet; b is a translation parameter that controls the position of the wavelet on the time axis; x(t) is the input time domain signal. In this embodiment, x(t) represents the acoustic signal collected by the microphone array; ψ a，b (t) is the wavelet basis function, It is a transformed version of the mother wavelet function in time and scale; is the complex conjugate of the wavelet basis function.

[0068] In this embodiment, the physical significance and advantages of the continuous wavelet transform used in the multi-scale signal analysis module are:

[0069] Multiscale analysis: By changing the scale parameter a, the wavelet transform can analyze the behavior of the signal at different frequencies; larger values ​​of the scale parameter a correspond to low-frequency analysis, and smaller values ​​of the scale parameter a correspond to high-frequency analysis; this allows the system to simultaneously analyze broadband noise generated by linear sound sources. Time-frequency localization: The translation parameter b allows the analysis of the characteristics of the signal at different time points; combined with the scale parameter a, the wavelet transform provides a time-frequency representation of the signal; this is particularly useful for analyzing non-stationary signals (such as transient noise in power lines). Adaptive time-frequency resolution: In the low frequency band, the time resolution is low but the frequency resolution is high; in the high frequency band, the time resolution is high but the frequency resolution is low; this adaptability allows the analysis to achieve an optimal balance in both the time and frequency domains. Noise feature extraction: The wavelet transform can effectively extract characteristic patterns in the signal, such as specific frequency components generated by power line faults; wavelet coefficients at different scales can reveal the multiscale structure of the signal, which helps to identify different types of noise sources. Synergy with variable density arrays: The multiscale characteristics of the wavelet transform complement the spatial sampling strategy of the variable density microphone array; the low frequency band corresponds to the part of the array with larger spacing, and the high frequency band corresponds to the part with smaller spacing.

[0070] In an embodiment, the multi-scale analysis method of the multi-scale signal analysis module enables the system to effectively process the complex sound fields generated by linear sound sources at different distances and directions; can separate and identify different types of power line noise (such as corona discharge, mechanical vibration, etc.); can improve the system's noise source positioning and feature extraction capabilities in complex environments; by combining this time-frequency analysis method with a spatial sampling strategy, comprehensive acoustic signal analysis in three dimensions of time, frequency and space is achieved, greatly improving the performance and adaptability of the system.

[0071] Dynamic Reconstruction Adaptive Processing: Based on real-time analysis of the noise environment and target location, the system dynamically adjusts the effective array configuration. This includes selectively activating or deactivating certain microphones and adjusting processing weights. The response time of the reconstruction algorithm is typically controlled within 100ms to adapt to rapidly changing environments.

[0072] The system in this embodiment can also implement the following contents:

[0073] Sound source localization and feature extraction, combined with near-field acoustic model and multi-scale analysis results, the system accurately locates the noise source. Specifically, delay estimation: the generalized cross-correlation method (GCC-PHAT) is used to calculate the delay between different microphone pairs. Triangulation: an overdetermined set of equations is constructed based on delay information and array geometry. Least squares solution: the weighted least squares method is applied to solve the coordinates of the sound source position. Iterative optimization: the Newton-Raphson method is used for position refinement. For typical overhead lines, the system can achieve a positioning accuracy of ±0.5m. The feature extraction process includes: Spectral analysis: the short-time Fourier transform (STFT) is used to obtain the time-varying spectrum. Modulation feature extraction: the envelope spectrum of the spectrum is calculated to identify periodic patterns. Statistical features: the peak factor, skewness, kurtosis and other statistical quantities of the signal are extracted. Time-frequency image features: the time-frequency graph is converted into an image to extract texture and shape features. These features are used for subsequent fault diagnosis and pattern recognition.

[0074] Data fusion and intelligent analysis integrates acoustic data, environmental parameters (such as temperature, humidity, wind speed) and historical data, and uses machine learning algorithms (such as support vector machines or deep neural networks) for comprehensive analysis. This enables the system to distinguish between normal operating noise and potential fault signals with an accuracy of more than 95%.

[0075] Real-time monitoring and early warning, the system runs 24 hours a day and monitors the line status in real time. When an abnormal noise pattern is detected, the early warning mechanism is triggered immediately. The early warning information includes the abnormal type, location, severity, etc., which is transmitted to the control center through a secure network.

[0076] Adaptive learning and optimization: The system continuously learns and optimizes its performance. By accumulating data and feedback, the algorithm parameters and fault models are regularly updated to continuously improve the accuracy and reliability of the system.

