A pick wear detection method and system for a roadheader

Abstract

The application discloses a pick wear detection method and system for a heading machine, and relates to the technical field of mine machinery state monitoring. The method comprises the following steps: acquiring a pick tool seat and a cutter head, the cutter head being fixed to the tool seat through a plurality of rotating shafts, and a plurality of high-bandwidth acoustic emission sensors being arranged on the rotating shafts; identifying a rotating detection signal by using the high-bandwidth acoustic emission sensors, and extracting features such as acoustic emission pulse starting time and peak amplitude; constructing a sound source spatial positioning model to obtain a sound source positioning result; mapping the sound source positioning result to a rotating shaft three-dimensional model to identify hidden wear, and outputting a risk level. The application solves the technical problem that the traditional detection method is difficult to accurately identify the hidden wear of the rotating shaft of the pick of the heading machine, and cannot comprehensively and accurately obtain the wear information to meet the precise evaluation requirement, achieves the technical effect of accurately identifying the hidden wear of the rotating shaft of the pick of the heading machine and evaluating the risk level, and improves the accuracy and comprehensiveness of the pick wear detection.

Description

Technical Field

[0001] This invention relates to the field of mining machinery condition monitoring technology, and in particular to a method and system for detecting wear of cutting teeth in tunneling machines. Background Technology

[0002] In mining operations, the wear condition of tunneling machine cutting teeth directly affects tunneling efficiency, equipment safety, and mining costs; therefore, accurate detection is crucial for ensuring production. Current technologies for detecting tunneling machine cutting tooth wear primarily rely on manual inspections or traditional sensor monitoring. While these methods are effective in situations where equipment is stopped or under simple operating conditions, their limitations become apparent in complex mining environments as mining intensity increases. Traditional methods struggle to detect the hidden wear on the cutting tooth's rotating shaft, failing to comprehensively acquire wear data, resulting in inaccurate detection and failing to meet the demands for precise assessment and timely maintenance of cutting tooth wear. Summary of the Invention

[0003] This application provides a method and system for detecting wear on the cutting teeth of a tunneling machine, which solves the technical problem that traditional detection methods are unable to accurately identify the hidden wear on the rotating shaft of the cutting teeth of the tunneling machine, and cannot comprehensively and accurately obtain wear information to meet the needs of precise assessment.

[0004] The first aspect of this application provides a method for detecting wear of cutting teeth in a tunneling machine. The method includes: acquiring a cutting tooth tool holder and a cutting tooth head, the cutting tooth head being fixed to the cutting tooth tool holder via multiple rotating shafts; setting multiple high-bandwidth acoustic emission sensors on the multiple rotating shafts; using the multiple high-bandwidth acoustic emission sensors to identify multiple rotation detection signal data of the multiple rotating shafts; extracting the acoustic emission pulse start time and acoustic emission pulse characteristics of the multiple rotation detection signal data, wherein the acoustic emission pulse start time is the arrival time of the acoustic emission signal, and the acoustic emission pulse characteristics include peak amplitude, duration, spectral center frequency, energy, and envelope shape; constructing a sound source spatial localization model, wherein the sound source spatial localization model locates the sound source by analyzing the acoustic emission pulse start time and acoustic emission pulse characteristics to obtain a sound source localization result; mapping the sound source localization result to a three-dimensional structural model of the rotating shafts for latent wear identification, and outputting a latent wear risk level.

[0005] A second aspect of this application provides a cutter wear detection system for a tunneling machine, the system comprising: an acoustic emission sensor deployment module for acquiring a cutter tool holder and a cutter head, the cutter head being fixed to the cutter tool holder via multiple rotating shafts, and multiple high-bandwidth acoustic emission sensors being installed on the multiple rotating shafts; an acoustic emission pulse acquisition module for identifying multiple rotation detection signal data of the multiple rotating shafts using the multiple high-bandwidth acoustic emission sensors, extracting the acoustic emission pulse start time and acoustic emission pulse characteristics of the multiple rotation detection signal data, the acoustic emission pulse start time being the arrival time of the acoustic emission signal, and the acoustic emission pulse characteristics including peak amplitude, duration, spectral center frequency, energy, and envelope shape; a sound source localization result acquisition module for constructing a sound source spatial localization model, the sound source spatial localization model obtaining a sound source localization result by locating the sound source based on the acoustic emission pulse start time and acoustic emission pulse characteristics; and a latent wear risk level acquisition module for mapping the sound source localization result to a three-dimensional structural model of the rotating shaft for latent wear identification and outputting a latent wear risk level.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] This application utilizes high-bandwidth acoustic emission sensors installed on multiple rotating shafts of the cutting tooth to identify the rotation detection signal data of the rotating shafts. The starting time and pulse characteristics of the acoustic emission pulses are extracted, and a spatial localization model of the sound source is constructed to obtain the sound source localization result. This result is then mapped onto a three-dimensional structural model of the rotating shaft for latent wear identification, outputting the latent wear risk level. This allows for precise detection of the latent wear of the tunneling machine cutting tooth, making the cutting tooth wear detection results more accurate and reliable. This achieves the technical effect of accurately identifying and assessing the risk level of latent wear on the rotating shaft of the tunneling machine cutting tooth, improving the accuracy and comprehensiveness of cutting tooth wear detection. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic flowchart of a method for detecting wear of cutting teeth in a tunneling machine, provided in an embodiment of this application.

[0010] Figure 2 This is a schematic diagram of the structure of a cutter tooth wear detection system for a tunneling machine provided in an embodiment of this application.

[0011] Figure labeling: 1. Acoustic emission sensor deployment module; 2. Acoustic emission pulse acquisition module; 3. Sound source localization result acquisition module; 4. Latent wear risk level acquisition module. Detailed Implementation

[0012] This application provides a method and system for detecting wear on the cutting teeth of a tunneling machine, which solves the technical problem that traditional detection methods are unable to accurately identify the hidden wear on the rotating shaft of the cutting teeth of the tunneling machine, and cannot comprehensively and accurately obtain wear information to meet the needs of precise assessment.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0015] Example 1, as Figure 1 As shown, a method for detecting wear of cutting teeth in a tunneling machine, wherein the method includes:

[0016] Step A100: Obtain a cutting tool holder and a cutting tool head. The cutting tool head is fixed to the cutting tool holder via multiple rotating shafts, and multiple high-bandwidth acoustic emission sensors are arranged on the multiple rotating shafts.

