Elevator noise source detection and identification method and device

Abstract

The invention discloses an elevator noise source detection and identification method and device. The method comprises the following steps that acoustic signals of the car top, the car top shaft space, the car bottom and the car bottom shaft space are collected, and vibration signals generated by a traction machine, a guide rail and a car in the elevator running process are collected; estimating time delay between acoustic signals through a generalized cross-correlation algorithm, establishing a nonlinear equation set about a sound source position, and solving to obtain a sound source estimation position; obtaining a sound source position through a beam forming algorithm in combination with the sound source estimation position so as to determine a noise source area; if the position of the sound source points to the interior of the car frame or the interior of the traction machine, the complex sound pressure of an area adjacent to the position of the sound source is measured, and sound field visualization is achieved through near-field acoustical holography so as to locate a noise source area; and judging whether the noise source is an impact sound source or a wind noise source on the same time axis according to the acoustic signal and vibration signal peak conditions of the noise source region. According to the technical scheme, the elevator noise detection efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of elevator detection technology, and in particular to a method and device for detecting and identifying elevator noise sources. Background Technology

[0002] Elevators are essential pieces of equipment in buildings for vertical transportation, and their operating noise levels directly affect passenger comfort and the quality of the building's acoustic environment. Elevator noise testing is a crucial step in ensuring elevator safety, enabling the timely detection of potential mechanical failures and guaranteeing compliant equipment operation.

[0003] However, current testing practices primarily rely on manual methods: on the one hand, inspectors use handheld sound level meters to perform single-point measurements inside the elevator car or at specific locations in the hoistway. This method only obtains local sound pressure levels and cannot reflect the propagation characteristics and distribution patterns of noise in three-dimensional space. On the other hand, technicians subjectively determine the source of noise through experience-based "listening," which makes it difficult to accurately distinguish the specific components generating the noise. For example, it cannot effectively identify whether the noise originates from inside the car frame, the traction machine system, or the guide rail structure. Furthermore, traditional methods rely excessively on accumulated personal experience during fault diagnosis, resulting in a lack of objective data support. This not only prolongs maintenance cycles but may also lead to overlooking critical hidden dangers due to misjudgments, such as mistaking impact noise sources for wind noise sources, thus delaying the implementation of targeted maintenance measures. These problems result in vague noise source localization and low fault handling efficiency, failing to meet the urgent needs of modern elevator safety inspections for precision and automation. Summary of the Invention

[0004] The purpose of this application is to provide a method and device for detecting and identifying elevator noise sources, which can more accurately locate elevator noise sources and improve detection efficiency and accuracy.

[0005] To achieve the above objectives, in a first aspect, this application provides a method for detecting and identifying elevator noise sources, which includes the following steps:

[0006] Acoustic signals are collected from the top of the car, the top hoistway space, the bottom of the car, and the bottom hoistway space. Vibration signals generated by the traction machine, guide rails, and car during elevator operation are also collected. The time delay between acoustic signals is estimated by generalized cross-correlation algorithm, a set of nonlinear equations about the location of the sound source is established, and the estimated location of the sound source is obtained by solving the equations. The location of the sound source is determined by combining the estimated sound source location with a beamforming algorithm to identify the noise source region. If the sound source is located inside the car frame or the traction machine, the complex sound pressure in the vicinity of the sound source is measured, and the sound field is visualized using near-field acoustic holography to locate the noise source area. Based on the peak values ​​of acoustic and vibration signals in the noise source area on the same time axis, the noise source can be determined to be either an impact noise source or a wind noise source.

[0007] Secondly, this application also provides a computer device, including a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned elevator noise source detection and identification method.

[0008] Thirdly, this application also provides a computer-readable storage medium storing an executable program, which, when executed by a processor, implements the above-mentioned elevator noise source detection and identification method.

[0009] This application's technical solution estimates the time delay between acoustic signals through generalized cross-correlation and uses beamforming for localization. It also visualizes and more accurately locates noise sources through near-field acoustic holography. Finally, it combines vibration and acoustic signals to determine the type of noise source for diagnosis. This solves the problems of ambiguous and inefficient noise source localization in existing technologies, accurately locates elevator noise sources, improves detection efficiency and accuracy, and does not rely on the experience of the detection personnel, thus improving the reliability of noise detection.