[0077] Real-time measurement of environmental parameters and environmental compensation of acoustic measurement. The system integrates temperature, humidity and wind speed sensors for real-time measurement of environmental parameters, specifically:

[0078] Temperature compensation: According to the measured temperature, adjust the sound speed calculation formula c = 331.3 + 0.606T, where T is the Celsius temperature.

[0079] Humidity compensation: Use the modified sound velocity formula c humid =c*(1+0.0016h), where h is the relative humidity percentage.

[0080] Wind speed compensation: Consider the wind speed vector in the sound source localization algorithm and correct the sound wave propagation path.

[0081] Atmospheric absorption compensation: According to ISO 9613-1 standard, the atmospheric absorption coefficient at different frequencies is calculated and compensated in signal processing. These compensation measures significantly improve the accuracy and reliability of the system under different environmental conditions.

[0082] This embodiment can also have the following variations: the array shape can be expanded from a partial cylindrical surface to a complete cylindrical surface or a spherical surface to adapt to different monitoring scenarios. For example, in complex terrain, a 360-degree omnidirectional cylindrical array can be used. Fiber optic sensing technology can be integrated to achieve joint monitoring of acoustics and vibration. It is particularly suitable for detecting mechanical faults of the line, such as looseness or breakage. The system can be deeply integrated with artificial intelligence algorithms to achieve automatic fault diagnosis and predictive maintenance. For example, a long short-term memory network (LSTM) is used to predict the development trend of potential faults. A distributed array network can be developed to deploy multiple collaborative array systems in a large range of lines to achieve wider coverage and more accurate positioning. The system can integrate a visualization module to generate intuitive noise maps and fault diagnosis reports, which are convenient for maintenance personnel to quickly identify problem areas.

[0083] Embodiment 3:

[0084] This embodiment provides a monitoring system for monitoring a linear sound source, which further supplements and verifies the microphone array in Embodiment 1 and the monitoring system in Embodiment 2.

[0085] Partial cylindrical microphone array arrangement used:

[0086] For the first time, the microphone array is arranged on a partial cylindrical surface with a range of about 30 degrees, which is specially optimized for linear sound sources. It innovatively combines the cylindrical geometry with the monitoring requirements of linear sound sources to form a unique sound field acquisition structure.

[0087] The partial cylindrical microphone array used significantly improves the spatial coverage of linear sound sources and expands the effective monitoring angle by about 40% compared to the planar array. The spatial coverage function can be expressed as:

[0088]

[0089] in, represents the array pattern, θ and The two angle variables are the circumferential direction and the axial direction of the cylinder, respectively, and are arranged on a partial cylindrical surface. The integral value increases by about 40%.

[0090] An optimal balance between near-field and far-field performance is achieved; the near-field positioning accuracy is improved by about 50%, which can be expressed by the Cramer-RaoLow Boun lower bound (CRLB):

[0091]

[0092] Among them, CRLB(θ) represents the lower bound of the variance of the estimated direction or position θ of the sound source, which represents the minimum variance that any unbiased estimator can achieve under given conditions; σ 2 represents the noise square, which is the noise variance in the received signal. A larger noise variance will increase CRLB, indicating a decrease in estimation accuracy. T represents the observation time, which is the observation time or signal duration. A longer observation time usually reduces CRLB and improves estimation accuracy. α represents the signal amplitude. |α| 2 Represents the square of the signal amplitude. A stronger signal will reduce CRLB and improve the estimation accuracy. s Represents the sampling frequency. A higher sampling frequency usually provides more information, thus reducing CRLB. represents the summation term, which indicates the sensitivity of the array manifold (the set of phase differences of each array element to sound waves in different directions) to θ. The larger this term is, the more sensitive the array is to directional changes, which is beneficial to improving the estimation accuracy. n (θ) represents the array flow, which is the response of the nth array element to a direction.

[0093] In this embodiment, CRLB provides a theoretical optimal performance limit. The variance of any unbiased estimator cannot be less than this value, which helps to understand the theoretical limit of positioning accuracy under given conditions. By analyzing CRLB, we can understand how to optimize the array design to improve positioning accuracy, such as increasing the number of microphones, optimizing the microphone layout, and improving the signal-to-noise ratio. In the variable density distribution partial cylindrical array, CRLB can help evaluate the advantages of this unconventional design. Based on detailed CRLB analysis and numerical simulation, the variable density distribution partial cylindrical array design shows significant performance advantages in the extended near field range (<30m). In the low frequency range of 50-500Hz, the average positioning accuracy can be theoretically improved by about 30-50%, and the specific improvement varies with the distance; this improvement is mainly due to the ability of our array design to optimize the capture of the near-field effects of linear sound sources. CRLB is usually a function of frequency, which helps to understand the performance of the array in different frequency bands, which is particularly useful for analyzing multi-scale sampling strategies. CRLB provides a benchmark for evaluating the performance of actual algorithms. If the performance of an algorithm is close to CRLB, it can be said to be near-optimal. The CRLB formula clearly shows the impact of various parameters (such as signal-to-noise ratio, sampling rate, and number of array elements) on performance, which helps make wise trade-off decisions in system design. When dealing with the near-field effect of a linear sound source, it is necessary to modify a n The expression for (θ) allows quantifying the impact of near-field effects on positioning accuracy.