[0017] In this embodiment, the cutting tool holder is a component used to fix the cutting tool head. The cutting tool head is connected to it through multiple rotating shafts, together forming the core structure of the tunneling machine cutting tool. Its geometric dimensions and positional relationships are important data for the 3D modeling of the cutting tool. The cutting tool head is the part of the tunneling machine cutting tool that directly participates in rock breaking and coal mining operations. It is fixed to the cutting tool holder through multiple rotating shafts. Its connection structure with the tool holder and its operating status directly affect the wear of the cutting tool, and it is also a key component that needs to be included in the 3D modeling.

[0018] Specifically, in mining operations, the cutting head and tool holder of a tunneling machine are core components that directly participate in rock breaking and coal mining. These two components are connected and move relative to each other through multiple rotating shafts. These shafts are prone to wear during long-term friction and stress, especially latent wear, which is difficult to detect using traditional methods. Therefore, it is essential to obtain suitable cutting head and tool holders to ensure structural compatibility and guarantee stable operation of the rotating shafts during operation.

[0019] Since wear on rotating shafts is accompanied by specific acoustic emission signals, and traditional sensors have limited bandwidth, making it difficult to fully capture the wear-related broadband signals, multiple high-bandwidth acoustic emission sensors need to be installed on multiple rotating shafts. For example, for three rotating shafts, two high-bandwidth acoustic emission sensors are arranged on each shaft, for a total of six sensors, which are installed at different stress points on the rotating shafts to ensure that acoustic emission signals generated by friction, collision, etc., can be collected from multiple angles. These signals contain key information about the wear state.

[0020] Through the above steps, the high-bandwidth acoustic emission sensor can capture the acoustic emission signals during the operation of the rotating shaft in real time, providing raw data support for the subsequent extraction of features such as the start time, peak amplitude, and center frequency of the acoustic emission pulse, laying the foundation for accurate detection of hidden wear on the rotating shaft.

[0021] Step A200: Use the multiple high-bandwidth acoustic emission sensors to identify multiple rotation detection signal data of the multiple rotating shafts, and extract the acoustic emission pulse start time and acoustic emission pulse characteristics of the multiple rotation detection signal data. The acoustic emission pulse start time is the arrival time of the acoustic emission signal, and the acoustic emission pulse characteristics include peak amplitude, duration, spectral center frequency, energy, and envelope shape.

[0022] In this embodiment, the envelope shape is a characteristic of acoustic emission pulses, referring to the contour shape of the acoustic emission signal after processing, which can reflect the overall trend of the acoustic emission signal.

[0023] Optionally, during the operation of the cutting gear, acoustic emission signals generated by multiple rotating shafts due to friction, force deformation, etc., will propagate to the surroundings. However, traditional sensors have narrow bandwidths and cannot fully capture these broadband signals containing wear information. High-bandwidth acoustic emission sensors installed on multiple rotating shafts can receive signals simultaneously. The example above uses three rotating shafts, with two high-bandwidth acoustic emission sensors on each shaft, for a total of six, covering a bandwidth of 10kHz-1MHz. Through the high bandwidth characteristics of the sensors, the rotation detection signal data of different rotating shafts under different operating conditions can be accurately identified. This data contains the original information of the wear state of the rotating shafts.

[0024] Next, based on the identified rotation detection signal data, key information is further extracted. First, the start time of the acoustic emission pulse is determined, that is, the specific time when the acoustic emission signal arrives at each high-bandwidth acoustic emission sensor. For example, the first high-bandwidth acoustic emission sensor receives the signal at 0.002s, and the second high-bandwidth acoustic emission sensor receives the same signal at 0.003s. This time difference provides the basis for subsequent positioning.

[0025] Simultaneously, acoustic emission pulse features are extracted, including peak amplitude, duration, spectral center frequency, energy (calculated by integrating the square of the signal amplitude), and envelope shape (the signal profile obtained after filtering). These features reflect the wear state of the rotating shaft from different dimensions.

[0026] Furthermore, peak amplitude is the maximum amplitude value appearing in the acoustic emission signal, reflecting the signal strength. During extraction, time-domain analysis is performed on the rotation detection signal data acquired by the high-bandwidth acoustic emission sensor, and the maximum amplitude value is identified and extracted from the signal waveform; this is the peak amplitude.

[0027] Duration refers to the time interval from the start to the end of an acoustic emission pulse. During extraction, the start time of the acoustic emission pulse is first determined, i.e., the moment the signal first exceeds a preset threshold, and then the end time of the pulse is determined, i.e., the moment the signal last falls below that threshold. The time difference between the two is the duration. Specifically, the method for determining the start time in duration extraction is based on the moment the signal first exceeds the preset threshold. This is described from the perspective of actual detection operation: when a high-bandwidth acoustic emission sensor receives a signal, it needs to distinguish between valid signals and background noise using a preset threshold. When the signal first exceeds this threshold, it is determined to have arrived at the high-bandwidth acoustic emission sensor, consistent with the aforementioned logic that the start time of the acoustic emission pulse is the arrival time of the acoustic emission signal.

[0028] The center frequency of the spectrum is the central frequency point where energy is concentrated in the spectrum of an acoustic emission signal, reflecting the frequency characteristics of the signal. During extraction, a Fourier transform is performed on the rotation detection signal data to convert the time-domain signal into a frequency-domain signal, obtaining the power spectral density of the signal. Then, the centroid of the spectrum is calculated based on the power spectral density, and the frequency corresponding to this centroid is the center frequency of the spectrum.

[0029] Energy is the total energy contained in an acoustic emission signal. To extract it, the amplitude of the acoustic emission signal is squared, and the result is integrated over the pulse duration; the integral result is the energy.

[0030] The envelope shape is the overall contour of the acoustic emission signal after processing, reflecting the signal's changing trend. During extraction, the rotation detection signal data is filtered to remove noise interference, and then the envelope of the acoustic emission signal is extracted using methods such as Hilbert transform or extreme point connection. The shape of this envelope is the envelope shape.

[0031] The aforementioned acoustic emission pulse characteristics comprehensively characterize the acoustic emission pulse from dimensions such as signal strength, time span, frequency characteristics, energy magnitude, and overall trend, providing comprehensive feature support for subsequent sound source localization and latent wear identification.

[0032] By using a high-bandwidth acoustic emission sensor to identify signals and extract multi-dimensional features, comprehensive and accurate raw data support is provided for subsequent sound source localization and wear identification.

[0033] Step A300: Construct a spatial localization model for the sound source. The spatial localization model for the sound source is used to locate the sound source by analyzing the start time and characteristics of the acoustic emission pulse. The result of the sound source localization is obtained.

[0034] In one embodiment of this application, the sound source location and weight matrix are first initialized. The theoretical TDOA arrival time difference sample is obtained by combining the sensor calibration results and the propagation velocity model. After comparing and constructing the error vector, the weight matrix is ​​obtained by using the acoustic emission pulse feature sample. Then, the Jacobian matrix is ​​calculated and updated to the accuracy standard by iterative weighted least squares method to obtain the sound source spatial localization model. The specific steps are described in detail in A310-A340.