[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application, it can be implemented according to the contents of the specification. In order to make the above and other objects, features and advantages of this application more obvious and understandable, some embodiments are listed below for detailed description. Attached Figure Description

[0011] Figure 1 A flowchart of an elevator noise source detection and identification method according to at least one embodiment.

[0012] Figure 2 This is a schematic diagram of the structure of a computer device according to at least one embodiment. Detailed Implementation

[0013] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0014] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0015] In the process of elevator noise detection, traditional methods rely on manual handheld sound level meters for single-point measurement or on personal experience for auscultation. This method cannot accurately locate the noise source in the spatial dimension, resulting in inaccurate identification of the sound-generating component. Furthermore, the troubleshooting process is highly dependent on the subjective experience of maintenance personnel, which restricts the efficiency of system maintenance.

[0016] To address this, a method for detecting and identifying elevator noise sources is provided, comprising steps S1 to S7, as shown in the flowchart below. Figure 1 As shown. Overall, in steps S1 to S7, the time delay between acoustic signals is first estimated through generalized cross-correlation, then beamforming is used for localization, and near-field acoustic holography is used for noise source visualization and more precise localization. Finally, the vibration signal and acoustic signal are combined to determine the type of noise source for diagnosis. This solves the problems of ambiguous noise source localization and low efficiency in the prior art, and can accurately locate elevator noise sources, improving detection efficiency and accuracy. Moreover, it does not rely on the experience of the detection personnel, thus improving the reliability of noise detection.

[0017] Step S1 involves collecting acoustic and vibration signals. In step S1, acoustic signals are collected from the top of the car, the top hoistway space, the bottom of the car, and the bottom hoistway space. Vibration signals generated by the traction machine, guide rails, and car during elevator operation are also collected. The purpose of collecting acoustic and vibration signals is to provide raw data for subsequent noise source localization and identification.

[0018] Acoustic signals refer to signals collected by sensors such as microphones, reflecting pressure changes caused by sound waves propagating in a medium. In elevator noise detection, microphones capture various sounds generated during elevator operation, such as mechanical collision sounds and airflow sounds. In at least one embodiment, acoustic signals from the top of the car and the top hoistway space are collected by a microphone array (e.g., 4-8 microphones forming a planar array with a diameter of approximately 0.5-1 meter) installed on the top of the car (e.g., the car upper beam or around the car top guardrail), and acoustic signals from the bottom of the car and the bottom hoistway space are collected by a microphone array (e.g., 4-8 microphones forming a planar array with a diameter of approximately 0.5-1 meter) installed on the bottom of the car (e.g., the car bottom frame).

[0019] Vibration signals refer to signals that reflect the mechanical vibration state of an object, collected by devices such as accelerometers. In elevator noise detection, accelerometers capture the mechanical vibrations generated by elevator components (such as the traction machine, guide rails, and car) during operation, and these vibrations are often closely related to noise generation. In at least one embodiment, accelerometer arrays installed on the traction machine base, guide rail supports, car wall panels, and car frame respectively collect vibration signals from the traction machine, guide rails, and car. That is, the accelerometer array installed on the traction machine base collects the traction machine vibration signal, the accelerometer array installed on the guide rail supports collects the guide rail vibration signal, and the accelerometer array installed on the car wall panels and car frame collects the car vibration signal.

[0020] Microphones and accelerometers can be connected to a multi-channel synchronous data acquisition unit via wired or wireless means to unify the acquired acoustic and vibration signals into the acquisition unit for recording and analysis.

[0021] Step S2 is a preprocessing step. In at least one embodiment, before estimating the time delay between acoustic signals using a generalized cross-correlation algorithm (step S3), step S2 preprocesses the acoustic and vibration signals, including filtering and denoising, and frame segmentation. Filtering and denoising can be performed using common filtering algorithms, such as T1 wavelet filtering, SG filtering, and wavelet denoising filtering. Frame segmentation divides a continuous signal into several segments of equal length, each segment called a frame, to transform the signal into a short-time stationary segment more suitable for algorithm analysis. By performing filtering and denoising on the acoustic and vibration signals and frame segmentation, the accuracy of subsequent signal analysis is improved, the location of the sound source is determined more reliably, thereby improving the accuracy and robustness of the entire elevator noise source detection and identification method. This effectively avoids misjudgments and inefficiencies caused by problems with the quality of the original signal, providing a more accurate basis for elevator fault diagnosis and maintenance.