[0094] In this embodiment, by using CRLB, not only can the superiority of the array design in the present invention be theoretically proved, but also further optimization can be guided, providing a powerful tool for evaluating and improving the performance of the proposed variable density distribution partial cylindrical microphone array, especially when dealing with complex linear sound source scenes. It provides an effective spatial filtering effect, and through the optimized non-uniform sampling strategy, 50-70% of spatial aliasing can be theoretically suppressed without significantly increasing the computational complexity, thereby improving the performance of the system in complex sound fields.

[0095] A variable density microphone distribution strategy was adopted to develop a non-uniform distribution function, and an optimized layout was achieved in three dimensions: radial, angular, and height. The variable density sampling theory was applied to cylindrical arrays for the first time, creatively solving the problem of wideband monitoring. More array combination possibilities are provided, no longer limited to simple multiple relationship adjustments. The array element position is defined by the functions proposed in Example 1 and Example 2; the spatial sampling efficiency of the array is significantly improved, and the sampling points are effectively increased by 35% with the same number of microphones. The spatial sampling efficiency can be expressed as:

[0096]

[0097] Among them, ψ(r) is the spatial sampling function; r i is the position of the i-th microphone.

[0098] It achieves an effective combination of large aperture and compact structure. Based on the frequency-dependent effective aperture calculation model, the average effective aperture in the target frequency range can be increased by 20-30% while maintaining similar physical dimensions (no more than 15% increase). The directivity and resolution of the low-frequency band are significantly improved. Theoretical analysis shows that in the frequency band below 100Hz, the designed main beam width can be reduced by 30-50% compared with the traditional uniform array. The specific effect depends on the actual array parameter configuration.

[0099] Synergistic effect of partial cylindrical surface arrangement and variable density distribution: For the first time, the organic combination of partial cylindrical surface geometry and variable density distribution is realized, creating a new space-frequency joint optimization paradigm. Balanced performance is achieved in a wide frequency band, and the effective working frequency band is widened by about 1.5 octaves. The frequency band response balance can be expressed as:

[0100]

[0101] Where H(f) is the frequency response function of the system; [f1, f2] is the operating frequency band.

[0102] The system's anti-interference ability in complex noise environments is significantly improved, and the signal-to-noise ratio is increased by 20dB. The output signal-to-noise ratio can be expressed as:

[0103]

[0104] Where w is the weight vector; R s and R n are the covariance matrices of the signal and noise, respectively.

[0105] Adaptive spatial filtering for linear sound sources is implemented. Theoretical analysis and preliminary simulation results show that in typical application scenarios, the gain of the target sound source can be increased by 10-20dB, and the suppression of interference sources can be improved by 20-30dB. The specific performance may vary depending on the actual application environment.

[0106] Deep integration of multi-scale spatial sampling strategy and array arrangement: A multi-scale spatial sampling strategy is implemented, which is perfectly combined with the variable density distribution of some cylindrical surfaces. The sound field sampling of multiple spatial scales is creatively realized in a single physical structure. The effective working frequency band of the system is significantly broadened. Theoretical analysis shows that the effective frequency band coverage of the design can be expanded by 50-150% compared with the traditional design. The multi-scale sampling function can be expressed as:

[0107]

[0108] Among them, w i is the weight of the i-th sampling point; δ is the Dirac function; r*i is the sampling point position.

[0109] A larger effective aperture is achieved in the low frequency band while maintaining high spatial resolution in the high frequency band. Based on theoretical models and preliminary simulations, the overall performance improvement can reach 30-50%. The frequency-dependent effective aperture can be expressed as:

[0110]

[0111] where w(r, f) is the frequency-dependent array weighting function.

[0112] Collaboration with array arrangement: The partial cylindrical arrangement provides an ideal geometric structure for multi-scale sampling, enabling efficient distribution of sampling points of different scales. The variable density distribution strategy provides a mathematical basis for multi-scale sampling and achieves continuous variation of sampling density.