[0035] After constructing the spatial localization model of the sound source, when the rotating shaft of the cutting tooth generates a new acoustic emission signal, the first step is to obtain the acoustic emission pulse start time and acoustic emission pulse characteristics of the signal using a high-bandwidth acoustic emission sensor. For example, if the first high-bandwidth acoustic emission sensor receives the signal at 0.0021s and the second high-bandwidth acoustic emission sensor receives the same signal at 0.0023s, the pulse start time can be obtained. Simultaneously, acoustic emission pulse characteristics such as a peak amplitude of 150mV, a duration of 0.5ms, and a spectral center frequency of 300kHz are extracted. These data provide the raw input for localization.

[0036] The acoustic emission pulse start time and acoustic emission pulse characteristics are input into the constructed sound source spatial localization model. The model first calculates the TDOA between different sensors based on the pulse start time. For example, the time difference between the two sensors in the above example is 0.0002s. Combining the three-dimensional coordinates of each high-bandwidth acoustic emission sensor and the corresponding propagation velocity model, the spatial location range of the sound source is initially calculated.

[0037] Meanwhile, the sound source spatial localization model optimizes the localization process using acoustic emission pulse characteristics. For example, signals with higher peak amplitude (150mV) have higher reliability and are given greater weight by the model; signals with a spectral center frequency matching wear signal characteristics (around 300kHz) are prioritized for localization calculations to reduce interference from noise signals. Through iterative calculations within the sound source spatial localization model, the position estimate is continuously corrected, ultimately outputting accurate three-dimensional coordinates as the sound source localization result.

[0038] By inputting the acoustic emission pulse start time and features into the sound source spatial localization model, and through TDOA arrival time difference sample calculation, feature optimization, and iterative correction, the precise location of the sound source is achieved, providing accurate spatial coordinates for subsequent latent wear identification.

[0039] Step A400: Map the sound source localization result to the three-dimensional structural model of the rotating shaft to identify latent wear and output the latent wear risk level.

[0040] In this embodiment, latent wear refers to wear that is not easily detected by surface observation and is generated during operation of components such as the rotating shaft of the cutting gear.

[0041] Specifically, the sound source localization results are mapped onto the three-dimensional structural model of the rotating shaft to obtain spatially mapped localization points. After DBSCAN clustering, the activity level of each region is calculated by statistically analyzing the indicators. Regions exceeding the preset activity level are identified as latent wear regions, and the activity level is mapped to the latent wear risk level. The specific steps are explained in detail in A440-A470. Among them, the construction process of the three-dimensional structural model of the rotating shaft is explained in detail in steps A410-A430.

[0042] Furthermore, step A300 in the method provided in this application embodiment includes:

[0043] A310: Obtain the sensor spatial calibration results based on the spatial three-dimensional coordinates corresponding to the plurality of high-bandwidth acoustic emission sensors.

[0044] A320: Obtain the material properties and environmental parameters of the multiple rotating shafts, and establish a propagation velocity model under the material properties and environmental parameters.

[0045] A330: Collect rotation detection signal samples of historical acoustic emission events, and analyze the acoustic emission pulse start time samples of the rotation detection signal samples to obtain TDOA arrival time difference samples.

[0046] A340: Combine the sensor spatial calibration results, TDOA arrival time difference samples, acoustic emission pulse feature samples, and propagation velocity model to calculate the sound source localization result samples of the historical acoustic emission events until the accuracy of the sound source localization result samples reaches a preset threshold, thereby obtaining the sound source spatial localization model.

[0047] In this embodiment, TDOA is the time difference of arrival, which refers to the time difference between the arrival of the acoustic emission signal at different high-bandwidth acoustic emission sensors. It can be obtained by analyzing the acoustic emission pulse start time samples of historical acoustic emission events.

[0048] Specifically, when constructing a spatial localization model for a sound source, the spatial positions of multiple high-bandwidth acoustic emission sensors must first be determined. Those skilled in the art measure the specific coordinates of each sensor in a three-dimensional coordinate system. For example, the aforementioned six sensors are located at (x1, y1, z1), (x2, y2, z2), up to (x6, y6, z6), respectively. Then, a spatial calibration algorithm is used to correct errors in these coordinates, ultimately obtaining the sensor spatial calibration results. This determines the precise relative position of each sensor in physical space, providing a spatial reference for subsequent sound source localization.

[0049] Next, the velocity characteristics of the acoustic emission signal during propagation need to be considered. The material properties of the rotating shaft (such as the elastic modulus and density when using alloy steel) and environmental parameters (such as the temperature range of 25-35℃ and humidity conditions of 60-80% commonly encountered in mining operations) directly affect the speed of sound. For example, the speed of sound propagation in alloy steel at 25℃ is approximately 5000 m / s, and the speed of sound may change by about 10 m / s for every 10℃ increase in temperature. Combining these data, by establishing the correlation equation between material properties and environmental parameters, a propagation speed model for this scenario can be constructed, specifically:

[0050] First, based on the material properties of the rotating shaft, a baseline value for the sound velocity is determined according to the fundamental principles of material acoustic properties. This baseline value is determined by the material's own physical properties (such as elastic modulus and density). Then, the influence of environmental parameters (such as temperature and humidity) on the sound velocity is analyzed, clarifying the correction trend of the sound velocity when different environmental parameters change. Next, these influence patterns are transformed into quantified correction coefficients, establishing a correlation equation between the baseline value and the correction coefficients of each environmental parameter. This equation reflects the relationship between the sound velocity change under the combined influence of material properties and environmental parameters. Finally, by inputting specific material properties and environmental parameters through this equation, the sound propagation velocity under corresponding conditions can be obtained, thereby constructing a propagation velocity model adapted to this scenario.

[0051] Simultaneously, it is necessary to collect rotation detection signal samples from historical acoustic emission events, such as selecting 1000 sets of historical data containing different degrees of wear. The acoustic emission pulse start time samples in each set are analyzed. For example, the reception time tA of high-bandwidth acoustic emission sensor A and the reception time tB of high-bandwidth acoustic emission sensor B in the same event are extracted, and the difference between them Δt=|tA-tB| is calculated, thus obtaining a large number of TDOA arrival time difference samples. These samples reflect the time difference patterns of acoustic signals arriving at different sensors in historical events, providing crucial data support for training the sound source spatial localization model.