[0022] Step S3 involves a rough estimation of the sound source location. In step S3, the time delay between acoustic signals is estimated using a generalized cross-correlation algorithm. A system of nonlinear equations concerning the sound source location is established and solved to obtain the estimated sound source location. After acquiring multiple acoustic signals, cross-correlation analysis is performed on any two signals to determine their relative time delay. For example, the cross-correlation function of two signals can be calculated, and its peak value can be used to estimate the time delay. After obtaining a set of time delay estimates, a mathematical model describing the relationship between the sound source location and the time delay can be constructed based on the propagation speed of sound waves in the medium and the geometric positional relationship between the sensors. This model is typically a system of nonlinear equations, which can be solved using iterative optimization algorithms or numerical methods to obtain a preliminary estimate of the sound source location.

[0023] In at least one embodiment, estimating the time delay between any two acoustic signals using a generalized cross-correlation algorithm includes: For every two acoustic signals and Obtained through Fourier Transform (FFT) and Calculate the cross power spectrum Introducing the PHAT weighting function The weighted cross-power spectrum was obtained. Then, the inverse Fourier transform is performed to obtain the generalized cross-correlation function:

[0024] Find the peak value of the generalized cross-correlation function to estimate the time difference between two acoustic signals. .

[0025] Cross-power spectrum (CPS) is a measure of the correlation between two signals at different frequencies, representing the common energy distribution of the two signals in the frequency domain. Calculating CPS allows analysis of the similarity between two acoustic signals at different frequency components, providing a basis for subsequent time delay estimation. CPS can be obtained by multiplying the Fourier transform of one signal by the conjugate of the Fourier transform of the other signal.

[0026] To overcome the interference of noise and reverberation in the elevator operating environment on time delay estimation, a PHAT weighting function is introduced to weight the cross-power spectrum. The PHAT (Phase Transform) weighting function effectively suppresses amplitude information in the cross-power spectrum while highlighting its phase information, as phase information is more critical for time delay determination and is insensitive to noise. The main purpose of the PHAT weighting function is to suppress the influence of noise and reverberation by weighting the cross-power spectrum, thereby highlighting the phase information of the signal and improving the accuracy and robustness of time delay estimation. By normalizing the amplitude of the cross-power spectrum and retaining only phase information, the PHAT weighting function allows for the effective extraction of the relative phase relationship even when the signal amplitude is disturbed in complex environments. The PHAT weighting function is typically defined as the reciprocal of the cross-power spectrum; multiplying it by the cross-power spectrum yields the weighted cross-power spectrum. Compared to direct cross-correlation calculation, this significantly improves the accuracy and robustness of time delay estimation.

[0027] The inverse Fourier transform (IFT) is a mathematical operation that converts a frequency-domain signal back to a time-domain signal. After obtaining the weighted cross-power spectrum, the IFT is used to convert it back to the time domain, yielding the generalized cross-correlation function. This function represents the correlation between two acoustic signals in the time domain, and due to the PHAT weighting, its peak position can more accurately indicate the signal delay.

[0028] In at least one embodiment, a system of nonlinear equations concerning the location of the sound source is established, and the estimated location of the sound source is obtained by solving the system, including: By iterating through all acoustic signals, a set of time delay estimates is obtained. Based on the time delay estimate, a set of nonlinear equations about the sound source location is established geometrically:

[0029] in Location of the sound source Let i be the position of the i-th microphone. For the j-th microphone position, For the speed of sound, The sampling period; Solving the nonlinear equations yields the estimated location of the sound source. .

[0030] Specifically, the acoustic signals collected by the microphone array are fully utilized to obtain comprehensive time delay information for sound source localization. By pairwise combining all microphone pairs in the array and calculating their time delays, a set of time delay estimates containing sufficient redundancy can be constructed, thereby improving the robustness of localization. Specifically, an iterative loop can be used, selecting each microphone in the array as a reference point and calculating its time delay with all other microphones in the array. Geometrically, the difference in straight-line distance between the sound source location and the microphone location causes signal time delay. Combining this geometric relationship with time delay estimates, sound velocity, and other parameters, a system of nonlinear equations concerning the sound source location can be constructed.

[0031] Step S4 calculates the sound source location based on the estimated sound source location. In step S4, the sound source location is determined using a beamforming algorithm in conjunction with the estimated sound source location to identify the noise source region. After obtaining the preliminary estimated sound source location, this location information can guide the implementation of the beamforming algorithm. For example, a small area centered on the estimated sound source location can be defined as the scanning range of the beamforming algorithm, instead of blindly scanning the entire space.