[0113] Integrated optimization of dynamic reconstruction adaptive array algorithm and array layout: A dynamic reconstruction algorithm specifically adapted to variable density distribution on partial cylindrical surfaces has been developed. The software-defined array concept is innovatively combined with a unique physical layout. Dynamic adjustment of the array directionality is achieved, which can be optimized in real time according to the location of the noise source. Compared with fixed arrays, the design has significantly improved flexibility in adapting to different monitoring scenarios, and theoretically can improve scenario adaptability by 60-90%. Dynamic beamforming can be expressed as:

[0114]

[0115] Among them, w i (t) is the time-varying weight; k(θ, φ) is the wave number vector.

[0116] The integration of dynamic reconfiguration adaptive array algorithm and array layout has greatly improved the anti-interference ability of the system. Theoretical analysis and preliminary simulation show that the signal-to-noise ratio can be improved by 15-20dB in complex environments. The specific degree of improvement depends on the characteristics of the interference source and environmental conditions. The adaptive anti-interference performance can be expressed by the minimum variance distortion-free response (MVDR):

[0117]

[0118] Where R is the covariance matrix of the received signal; a is the steering vector of the desired signal.

[0119] Collaboration with array layout: Partial cylindrical layout provides more adjustment space for dynamic reconstruction, making beamforming more flexible. Variable density distribution provides more optimization parameters for dynamic reconstruction algorithm, enhancing adaptive capability.

[0120] Integration of a dedicated near-field acoustic model for linear sound sources: A near-field model for linear sound sources for partially cylindrical variable density arrays was developed. Theoretical analysis shows that the design can significantly improve the accuracy of near-field source positioning. In typical application scenarios, the positioning error can be reduced to 1 / 2 to 1 / 4 of the traditional method; the system's ability to distinguish near-field and far-field noise sources is enhanced, and preliminary simulation results show that the discrimination accuracy can be improved by 50-80%.

[0121] Synergy with array arrangement: The partial cylindrical arrangement provides a more suitable geometric structure for the near-field model and optimizes the collection of near-field sound waves. The variable density distribution enhances the resolution of the near-field model, especially when the sound source is close.

[0122] Deep integration of adaptive multi-scale signal processing algorithms: An adaptive signal processing algorithm for variable density distribution of some cylindrical surfaces has been developed. Efficient extraction of noise features of different scales has been achieved, and the efficiency of theoretical model prediction and analysis can be improved by 40-60%. The performance of the system in complex noise environments has been significantly improved. According to preliminary simulation results, the signal-to-noise ratio can be improved by 10-20dB in typical application scenarios.

[0123] Synergy with array arrangement: The variable density arrangement of some cylindrical surfaces provides rich spatial information for the multi-scale algorithm and enhances the depth of scale analysis. The algorithm makes full use of the characteristics of variable density distribution to achieve more sophisticated spatiotemporal analysis.

[0124] Through the above innovative design, especially the unique combination of partial cylindrical surface arrangement and variable density distribution, as well as the multi-scale sampling, dynamic reconstruction algorithm and adaptive signal processing technology deeply integrated with it, the present invention has achieved a qualitative leap in many aspects such as spatial coverage, bandwidth, directivity, resolution and anti-interference ability. This highly integrated design not only solves many problems faced by traditional microphone arrays when monitoring linear sound sources, but also creates an intelligent monitoring system that can adapt to complex environments and has excellent performance. It provides a highly optimized, high-performance and easy-to-deploy comprehensive solution for noise monitoring, fault diagnosis and preventive maintenance of power systems.

[0125] Embodiment 4:

[0126] This embodiment provides a monitoring method for monitoring linear sound sources, using the monitoring system for monitoring linear sound sources as described in Example 1, including: installing the system below a power line or other linear sound source; collecting acoustic data through a microphone array; using an adaptive beamforming algorithm to perform signal enhancement; using a multi-scale signal analysis module to perform time-frequency analysis on the enhanced signal; and determining the location and characteristics of the noise source through a sound source localization and feature extraction module.

[0127] The above description is only a preferred embodiment of the present embodiment and is not intended to limit the present embodiment. For those skilled in the art, the present embodiment may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present embodiment shall be included in the protection scope of the present embodiment.

Claims

1. A microphone array for monitoring a linear sound source, characterized in that: It includes a plurality of microphones distributed along a portion of a cylindrical surface, wherein the plurality of microphones are distributed in a variable density along the axial direction of the cylinder; When the microphones are distributed on the surface of a cylinder, the distance from the microphone to the axis of the cylinder is equal to the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, where the calculated angle is equal to the sine value of the ratio of the current microphone sequence to the number of microphones; the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, is determined by the power calculation value of the ratio of the current microphone sequence to the number of microphones multiplied by the array height, where the exponent of the power calculation is a nonlinear factor of the height distribution.