[0052] Finally, the initial estimated position and initial weight matrix of the sound source are initialized. The theoretical TDOA arrival time difference sample is obtained by combining the sensor spatial calibration results and the propagation velocity model. After comparison, the error vector is constructed. The weight matrix is ​​obtained by using acoustic emission pulse feature samples. Then, the Jacobian matrix is ​​calculated and iteratively updated using the iterative weighted least squares method until the accuracy reaches the standard, thus obtaining the sound source spatial localization model. The specific steps are explained in detail in A341-A345.

[0053] By determining the spatial location of the sensor, establishing a sound velocity model, and obtaining historical time difference samples, multi-dimensional basic data is provided for the construction of a sound source spatial localization model, ensuring that the model can perform accurate calculations based on accurate spatial benchmarks, propagation speeds, and historical patterns.

[0054] Furthermore, step A340 in the method provided in this application embodiment includes:

[0055] A341: Initialize and set the initial estimated position and initial weight matrix of the sound source.

[0056] A342: Based on the sensor spatial calibration results and propagation velocity model, obtain the theoretical TDOA arrival time difference sample.

[0057] A343: Compare the theoretical TDOA arrival time difference sample with the TDOA arrival time difference sample to construct an error vector.

[0058] A344: Obtain the weight matrix of the error vector using the acoustic emission pulse feature samples.

[0059] A345: Calculate the Jacobian matrix based on the error vector and the weight matrix, and update the Jacobian matrix using the iterative weighted least squares method until the accuracy of the sound source localization result sample reaches a preset threshold, thereby obtaining the sound source spatial localization model.

[0060] In this embodiment, the initial estimated location of the sound source is a possible location initialized based on the characteristics of the rotating shaft structure when calculating the sound source localization results sample of historical acoustic emission events. The error vector is a vector constructed by comparing the theoretical TDOA arrival time difference sample with the actually acquired TDOA arrival time difference sample, and arranging the differences of each corresponding TDOA in order. The Jacobian matrix is ​​a matrix calculated based on the error vector and the weight matrix, used to quantify the relationship between the error vector and the sound source location.

[0061] Optionally, when calculating the sound source localization results samples of historical acoustic emission events, initialization settings are first performed. Taking into account the structural characteristics of the rotating shaft, key parts prone to wear, such as the bushing contact area, are used as the initial estimated location of the sound source. At the same time, based on the historical signal stability of each sensor, an initial weight matrix is ​​set. For example, sensors with less signal interference are assigned a weight of 0.7, and sensors susceptible to vibration are assigned a weight of 0.3, providing basic parameters for subsequent calculations.

[0062] Next, based on the sensor spatial calibration results (i.e., the precise three-dimensional coordinates of each sensor) and the propagation velocity model (reflecting the relationship between sound speed and materials and the environment) obtained in steps A310 and A320, the theoretical TDOA arrival time difference sample can be further derived. Specifically, based on the spatial distance between the initial estimated position of the sound source and each high-bandwidth acoustic emission sensor, the theoretical time for the signal to reach each sensor is calculated using the propagation velocity model. Then, the theoretical TDOA arrival time difference sample is obtained through the theoretical time difference between each pair of sensors. For example, if the distance from the initial position to the first high-bandwidth acoustic emission sensor is 0.5 meters, the corresponding theoretical propagation time is 0.0001 seconds; the distance to the second high-bandwidth acoustic emission sensor is 0.6 meters, the theoretical propagation time is 0.00012 seconds, and the theoretical TDOA for both is 0.00002 seconds.

[0063] Subsequently, the theoretical TDOA arrival time difference samples are compared with the actual collected TDOA arrival time difference samples (i.e., the actual time differences measured in historical events). By calculating the difference between the corresponding TDOA for each group, these differences are arranged in order to construct an error vector. For example, if a theoretical TDOA is 0.00002 seconds and the actual TDOA is 0.000025 seconds, the difference of 0.000005 seconds is used as an element in the error vector, intuitively reflecting the deviation between theory and reality.

[0064] Then, the error impact value of the acoustic emission pulse feature sample is obtained, and the weight matrix of the error vector is output after mapping using the sigmoid function. The specific steps are explained in detail in A344-1.

[0065] Next, the Jacobian matrix is ​​calculated, quantifying the impact of changes in the sound source position on the error based on the error vector and weight matrix. The error vector reflects the deviation between the theoretical and actual TDOA (Time Difference of Arrival) samples, while the weight matrix reflects the reliability of different signals. Combining the two, the rate of change of the error vector for each position parameter, i.e., small changes in x, y, and z in three-dimensional coordinates, can be derived, forming the Jacobian matrix. For example, when the x-coordinate of the sound source changes by 0.1 mm, a certain element in the error vector changes by 0.00001 seconds. This rate of change is used as the corresponding element in the Jacobian matrix to depict the correlation between position and error.

[0066] Finally, the Jacobian matrix is ​​updated iteratively using the iterative weighted least squares method. First, based on the current Jacobian matrix and weight matrix, the correction amount for the sound source location is calculated, and the initial estimated location is adjusted. Then, based on the new location, the theoretical TDOA arrival time difference sample, error vector, and Jacobian matrix are recalculated, and the correction process is repeated. A preset accuracy threshold of 95% is set. After each iteration, the deviation between the positioning result and the actual location is calculated. If the accuracy reaches 95% or higher after a certain iteration, the iteration stops.

[0067] By calculating the Jacobian matrix to quantify the correlation between position and error, and combining iterative weighted least squares method to continuously optimize the sound source position estimation until the preset accuracy is achieved, a high-precision sound source spatial localization model is finally obtained.

[0068] Furthermore, step A344 in the method provided in this application embodiment includes:

[0069] A344-1: Obtain the error impact value of the acoustic emission pulse feature sample, perform sigmoid function mapping on the error impact value, and output the weight matrix.

[0070] In this embodiment, the sigmoid function is a sigma function used to map the error influence value of acoustic emission pulse feature samples, and then output a weight matrix.

[0071] Specifically, in acoustic emission pulse feature samples, features such as peak amplitude, duration, and spectral center frequency can reflect the degree to which the signal is affected by noise, mechanical interference, or attenuation. For example, signals with peak amplitudes below 100mV may be distorted due to noise interference, and signals with durations exceeding the normal range of 0.1-1ms may be affected by mechanical vibration. Samples with these abnormal features have higher error impact values. By analyzing these feature samples, the error impact value of each sample can be quantified. The error impact value of severely distorted signals is set at 0.8, and that of slightly interfered signals at 0.3, thus reflecting the degree of influence of different signals on the error vector.