[0032] In at least one embodiment, step S4 specifically includes: Centered on the estimated location of the sound source, a three-dimensional search space is defined as the range of beamforming scanning and discretized into grid points; The steering vector for constructing a sensor (microphone) array for acquiring acoustic signals:

[0033] Calculate the covariance matrix of the acquired acoustic signal:

[0034] in This is a column vector consisting of the instantaneous signal amplitudes detected by all sensors at the nth sampling point; Scan the defined three-dimensional search space and calculate the output power at each grid point:

[0035] Finding output power The peak value is used to determine the location of the sound source.

[0036] In this study, a three-dimensional search space is defined centered on the estimated sound source location to serve as the scanning range for beamforming. This space is discretized into grid points to delineate a region of interest for fine-grained searching, thereby reducing computational load and improving search efficiency. This three-dimensional search space can be a spherical or cubic space centered on the estimated sound source location, and its size can be determined based on the confidence level of the estimated sound source location or a preset search range. Discretizing this continuous three-dimensional space into grid points facilitates point-by-point scanning and computation by the computer; the density of the grid points directly affects the final positioning accuracy and computational efficiency.

[0037] A steering vector describes the phase and amplitude relationship of a sound wave as it propagates from a specific spatial location (e.g., a grid point) to each microphone in a sensor (microphone) array. The steering vector can be calculated based on the geometry of the microphone array (e.g., sensor positions, spacing) and the speed of sound, with each element representing the relative phase delay of the sound wave arriving at the corresponding microphone. In some implementations, a plane wave model can be used to construct the steering vector; in other implementations, particularly for near-field localization, a spherical wave model can be used, whose steering vector incorporates distance attenuation information.

[0038] The covariance matrix of the acquired acoustic signals is calculated, which describes the statistical correlation between the acoustic signals received by the microphone array. This matrix contains the power information of the signals, as well as the energy and phase relationships between the signals from different sensors. It is a key input used in beamforming algorithms to estimate the power distribution of the sound source.

[0039] The core computational process of beamforming algorithms involves scanning the defined three-dimensional search space and calculating the output power at each grid point. Higher output power indicates a greater likelihood that the grid point is a sound source. Common beamforming algorithms include delay-sum beamforming and minimum variance distortionless response (MVDR) beamforming.

[0040] Finding the peak output power to locate the sound source involves calculating the output power of all grid points in the three-dimensional search space to determine the most likely location of the sound source. The grid point with the highest output power is considered the precise location of the sound source. Finding the peak output power can be achieved using a simple maximum search algorithm, iterating through the output power of all grid points to find the maximum value and its corresponding coordinates. To further improve accuracy, after finding the initial peak, local interpolation or a finer grid search can be performed around that peak point to obtain the sound source location with sub-grid precision.

[0041] The elevator noise source detection and identification method of this application effectively combines the advantages of two localization methods by using the sound source estimation location obtained by the generalized cross-correlation algorithm as the initial search range of the beamforming algorithm. First, the generalized cross-correlation algorithm provides a preliminary sound source location estimate, which eliminates the need for the beamforming algorithm to perform a blind search throughout the entire elevator shaft space, thus significantly reducing the search range and improving computational efficiency. Second, the beamforming algorithm utilizes the spatial sampling information of the microphone array to perform high-resolution spatial spectrum estimation of the sound field by constructing a steering vector and calculating the covariance matrix. By scanning and calculating the output power of each grid point within a defined three-dimensional search space, the precise location of the sound source can be accurately identified. This combined approach overcomes the limitations of single methods in terms of localization accuracy or computational efficiency in complex acoustic environments, such as elevator shafts, thereby obtaining a more accurate noise source location.

[0042] Step S5 involves visualizing the sound field using near-field acoustic holography. In step S5, it is first determined whether the sound source is located inside the car frame or the traction machine, as these areas are difficult to access or visualize, making it difficult to pinpoint the noise source region. If the sound source is located inside the car frame or the traction machine, the complex sound pressure level in the vicinity of the sound source is measured, and near-field acoustic holography is used to visualize the sound field, thereby locating the noise source region.