2. A microphone array for monitoring a linear sound source as claimed in claim 1, characterized in that: The distribution of the microphones in space is: r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; z i is the position of the i-th microphone along the length direction of the axis; H is the array height; α is the nonlinear factor of the height distribution.

3. A monitoring system for monitoring linear sound sources, characterized in that: It includes an interconnected microphone array, a signal acquisition and preprocessing module, an adaptive beamforming processor, a multi-scale signal analysis module, and a sound source localization and feature extraction module; The microphone array includes a plurality of microphones distributed along a portion of a cylindrical surface, and the plurality of microphones are distributed with variable density along the axial direction of the cylinder; When the microphones are distributed on the surface of a cylinder, the distance from the microphone to the axis of the cylinder is equal to the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, where the calculated angle is equal to the sine value of the ratio of the current microphone sequence to the number of microphones; the product of the angle change amplitude of the microphone along the length direction of the axis and the calculated angle, plus the sum of the starting angles, is determined by the power calculation value of the ratio of the current microphone sequence to the number of microphones multiplied by the array height, where the exponent of the power calculation is a nonlinear factor of the height distribution.

4. A monitoring system for monitoring linear sound sources as claimed in claim 3, characterized in that: The distribution of the microphones in space is: r i =Rθ i =θ0+A·sin(2πi / N)z i =H·(i / N) α Among them, r i is the distance from the i-th microphone to the cylinder axis, i = 1, 2, ..., N, N is the number of microphones; R is the cylinder radius; θ i is the angle of the i-th microphone; θ0 is the starting angle; A is the angle change amplitude; z i is the position of the i-th microphone along the length direction of the axis; H is the array height; α is the nonlinear factor of the height distribution.

5. A monitoring system for monitoring linear sound sources as claimed in claim 3, characterized in that: The adaptive beamforming processor is used to adaptively suppress interference by using minimum variance distortion-free response beamforming; The objective function is to minimize the total power output by the beamformer.

6. A monitoring system for monitoring linear sound sources as claimed in claim 5, characterized in that: In W H Under the condition of a(θ0)=1: min W W H Rw; Where: W represents an N×1 complex column vector, representing the beamforming weight, N is the number of microphones in the microphone array, and each element ω i Corresponding to the complex weight of a microphone; W H represents the conjugate transpose of W; R represents an N×N spatial covariance matrix, whose elements R ij represents the covariance between the signals of the i-th microphone and the j-th microphone; a(θ0) represents an N×1 complex vector, which describes the phase relationship of the signal from the desired direction θ0 on each microphone, and θ0 represents the direction of the linear sound source.

7. A monitoring system for monitoring linear sound sources as claimed in claim 3, characterized in that: The multi-scale signal analysis module uses continuous wavelet transform to perform time-frequency analysis on the collected acoustic signal. The wavelet transform is: Among them, X(a, b) is the result of wavelet transform, which is a two-dimensional function. a is the scale parameter that controls the expansion and contraction of the wavelet; b is the translation parameter; x(t) is the acoustic signal collected by the microphone array; ψ a , b (t) is the wavelet basis function, It is a transformed version of the mother wavelet function in time and scale; is the complex conjugate of the wavelet basis function.

8. A monitoring system for monitoring linear sound sources as claimed in claim 7, characterized in that: By changing the scale parameter, the wavelet transform analyzes the behavior of the signal at different frequencies. Larger scale parameter values ​​correspond to low-frequency analysis, and smaller scale parameter values ​​correspond to high-frequency analysis; the translation parameter allows the analysis of the characteristics of the signal at different time points.

9. A monitoring system for monitoring linear sound sources as claimed in claim 3, characterized in that: The signal acquisition and preprocessing module uses an adaptive filter to remove environmental background noise, applies spectral subtraction technology to improve the signal-to-noise ratio, and performs frequency response correction according to the microphone response characteristics.

10. A monitoring method for monitoring a linear sound source, characterized in that: A linear sound source monitoring system as described in any one of claims 3 to 9 is used, comprising: installing the system below a power line or other linear sound source; collecting acoustic data through a microphone array; using an adaptive beamforming algorithm to perform signal enhancement; using a multi-scale signal analysis module to perform time-frequency analysis on the enhanced signal; and determining the location and characteristics of the noise source through a sound source localization and feature extraction module.

Images