[0072] Because the numerical range of error impact values ​​can be dispersed, directly using them for weight calculation can lead to unstable adjustment effects. Mapping these error impact values ​​using the sigmoid function transforms them into continuous values ​​between 0 and 1. For example, an error impact value of 0.8 outputs 0.2 after processing with the sigmoid function, indicating low signal reliability and a smaller weight; an error impact value of 0.3 outputs 0.6, indicating high signal reliability and a larger weight. These mapped values ​​are arranged according to the sensor or sample sequence to form a weight matrix, where each element corresponds to the reliability weight of a different signal.

[0073] By combining the influence of acoustic emission pulse characteristics to quantify errors and mapping them using the sigmoid function, a weight matrix that reflects the reliability of the signal is output, effectively reducing the interference of low-quality signals on positioning and improving the reliability of the error vector.

[0074] Furthermore, step A400 in the method provided in this application embodiment includes:

[0075] A410: Obtain 3D modeling data of the cutting tooth structure, including the geometric dimensions and positional relationships of the tool holder, nesting parts, pin holes, cutting tooth tool holder, and cutting tooth tip.

[0076] A420: Obtain the three-dimensional working model of the cutting tooth from the three-dimensional modeling data of the cutting tooth structure.

[0077] A430: Divide the hidden wear monitoring area of ​​the three-dimensional working model of the cutting tooth, and extract the hidden wear monitoring area as the three-dimensional structural model of the rotating shaft.

[0078] Specifically, when constructing the 3D structural model of the rotating shaft, the first step is to comprehensively acquire the 3D modeling data of the cutting tooth structure. This involves scanning each component of the cutting tooth using a 3D scanning device, or extracting key parameters from design drawings to obtain the geometric dimensions of the tool holder, nesting parts, pin holes, cutting tooth tool holder, and cutting tooth tip. Simultaneously, the positional relationships between each component are recorded, such as the vertical distance between the tool holder and the cutting tooth tool holder, and the relative position of the pin hole within the tool holder, providing a complete data foundation for subsequent modeling.

[0079] Next, based on the acquired 3D modeling data of the cutting tooth structure, the model was constructed using the 3D modeling software SolidWorks. The geometric dimensions and positional relationships of each component were input into the 3D modeling software. Through Boolean operations, constraint associations, and other operations, the tool holder, nesting parts, pin holes, cutting tooth tool holder, and cutting tooth head were assembled according to the actual working state, forming a 3D working model of the cutting tooth that includes the overall structure of the cutting tooth. This model can accurately reflect the spatial form of the cutting tooth during operation, such as the connection method between the cutting tooth head and the rotating shaft, and the relative range of motion of each component.

[0080] Based on the 3D working model of the cutting gear, latent wear monitoring areas are defined according to the wear patterns of the rotating shaft. Typically, areas prone to latent wear, such as the contact area between the rotating shaft and the cutting gear tool holder, and areas of concentrated stress on the shaft (e.g., shaft sections 20-30mm from the end), are selected as latent wear monitoring areas. These latent wear monitoring areas are extracted from the overall 3D working model of the cutting gear using the region division function of 3D modeling software, forming an independent 3D structural model of the rotating shaft. This model only contains the areas requiring focused monitoring, reducing the complexity of subsequent data processing.

[0081] By acquiring comprehensive modeling data, constructing an overall three-dimensional operational model, and extracting key monitoring areas, a three-dimensional structural model of the rotating shaft that accurately reflects the wear-prone parts of the rotating shaft is constructed, providing a precise spatial reference for mapping sound source localization results and identifying latent wear.

[0082] Furthermore, step A400 in the method provided in this application embodiment includes:

[0083] A440: Map the sound source localization result to the three-dimensional structural model of the rotating shaft and output the spatial mapping localization point.

[0084] A450: Performs DBSCAN spatial clustering on continuously acquired spatially mapped positioning points and outputs multiple clustering regions.

[0085] A460: Calculate the location point density index, location point cumulative energy value, and location point frequency for each cluster region in the multiple cluster regions, and calculate the multiple activity levels corresponding to the multiple cluster regions.

[0086] A470: Identify clustered regions with activity levels greater than a preset threshold as latent wear regions, and map the activity level to output a latent wear risk level.

[0087] Specifically, when mapping the sound source localization results to the three-dimensional structural model of the rotating axis, the spatial coordinate systems of the two are first unified. The three-dimensional coordinates obtained from the sound source localization are converted into the local coordinates of the three-dimensional structural model of the rotating axis, so that each localization result accurately corresponds to the specific position in the model. Finally, a series of spatially mapped localization points are output, which intuitively reflect the spatial distribution of the acoustic emission signal source on the rotating axis.

[0088] Next, when performing DBSCAN spatial clustering (density-based spatial clustering for noisy applications) on the continuously acquired spatially mapped positioning points, reasonable clustering parameters were set: with a neighborhood radius of 3 mm, a region containing at least 8 positioning points was divided into a clustering unit. This clustering method can aggregate scattered positioning points into multiple clustering regions, such as obtaining three main clustering regions: the middle of the shaft, the vicinity of the shaft end, etc., each region representing a potential wear location where acoustic emission signals are concentrated.

[0089] Then, the indicators for each cluster region are calculated. The location point density indicator is the ratio of the number of location points in the region to the region's volume. For example, if a cluster region contains 120 points and has a volume of 200 mm³, the density is 120 / 200 = 0.6 points / mm³. The cumulative location point energy is the sum of the acoustic emission energy of all points in the region, with an example cumulative value of 5000 mV·s. The location point frequency is the number of location points in the region per unit time, assuming 8 per minute. The activity level is calculated by weighting these three indicators in a 4:3:3 ratio. For example, the activity level of the above region is 0.6 × 0.4 + (5000 / 10000) × 0.3 + (8 / 10) × 0.3 = 0.57, where 10000 and 10 are normalization bases set to standardize the cumulative location point energy and location point frequency to the 0-1 range.

[0090] Finally, a preset activity threshold of 0.4 is set, and clusters with activity levels greater than this value are identified as latent wear areas. The activity values ​​are then mapped to risk level classification standards: 0.4-0.6 corresponds to low risk (risk level 1); 0.6-0.8 corresponds to medium risk (risk level 2); and above 0.8 corresponds to high risk (risk level 3). For example, the region with an activity level of 0.57 is output as low latent wear risk level 1.

[0091] Through spatial mapping, cluster analysis, index calculation, and risk classification, the system accurately identifies the hidden wear areas of the rotating shaft and assesses the risk level, providing a reliable basis for the detection of cutter wear.

[0092] Furthermore, step A400 in the method provided in this application embodiment further includes step A480, which further includes:

[0093] A481: Collect acoustic emission pulse feature samples and known wear type labels from historical acoustic emission events.