[0043] In at least one embodiment, if the sound source is located inside the car frame or the traction machine, the complex sound pressure in the vicinity of the sound source is measured, and near-field acoustic holography is used to visualize the sound field in order to locate the noise source region, including: If the sound source is located inside the car frame or the traction machine, the complex sound pressure level in the vicinity of the sound source location is measured using a nearby microphone array. The distance between microphones in the array must be less than the wavelength of the sound wave in order to capture evanescent waves on the surface of the sound source. The wavenumber spectrum of the microphone array surface was obtained by two-dimensional Fourier transform. It contains information about the direction of sound field propagation and energy distribution in space; By utilizing the principle of inverse sound field propagation, multiplying the wavenumber by the transfer function in the wavenumber domain yields the wavenumber spectrum of the sound pressure at the sound source surface: ,in This is the distance from the microphone array surface to the sound source surface; Inversely, the normal vibration velocity of the sound source surface is obtained: ,in Angular frequency, For the density of the propagation medium, Normal wave number; Finally, the vibration distribution of the sound source surface is obtained through inverse Fourier transform. A visual acoustic image is generated based on the vibration distribution, and the noise source region is determined from the acoustic image.

[0044] The method involves measuring the complex sound pressure level (SPL) in the vicinity of the sound source using a microphone array positioned nearby, aiming to obtain sound field information in the area surrounding the noise source. Complex SPL refers to a physical quantity that includes both sound pressure amplitude and phase, and it can completely describe the propagation state of sound waves in space. Measurement using a microphone array allows for the simultaneous acquisition of SPL information from multiple spatial points, providing a data foundation for subsequent sound field reconstruction. The microphone array can be a planar array, a conformal array, or an arbitrary shape array, and its arrangement can be adjusted as needed according to the geometry and spatial constraints of the area to be measured.

[0045] Multiplying the transfer function in the wavenumber domain yields the sound pressure wavenumber spectrum at the source surface, which is used to propagate the sound field information from the measurement plane back to the source plane. The transfer function describes the attenuation and phase change of the sound wave as it propagates from the source plane to the measurement plane. In the wavenumber domain, the propagation characteristics of the sound field can be represented by simple multiplication operations, thus enabling the reverse propagation of the sound field.

[0046] In near-field acoustic holography (NAH) processing, , and These are three fundamental physical quantities, and their calculation methods are clear and fixed, as follows: 1. Density of the propagation medium

[0047] Value: For air, under standard conditions (20°C), it is usually taken as...

[0048] If higher accuracy is required, corrections can be made based on temperature and humidity, but this constant is generally used directly in engineering.

[0049] 2. Angular frequency

[0050] Meaning: The time angular frequency of a sound wave.

[0051] Calculation: Directly obtained from the analysis frequency f (in Hz):

[0052] For example, when f=1000Hz .

[0053] 3. Normal wavenumber

[0054] Meaning: The component of the wavenumber vector perpendicular to the measurement surface (z-direction), which determines the propagation or attenuation characteristics of the sound wave along the z-direction.

[0055] Calculation steps: 3.1 Total wavenumber k:

[0056] Where c is the speed of sound (usually taken as c = 340 m / s in air).

[0057] 3.2 Transverse wavenumber , : In the plane NAH based on spatial Fourier transform, they are determined by spatial sampling:

[0058] in , For measuring the surface dimensions, m and n are integer indices (corresponding to the discrete wavenumbers after FFT). In practice, m and n are generated through the FFT frequency axis.

[0059] 3.3 Calculation : · like (Propagation wave):

[0060] at this time Let be a real number, and let be the sound wave propagating along the z-direction.

[0061] · like (Fleeting Waves):

[0062] at this time Since it is a purely imaginary number, the sound wave decays exponentially along the z-direction, and the decay constant is determined by the imaginary part.

[0063] In numerical implementation, complex number operations can be used to process them directly:

[0064] In environments that support complex numbers (such as MATLAB and Python), taking the square root of a negative number will automatically yield an imaginary result.

[0065] The purpose of retrieving the normal velocity of the sound source surface is to derive the vibration characteristics of the sound source surface from the sound pressure wavenumber domain spectrum. Normal velocity is a physical quantity describing the vibration intensity of the sound source surface, and it is closely related to the radiation efficiency of the sound source and the noise generation mechanism. The inverse Fourier transform yields the vibration distribution of the sound source surface, aiming to convert the normal velocity spectrum from the wavenumber domain back to the spatial domain, thus obtaining the specific vibration distribution of the sound source surface. Generating an acoustic image based on the vibration distribution can be achieved using existing general-purpose data visualization tools, such as MATLAB, Python's matplotlib library, or LabVIEW, by mapping the vibration distribution data onto a two-dimensional or three-dimensional graphical interface and combining it with an appropriate color mapping scheme. Analysis of the generated visualized acoustic image clearly identifies the areas with the highest vibration intensity, which are the main sources of noise. The generated visualized acoustic image provides high-resolution spatial information, enabling precise identification of specific sound-generating components or defects, even in structurally complex internal areas or hard-to-reach regions. The analysis of sonograms can be performed visually by professionals or automatically by image processing algorithms, such as threshold segmentation, region growing, or cluster analysis, to automatically identify and delineate high-vibration areas, thereby automatically determining the noise source region.

[0066] In step S5, if the sound source location is not pointing to the inside of the car frame or the traction machine, the sound source location is relatively easy to reach, such as the guide rail, and there is no need to use near-field acoustic holography for sound field visualization. The location pointed to by the sound source location is the noise source area.

[0067] Step S5, combined with the beamforming localization method described in step S4, forms a coarse-to-fine localization strategy. When the beamforming algorithm can only provide a general noise source area (such as inside the elevator car or traction machine), step S5 can further focus the noise, providing high-resolution sound field visualization. This allows for precise identification of specific sound-generating components or defect locations within these complex and difficult-to-observe areas. This combination enables elevator noise source detection and identification to extend beyond the macroscopic level to the microscopic component level, significantly improving the accuracy and efficiency of fault diagnosis.

[0068] Step S6 involves simultaneously analyzing the collected acoustic and vibration signals within the located noise source area to distinguish the physical causes of the noise, aiming to provide a more refined basis for subsequent fault diagnosis. In step S6, the noise source is determined to be an impact noise source or a wind noise source based on the peak values ​​of the acoustic and vibration signals in the noise source area on the same time axis.

[0069] When determining whether a noise source is an impact noise source or a wind noise source, if the vibration signal also reaches a peak before the acoustic signal peaks, or if the vibration signal reaches a peak at the same time as the acoustic signal peaks, then the noise source is determined to be an impact noise source. Impact noise sources usually originate from the instantaneous contact, collision, or impact of mechanical parts. Therefore, the peak value of the vibration signal is often highly correlated with the peak value of the acoustic signal in time, and the vibration peak value may even slightly precede the acoustic peak value. This determination can be achieved by performing time synchronization analysis on the acoustic and vibration signals. For example, by calculating the cross-correlation function of the signals, or by setting a time window before the time point when the acoustic signal peak occurs, and detecting whether there is a significant peak value of the vibration signal within that window.

[0070] If the vibration signal does not peak before or after the acoustic signal peaks, the noise source is identified as a wind noise source. Wind noise sources are typically caused by airflow disturbances, aerodynamic effects, or ventilation system problems, and are mainly manifested as changes in the acoustic signal without significant mechanical vibration peaks. Therefore, when the acoustic signal peaks, if the amplitude of the vibration signal remains at a low level without significant peaks for a certain period before and after the peak, the noise source can be inferred to be a wind noise source. This can be achieved by setting a time interval before and after the acoustic signal peak, monitoring the amplitude of the vibration signal within this interval, and comparing it with a preset threshold.

[0071] This time-series-based judgment mechanism enables a deeper identification of the physical causes of noise within the located noise source area, thus providing a more accurate basis for subsequent fault diagnosis and repair.

[0072] Step S7 is the output and data storage step. In step S7, based on the previously obtained noise source results (impact noise source or wind noise source), it is determined whether a fault has occurred. For example, if an impact noise source is present near a location where impact noise should not exist, it is determined to be a fault. Or, if all noise is from wind noise sources, it is determined to be working normally. When it is determined to be normal, the normal status is output, and no elevator maintenance is required. When a fault is determined, the detection conclusion is output for maintenance personnel to refer to during repair, and relevant data is stored in the fault database, including acoustic signal segments, vibration signal segments, time-frequency domain feature vectors extracted in the previous steps, coordinate positions, fault types, judgment conclusions, confidence levels, elevator metadata, etc. This stored data can be used by subsequent maintenance personnel for reference and auxiliary diagnosis, or by the algorithm model to call the data through the database interface for operations such as labeling, retrieval, machine learning iteration, and fault prediction.

[0073] In embodiments of this application, a computer device 90 is also provided, such as... Figure 2As shown, the device includes a processor 91, a memory 92, and a computer program 93 stored in the memory 92 and executable on the processor. When the processor 91 executes the computer program 93, it implements the elevator noise source detection and identification method of one or more embodiments described above. By deploying the elevator noise source detection and identification method on the computer device 90, the automated and efficient execution of the detection and identification method of this application is achieved.