[0094] A482: Train a latent wear classification model based on the acoustic emission pulse feature samples and known wear type labels.

[0095] A483: Input the acoustic emission pulse features into the latent wear classification model for identification, and output multiple latent wear type labels corresponding to the multiple clustering regions.

[0096] A484: Calculate the activity level by weighting the multiple latent wear type labels and output the latent wear risk level.

[0097] In one embodiment, firstly, acoustic emission pulse feature samples and known wear type labels from historical acoustic emission events are collected. Event data covering different wear states are selected, for example, 500 sets of acoustic emission pulse feature samples containing peak amplitude, duration, and spectral center frequency are collected. At the same time, each set of acoustic emission pulse feature samples is labeled with a clear known wear type label, such as abrasive wear, fatigue wear, adhesive wear, etc., with each type containing at least 100 sets of samples to ensure the diversity of samples and the accuracy of labels, providing sufficient basic data for training the latent wear classification model.

[0098] Then, when training the latent wear classification model based on the collected acoustic emission pulse feature samples and known wear type labels, the samples were first divided into training and test sets in an 8:2 ratio. A random forest algorithm was used to construct the model, with acoustic emission pulse features such as peak amplitude, duration, and spectral center frequency as input variables, and known wear type labels as output variables. The model parameters were iteratively optimized using the training set, for example, by adjusting the number of decision trees to 50 and setting the maximum depth to 10. The model performance was verified using the test set. Training was stopped when the classification accuracy reached above 95%, resulting in a stable latent wear classification model.

[0099] Next, when the newly acquired acoustic emission pulse features are input into the trained latent wear classification model for identification, the model will match the features with the correlation patterns between various wear types. For example, if the acoustic emission pulse features of a certain cluster region are a peak amplitude of 150mV, a duration of 0.8ms, and a spectral center frequency of 300kHz, the model will determine the wear type corresponding to this region as abrasive wear by calculating the feature similarity and output the corresponding latent wear type label, thus achieving accurate identification of the wear type for each cluster region.

[0100] Finally, a wear type weight mapping table is constructed to obtain the weights corresponding to multiple implicit wear type labels. Then, the activity of multiple cluster regions is comprehensively weighted according to these weights to output the implicit wear risk level. The specific steps are explained in detail in A484-1-A484-2.

[0101] By collecting samples, scientifically training models, and accurately identifying features, we have achieved effective classification of latent wear types in multiple clustered regions, laying the foundation for subsequent weight calculation based on wear types and outputting more reasonable latent wear risk levels.

[0102] Furthermore, step A484 in the method provided in this application embodiment includes:

[0103] A484-1: Construct a wear type weight mapping table, and obtain multiple weights corresponding to the multiple implicit wear type labels based on the wear type weight mapping table.

[0104] A484-2: Calculate the comprehensive weight of the activity levels corresponding to the multiple clustering regions according to the multiple weights, and output the implicit wear risk level.

[0105] Optionally, when constructing the wear type weight mapping table, weight values ​​are set according to the degree of impact of different latent wear types on equipment operation. Combining historical fault data and the experience of experts in this field, common wear types (such as abrasive wear, fatigue wear, and adhesive wear) are sorted according to their development speed and severity, forming a wear type weight mapping table containing wear types and corresponding weights, as shown in Table 1, providing a standard basis for subsequent weight acquisition.

[0106] Table 1: Wear Type Weight Mapping Table

[0107]

[0108] Next, when obtaining the weights corresponding to multiple implicit wear type labels based on the wear type weight mapping table, it is necessary to match the wear type label of each cluster region with the mapping table. For example, if a certain cluster region is identified as abrasive wear, its corresponding weight of 0.8 is extracted from the mapping table; if another region is fatigue wear, its weight of 0.6 is extracted, ensuring that the wear type of each region can correspond to an accurate weight value, providing basic parameters for comprehensive calculation.

[0109] Finally, when calculating the overall weight of the activity levels of multiple cluster regions according to multiple weights, a weighted summation method is used. Assuming the activity levels of the three cluster regions are 0.7, 0.5, and 0.4 respectively, with corresponding weights of 0.8, 0.6, and 0.4, the overall value is calculated as 0.7×0.8+0.5×0.6+0.4×0.4=0.56+0.3+0.16=1.02. This overall value is then mapped to the risk level classification standard preset in step A470. An activity level greater than 0.8 corresponds to high risk, and the implicit wear and tear risk in this scenario is output as high risk, level 3.

[0110] By constructing a wear type weight mapping table, matching weights, and performing comprehensive calculations, differentiated consideration of the impact of different wear types is achieved, making the output implicit wear risk level more consistent with the actual wear hazard level and improving the accuracy of the assessment.

[0111] Furthermore, step A400 in the method provided in this application embodiment further includes step A490, which further includes:

[0112] A491: Obtain the surface wear detection data of the cutting tooth head, identify visible wear according to the surface wear detection data, and output the visible wear risk level.

[0113] A492: Obtain the comprehensive wear risk level based on the implicit wear risk level and the explicit wear risk level.

[0114] In one embodiment, when acquiring surface wear detection data of the cutting tooth tip, laser scanning or high-definition image recognition technology can be used to collect data on the changes in geometric parameters of the tip surface, such as the area of ​​the wear region, wear depth, and degree of edge dulling. These data directly reflect the visible wear state of the cutting tooth tip and provide a quantitative basis for subsequent identification.

[0115] Next, when identifying visible wear based on surface wear detection data, it is necessary to establish visible wear evaluation criteria, as shown in Table 2. The collected detection data is then compared with the criteria to output the corresponding visible wear risk level.

[0116] Table 2: Evaluation Criteria for Visible Wear

[0117]

[0118] Then, when obtaining the comprehensive wear risk level based on the latent wear risk level and the visible wear risk level, it is necessary to set the weight allocation for the two. Considering that visible wear directly affects operational efficiency, it can be assigned a weight of 60%, while latent wear, due to its potential hazards, can be assigned a weight of 40%. For example, if the latent wear risk level is 2 (medium) and the visible wear risk level is 2 (medium), the comprehensive risk level is calculated as 2×0.4+2×0.6=2, outputting a medium comprehensive wear risk level; if the latent wear risk level is 3 (high) and the visible wear risk level is 2 (medium), then the comprehensive level is 3×0.4+2×0.6=2.4, which, after rounding, outputs a slightly above-medium comprehensive risk level.

[0119] By acquiring visible wear data and identifying its levels, and combining this with a weighted average of latent wear risk levels, a comprehensive assessment of the wear status of the cutting teeth is achieved. This makes the output comprehensive risk level more reflective of the actual wear hazards, providing a more accurate basis for equipment maintenance.

[0120] In summary, the method for detecting wear of cutting teeth in a tunneling machine provided in this application has the following technical effects:

[0121] This application acquires rotation detection signal data by setting multiple high-bandwidth acoustic emission sensors on the rotating shaft of the cutting tooth. After processing such as extracting the acoustic emission pulse start time and characteristics and constructing a sound source spatial localization model, the sound source localization result is obtained. The activity level and risk level of the latent wear area are calculated, and comprehensive weight calculation is performed in combination with the visible wear detection data. This allows for accurate identification of the latent and visible wear of the cutting tooth of the tunneling machine, making the cutting tooth wear detection results more accurate and reliable. It achieves the technical effect of accurately identifying the latent wear of the rotating shaft of the tunneling machine cutting tooth and assessing the risk level, thus improving the accuracy and comprehensiveness of the cutting tooth wear detection.

[0122] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a cutter tooth wear detection system for a tunneling machine, the system comprising:

[0123] Acoustic emission sensor deployment module 1 is used to acquire a cutting tool holder and a cutting tool head. The cutting tool head is fixed to the cutting tool holder through multiple rotating shafts, and multiple high-bandwidth acoustic emission sensors are set on the multiple rotating shafts.

[0124] Acoustic emission pulse acquisition module 2 uses the multiple high-bandwidth acoustic emission sensors to identify multiple rotation detection signal data of the multiple rotating shafts, and extracts the acoustic emission pulse start time and acoustic emission pulse characteristics of the multiple rotation detection signal data. The acoustic emission pulse start time is the arrival time of the acoustic emission signal, and the acoustic emission pulse characteristics include peak amplitude, duration, spectral center frequency, energy, and envelope shape.

[0125] The sound source localization result acquisition module 3 is used to construct a sound source spatial localization model. The sound source spatial localization model obtains the sound source localization result by locating the sound source based on the start time and characteristics of the acoustic emission pulse.

[0126] The latent wear risk level acquisition module 4 is used to map the sound source localization result to the three-dimensional structural model of the rotating shaft for latent wear identification and output the latent wear risk level.

[0127] Furthermore, the sound source localization result acquisition module 3 is used to perform the following steps:

[0128] The spatial calibration results of the sensors are obtained based on the spatial three-dimensional coordinates corresponding to the multiple high-bandwidth acoustic emission sensors; the material properties and environmental parameters of the multiple rotating axes are obtained, and a propagation velocity model under the material properties and environmental parameters is established; rotation detection signal samples of historical acoustic emission events are collected, and the acoustic emission pulse start time samples of the rotation detection signal samples are analyzed to obtain TDOA arrival time difference samples; the sound source localization result samples of the historical acoustic emission events are calculated by combining the sensor spatial calibration results, TDOA arrival time difference samples, acoustic emission pulse feature samples and propagation velocity model until the accuracy of the sound source localization result samples reaches a preset threshold, and a sound source spatial localization model is obtained.

[0129] Furthermore, the sound source localization result acquisition module 3 is used to perform the following steps:

[0130] Initialize and set the initial estimated location and initial weight matrix of the sound source; obtain theoretical TDOA arrival time difference samples based on the sensor spatial calibration results and propagation velocity model; construct an error vector by comparing the theoretical TDOA arrival time difference samples with the actual TDOA arrival time difference samples; obtain the weight matrix of the error vector using the acoustic emission pulse feature samples; calculate the Jacobian matrix based on the error vector and the weight matrix, and update the Jacobian matrix using the iterative weighted least squares method until the accuracy of the sound source localization result samples reaches a preset threshold, thereby obtaining the sound source spatial localization model.

[0131] Furthermore, the sound source localization result acquisition module 3 is used to perform the following steps:

[0132] Obtain the error impact value of the acoustic emission pulse feature sample, perform sigmoid function mapping on the error impact value, and output the weight matrix.

[0133] Furthermore, the latent wear risk level acquisition module 4 is used to perform the following steps:

[0134] Obtain three-dimensional modeling data of the cutting tooth structure, including the geometric dimensions and positional relationships of the tool holder, nesting parts, pin holes, cutting tooth tool seat, and cutting tooth head; obtain a three-dimensional working model of the cutting tooth from the three-dimensional modeling data of the cutting tooth structure; divide the hidden wear monitoring area of ​​the three-dimensional working model of the cutting tooth, and extract the hidden wear monitoring area as the three-dimensional structural model of the rotating shaft.

[0135] Furthermore, the latent wear risk level acquisition module 4 is used to perform the following steps:

[0136] The sound source localization results are mapped to the three-dimensional structural model of the rotating shaft to output spatially mapped localization points. DBSCAN spatial clustering is performed on the continuously acquired spatially mapped localization points to output multiple cluster regions. The localization point density index, localization point energy cumulative value, and localization point frequency corresponding to each of the multiple cluster regions are statistically analyzed, and multiple activity levels corresponding to the multiple cluster regions are calculated. Cluster regions with activity levels greater than a preset value are identified as latent wear regions, and the activity levels are mapped and output as latent wear risk levels.

[0137] Furthermore, the latent wear risk level acquisition module 4 is used to perform the following steps:

[0138] Acoustic emission pulse feature samples and known wear type labels from historical acoustic emission events are collected; a latent wear classification model is trained based on the acoustic emission pulse feature samples and known wear type labels; the acoustic emission pulse features are input into the latent wear classification model for identification, and multiple latent wear type labels corresponding to the multiple clustering regions are output; the activity level is weighted according to the multiple latent wear type labels, and the latent wear risk level is output.

[0139] Furthermore, the latent wear risk level acquisition module 4 is used to perform the following steps:

[0140] Construct a wear type weight mapping table, and obtain multiple weights corresponding to the multiple implicit wear type labels based on the wear type weight mapping table; calculate the comprehensive weight of multiple activity levels corresponding to the multiple clustering regions according to the multiple weights, and output the implicit wear risk level.

[0141] Furthermore, the latent wear risk level acquisition module 4 is used to perform the following steps:

[0142] Obtain surface wear detection data of the cutting tooth head, identify visible wear according to the surface wear detection data, and output the visible wear risk level; obtain the comprehensive wear risk level based on the hidden wear risk level and the visible wear risk level.