[0074] Computer device 90 refers to an electronic device capable of executing instructions, processing data, and storing information. It can be a general-purpose computer system, such as a personal computer, server, or workstation, or an embedded system or dedicated computing platform customized for a specific application. This device provides the necessary hardware foundation and operating environment for complex computing tasks.

[0075] Processor 91 is the core computing unit of a computer device, responsible for interpreting and executing instructions in computer programs, performing arithmetic and logical operations, and data processing. Processor 91 can be a central processing unit (CPU) for general computing and control tasks; it can also be a graphics processing unit (GPU), which excels at parallel computing, especially suitable for training and inference of deep learning models; or it can be a field-programmable gate array (FPGA) or application-specific integrated circuit (ASIC) for implementing highly optimized specific algorithms.

[0076] Memory 92 is a hardware component used to store data and computer programs 93. Memory 92 may include random access memory (RAM) for temporary storage of running programs and data for fast access by the processor; it may also include read-only memory (ROM) or non-volatile memory (such as solid-state drives (SSDs), hard disk drives (HDDs), or flash memory) for long-term storage of the operating system, applications, and large amounts of data.

[0077] The computer program 93, stored in memory 92 and executable on the processor, is a collection of instructions stored in memory 92 in a form that can be understood and executed by processor 91. When processor 91 loads and executes these instructions, it operates according to a predetermined logical flow. This program can be an executable file compiled from a high-level language, interpreted code written in a scripting language, or low-level instructions existing as firmware. When processor 91 runs according to the instructions of computer program 93 stored in memory 92, it executes the steps and operations defined by the program one by one. Processor 91 is responsible for coordinating data flow, performing computational tasks, and managing storage resources, thereby fully realizing all aspects of the elevator noise source detection and identification method.

[0078] The computer device 90 can also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), one or more devices that enable a user to interact with the electronic device, and / or any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication can be performed via an input / output (I / O) interface. Furthermore, the computer device 90 can also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter. The network adapter communicates with other modules of the computer device 90 via a bus. It should be understood that other hardware and / or software modules can be used in conjunction with the computer device 90, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0079] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this application.

[0080] In embodiments of this application, a computer-readable storage medium is also provided, which stores an executable program. When the executable program is executed by a processor, it implements the elevator noise source detection and identification method described above. When the executable program is run, the program code is used to cause the processor to perform the steps described in this specification according to the various exemplary embodiments of this application.

[0081] A computer-readable storage medium is a physical medium capable of storing digital data or instructions that can be read and executed by a computer system. This medium can be a non-volatile storage medium, such as a hard disk drive (HDD), solid-state drive (SSD), flash memory, or optical disc (CD-ROM, DVD-ROM), used for long-term storage of executable programs. Alternatively, it can be a volatile storage medium, such as random access memory (RAM), used for temporarily storing instructions and data during program execution.

[0082] An executable program is a collection of instructions that, after being compiled or interpreted, can be directly executed by a computer's processor to complete the various tasks of the detection and recognition method. This program can be written in high-level programming languages ​​(such as Python, C++, MATLAB, etc.) and utilize deep learning frameworks (such as TensorFlow, PyTorch, Keras, etc.) to implement neural network models and algorithmic logic.

[0083] A processor is the core component of a computer system, responsible for reading and executing instructions from computer programs to perform arithmetic, logical, and control operations. This processor can be a central processing unit (CPU), providing general-purpose computing power; it can be a graphics processing unit (GPU), particularly suitable for massively parallel computing tasks in deep learning to accelerate the training and inference processes of neural networks; or it can be an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA), optimized for specific computing tasks.

[0084] It should be understood that other embodiments of this application will readily conceive of by those skilled in the art upon consideration of the specification and practice of the application disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. This application is not limited to the methods / structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The specification and embodiments are to be considered exemplary only, and the scope of this application is defined only by the appended claims.