[0143] The cutter wear detection system for tunneling machines provided in this embodiment of the invention can execute the cutter wear detection method for tunneling machines provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0144] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0145] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A pick wear detection method for a heading machine, characterized by, The method comprises: obtaining a cutting tooth tool seat and a cutting tooth cutter head, the cutting tooth cutter head being fixed to the cutting tooth tool seat through a plurality of rotating shafts, a plurality of high-bandwidth acoustic emission sensors being arranged on the plurality of rotating shafts; identifying a plurality of rotating detection signal data of the plurality of rotating shafts by using the plurality of high-bandwidth acoustic emission sensors, extracting acoustic emission pulse starting time and acoustic emission pulse characteristics of the plurality of rotating detection signal data, the acoustic emission pulse starting time being an arrival time of an acoustic emission signal, and the acoustic emission pulse characteristics including peak amplitude, duration, spectral center frequency, energy and envelope shape; constructing a sound source spatial positioning model, the sound source spatial positioning model being obtained by sound source positioning on the acoustic emission pulse starting time and the acoustic emission pulse characteristics; mapping the sound source positioning result to a rotating shaft three-dimensional structure model for implicit wear identification, and outputting an implicit wear risk level; The method for constructing the sound source spatial positioning model comprises: obtaining a sensor space calibration result according to a space three-dimensional coordinate corresponding to the plurality of high-bandwidth acoustic emission sensors; obtaining material properties and environmental parameters of the plurality of rotating shafts, and establishing a propagation speed model under the material properties and the environmental parameters; collecting rotating detection signal samples of historical acoustic emission events, and analyzing acoustic emission pulse starting time samples of the rotating detection signal samples to obtain TDOA arrival time difference samples; combining the sensor space calibration result, the TDOA arrival time difference samples, acoustic emission pulse characteristic samples and the propagation speed model to calculate sound source positioning result samples of the historical acoustic emission events, until an accuracy of the sound source positioning result samples reaches a preset threshold, to obtain a sound source spatial positioning model; The method for combining the sensor space calibration result, the TDOA arrival time difference samples, the acoustic emission pulse characteristic samples and the propagation speed model to calculate the sound source positioning result samples of the historical acoustic emission events comprises: initializing an estimated sound source initial position and an initial weight matrix; obtaining theoretical TDOA arrival time difference samples according to the sensor space calibration result and the propagation speed model; comparing the theoretical TDOA arrival time difference samples and the TDOA arrival time difference samples to construct an error vector; obtaining a weight matrix of the error vector by using the acoustic emission pulse characteristic samples; calculating a Jacobian matrix according to the error vector and the weight matrix, and updating and iterating the Jacobian matrix by using an iterative weighted least squares method, until the accuracy of the sound source positioning result samples reaches the preset threshold, to obtain the sound source spatial positioning model.

2. The method of claim 1, wherein, The method for obtaining the weight matrix of the error vector by using the acoustic emission pulse characteristic samples comprises: obtaining an error influence value of the acoustic emission pulse characteristic samples, performing sigmoid function mapping on the error influence value, and outputting a weight matrix.

3. The method of claim 1, wherein, The method for constructing a rotating shaft three-dimensional structure model comprises: obtaining cutting tooth structure three-dimensional modeling data, including geometric sizes and positional relationships of a tool holder, a nesting piece, a pin hole, a cutting tooth tool seat and a cutting tooth cutter head; obtaining a cutting tooth three-dimensional operation model from the cutting tooth structure three-dimensional modeling data; Divide the hidden wear monitoring area of the three-dimensional working model of the cutting tooth, and extract the hidden wear monitoring area as a three-dimensional structure model of the rotating shaft.

4. The method of claim 1, wherein, The method for identifying hidden wear by mapping the acoustic source positioning result to the three-dimensional structure model of the rotating shaft comprises the following steps: Mapping the acoustic source positioning result to the three-dimensional structure model of the rotating shaft, and outputting a spatial mapping positioning point; DBSCAN spatial clustering is performed on the spatial mapping positioning points obtained by continuous acquisition, and a plurality of clustering areas are outputted; The density index, energy cumulative value and frequency of the positioning points corresponding to each clustering area in the plurality of clustering areas are counted, and a plurality of activity degrees corresponding to the plurality of clustering areas are calculated; A clustering area greater than a preset activity degree is identified as a hidden wear area, and the activity degree is mapped and outputted as a hidden wear risk level.

5. The method of claim 4, wherein, The method for identifying hidden wear by mapping the acoustic source positioning result to the three-dimensional structure model of the rotating shaft further comprises the following steps: Acoustic emission pulse feature samples and known wear type labels of historical acoustic emission events are acquired; A hidden wear classification model is trained according to the acoustic emission pulse feature samples and the known wear type labels; The acoustic emission pulse feature is inputted into the hidden wear classification model for identification, and a plurality of hidden wear type labels corresponding to the plurality of clustering areas are outputted; The activity degree is weighted according to the plurality of hidden wear type labels, and a hidden wear risk level is outputted.

6. The method of claim 5, wherein, The method for weighting the activity degree according to the plurality of hidden wear type labels comprises the following steps: A wear type weight mapping table is constructed, and a plurality of weights corresponding to the plurality of hidden wear type labels are obtained based on the wear type weight mapping table; The plurality of activity degrees corresponding to the plurality of clustering areas are comprehensively weighted according to the plurality of weights, and a hidden wear risk level is outputted.

7. The method of claim 1, wherein, After the hidden wear risk level is outputted, the method further comprises the following steps: Surface wear detection data of the cutting tooth bit are acquired, dominant wear identification is performed according to the surface wear detection data, and a dominant wear risk level is outputted; A comprehensive wear risk level is obtained according to the hidden wear risk level and the dominant wear risk level.

8. A pick wear detection system for a boring machine, characterized by, A cutting tooth wear detection method for a heading machine for implementing any one of claims 1-7, the system comprising: An acoustic emission sensor arrangement module (1) is used to acquire a cutting tooth tool seat and a cutting tooth bit, the cutting tooth bit is fixed to the cutting tooth tool seat through a plurality of rotating shafts, and a plurality of high-bandwidth acoustic emission sensors are arranged on the plurality of rotating shafts; An acoustic emission pulse acquisition module (2) is used to identify a plurality of rotating detection signal data of the plurality of rotating shafts by using the plurality of high-bandwidth acoustic emission sensors, extract acoustic emission pulse start time and acoustic emission pulse features of the plurality of rotating detection signal data, the acoustic emission pulse start time is the arrival time of the acoustic emission signal, and the acoustic emission pulse features include peak amplitude, duration, spectral center frequency, energy and envelope shape; An acoustic source positioning result acquisition module (3) is used to construct an acoustic source spatial positioning model, the acoustic source spatial positioning model is obtained by performing acoustic source positioning on the acoustic emission pulse start time and the acoustic emission pulse features, and an acoustic source positioning result is obtained. The implicit wear risk level acquisition module (4) is configured to map the sound source positioning result to a three-dimensional structure model of the rotating shaft to identify implicit wear and output an implicit wear risk level.

Images