Claims

1. A method for detecting and identifying elevator noise sources, characterized in that, It includes the following steps: Acoustic signals are collected from the top of the car, the top hoistway space, the bottom of the car, and the bottom hoistway space. Vibration signals generated by the traction machine, guide rails, and car during elevator operation are also collected. The time delay between acoustic signals is estimated by generalized cross-correlation algorithm, a set of nonlinear equations about the location of the sound source is established, and the estimated location of the sound source is obtained by solving the equations. The location of the sound source is determined by combining the estimated sound source location with a beamforming algorithm to identify the noise source region. If the sound source is located inside the car frame or the traction machine, the complex sound pressure in the vicinity of the sound source is measured, and the sound field is visualized using near-field acoustic holography to locate the noise source area. Based on the peak values ​​of acoustic and vibration signals in the noise source area on the same time axis, the noise source can be determined to be either an impact noise source or a wind noise source.

2. The elevator noise source detection and identification method as described in claim 1, characterized in that, The estimation of time delay between acoustic signals using the generalized cross-correlation algorithm includes: For every two acoustic signals and Obtained through Fourier transform and Calculate the cross power spectrum Introducing the PHAT weighting function The weighted cross-power spectrum was obtained. Then, the inverse Fourier transform is performed to obtain the generalized cross-correlation function: Find the peak value of the generalized cross-correlation function to estimate the time difference between two acoustic signals. .

3. The elevator noise source detection and identification method as described in claim 2, characterized in that, The process of establishing a system of nonlinear equations about the location of the sound source and solving it to obtain an estimated location of the sound source includes: By iterating through all acoustic signals, a set of time delay estimates is obtained. Based on the time delay estimate, a set of nonlinear equations about the sound source location is established geometrically: in Location of the sound source For the speed of sound, The sampling period; Solving the nonlinear equations yields the estimated location of the sound source. .

4. The elevator noise source detection and identification method as described in claim 1, characterized in that, The step of determining the noise source region by combining the estimated sound source location with a beamforming algorithm includes: Centered on the estimated location of the sound source, a three-dimensional search space is defined as the range of beamforming scanning and discretized into grid points; The steering vector for constructing the sensor array that acquires acoustic signals: Calculate the covariance matrix of the acquired acoustic signal: in This is a column vector consisting of the instantaneous signal amplitudes detected by all sensors at the nth sampling point; Scan the defined three-dimensional search space and calculate the output power at each grid point: Finding output power The peak value is used to determine the location of the sound source.

5. The elevator noise source detection and identification method as described in claim 1, characterized in that, If the sound source is located inside the car frame or the traction machine, the complex sound pressure in the vicinity of the sound source is measured, and near-field acoustic holography is used to visualize the sound field in order to locate the noise source region, including: If the sound source is located inside the car frame or the traction machine, the complex sound pressure in the vicinity of the sound source location is measured by a microphone array placed nearby. ; The wavenumber spectrum of the microphone array surface was obtained by two-dimensional Fourier transform. ; Multiplying by the transfer function in the wavenumber domain yields the wavenumber domain spectrum of the sound pressure at the source surface: ,in This is the distance from the microphone array surface to the sound source surface; Inversely, the normal vibration velocity of the sound source surface is obtained: Finally, the vibration distribution of the sound source surface is obtained through inverse Fourier transform. A visual acoustic image is generated based on the vibration distribution, and the noise source region is determined from the acoustic image.

6. The elevator noise source detection and identification method as described in claim 1, characterized in that, When determining whether a noise source is an impact noise source or a wind noise source, if the vibration signal also reaches a peak before the acoustic signal reaches a peak, or if the vibration signal also reaches a peak at the same time as the acoustic signal reaches a peak, then the noise source is determined to be an impact noise source; if the vibration signal does not reach a peak before or after the acoustic signal reaches a peak, then the noise source is determined to be a wind noise source.

7. The elevator noise source detection and identification method as described in claim 1, characterized in that, Before estimating the time delay between acoustic signals using the generalized cross-correlation algorithm, the acoustic and vibration signals are preprocessed, including filtering and denoising, and frame segmentation.

8. The elevator noise source detection and identification method as described in claim 1, characterized in that, Acoustic signals from the top of the car and the top hoistway space are collected using a microphone array on the top of the car, and acoustic signals from the bottom of the car and the bottom hoistway space are collected using a microphone array at the bottom of the car. Vibration signals from the traction machine, guide rails, and car are collected using an array of acceleration sensors installed on the traction machine base, guide rail supports, car wall panels, and car frame, respectively.

9. A computer device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the elevator noise source detection and identification method as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing an executable program, characterized in that, When the executable program is executed by the processor, it implements the elevator noise source detection and identification method as described in any one of claims 1 to 8.

Images