Impact force detection device for a pdc compact

Abstract

The application relates to the field of mineral resource exploration, and discloses an impact force detection device for a PDC (Polycrystalline Diamond Compact) composite sheet. The device comprises a PDC composite sheet clamping module for clamping the PDC composite sheet above a rock sample; a cutting force detection module connected with the PDC composite sheet clamping module and capable of detecting the cutting force generated in a cutting experiment; a high-frequency acoustic emission signal detection module connected with one side of the PDC composite sheet and capable of detecting the acoustic emission signal within 100 kHz-1 MHz generated in the cutting experiment; and a processor configured to acquire the cutting force and the acoustic emission signal, perform noise reduction and feature extraction on the acoustic emission signal to obtain a feature vector of the noise-reduced acoustic emission signal data, acquire an impact force model, and input the cutting force and the feature vector to calculate the impact force generated in the cutting experiment. The high-frequency acoustic emission signal detection module of the device is directly connected with the PDC composite sheet, can output the impact force data containing high-frequency information, and provides a basis for the impact resistance performance evaluation and optimization of the PDC composite sheet.

Description

Technical Field

[0001] This application belongs to the field of mineral resource exploration technology, and specifically relates to an impact force detection device for PDC composite sheets. Background Technology

[0002] With the continuous advancement of deep-earth drilling projects (such as drilling projects in the Tarim Basin that are heading towards ultra-deep wells at depths of 10,000 meters), the rock conditions faced by drilling are becoming increasingly extreme. In particular, when drilling into formations with great depth, strong compaction, high hardness, strong abrasiveness, and no clay (such as pure dense dolomite formations), PDC composite drill bits are prone to high-frequency stick-slip vibrations and instantaneous violent impacts during rock breaking. These high-frequency micro-vibrations occurring at the microsecond level can lead to early breakage, thermal cracking, or abnormal wear of the PDC composite bits, which seriously restricts the improvement of the mechanical rate of penetration (ROP) in deep formations.

[0003] Currently, the conventional method for evaluating the rock-breaking performance of PDC composite blades is to conduct cutting experiments on a vertical lathe and install a conventional three-dimensional force gauge under the tool holder to measure stress data. However, traditional macroscopic force gauges have serious physical limitations. Their natural frequency is low (usually within a few kilohertz), and the stress wave is significantly mechanically attenuated and filtered by the complex machine tool tool holder structure and fixture stiffness during transmission from the PDC cutting teeth to the force gauge. This low-frequency response characteristic completely fails to capture the high-frequency micro-vibrations and transient shock waves released by the PDC composite blades when brittlely fracturing dense hard rock. This results in the stress data measured by traditional methods being a significantly smoothed average curve, completely masking the true rock-breaking impact peak value, and failing to provide accurate data support for the impact stress analysis, failure mechanism, and life assessment of PDC composite blades. Summary of the Invention

[0004] The purpose of this application is to provide an impact force detection device for PDC composite sheets, which can detect high-frequency stress wave information and calculate and output real impact force data containing high-frequency stress wave information.

[0005] To achieve the above objectives, this application provides an impact force testing device for PDC composite sheets, the impact force testing device comprising: PDC composite sheet clamping module, used to clamp the PDC composite sheet located above the rock sample; The cutting force detection module is connected to the PDC composite sheet clamping module and is used to detect the cutting force generated when the PDC composite sheet impacts and cuts the rock sample. The high-frequency acoustic emission signal detection module is connected to the side wall on one side of the axial direction of the PDC composite sheet and is used to detect the acoustic emission signal of a preset frequency generated when the PDC composite sheet impacts and cuts the rock sample. The preset frequency range is 100kHz-1MHz. The processor is communicatively connected to the cutting force detection module and the high-frequency acoustic emission signal detection module and is configured as follows: Acquire cutting force and acoustic emission signals; The acoustic emission signal is denoised to obtain denoised emission signal data; Feature extraction processing is performed on the noise-reduced transmitted signal data to obtain a feature vector of the noise-reduced transmitted signal data; Obtain the impact force model; The cutting force and eigenvector are input into the impact force model to calculate the impact force generated when the PDC composite sheet performs impact cutting on the rock sample.

[0006] In the embodiments of this application, the feature vector includes the mean of the noise-reduced transmitted signal data, the variance of the noise-reduced transmitted signal data, the root mean square of the noise-reduced transmitted signal data, the peak value of the noise-reduced transmitted signal data, the skewness of the noise-reduced transmitted signal data, the kurtosis of the noise-reduced transmitted signal data, the waveform factor of the noise-reduced transmitted signal data, the impulse factor of the noise-reduced transmitted signal data, the dominant frequency of the noise-reduced transmitted signal data, the centroid frequency of the noise-reduced transmitted signal data, the mean square frequency of the noise-reduced transmitted signal data, and the energy proportion of a specific high-frequency band of the noise-reduced transmitted signal data.

[0007] In the embodiments of this application, the impact force model is as follows:

[0008] in, It is impact force; It is an eigenvector; It is a local feature extraction layer; It is an output layer that fuses hidden state vectors; It is the attention mechanism layer.

[0009] In the embodiments of this application, the cutting force and eigenvector are input into the impact force model to calculate the impact force generated when the PDC composite sheet performs impact cutting on the rock sample, including: The feature vector is input into the local feature extraction layer of the impact force model to obtain the local features of the noise-reduced transmitted signal data; The cutting force and local features are input into the fusion hidden state vector output layer of the impact force model to obtain the fusion hidden state vector at each time step. The fusion hidden state vector is formed by fusing local features and cutting force. The fused hidden state vectors of each time step are input into the attention mechanism layer of the impact force model to obtain the impact force of each time step when the PDC composite sheet performs impact cutting on the rock sample.

[0010] In the embodiments of this application, the impact force model is constructed in the following manner: Obtain the initial model of the impact force model; Obtain the cutting force training sample time series and the acoustic emission signal training sample time series, wherein the cutting force training sample time series includes a set of cutting force training sample values ​​for N time steps, and the acoustic emission signal training sample time series includes a set of acoustic emission signal training sample values ​​for N time steps. The objective loss function for constructing the impact force model is determined according to the following formula:

[0011] in, It is the calculated value of the target loss function, which is the mean square error loss value of the impact force; It is the number of sample values; It is the first The smooth predictive impact of a sample; It is the first One cutting force; The initial model of the impact force model is trained based on the time series of cutting force training samples, the time series of acoustic emission signal training samples, and the target loss function, and the trained model of the impact force model is obtained. If the target loss function satisfies the preset convergence condition, the training model of the impact force model is stopped and the impact force model is obtained.

[0012] In the embodiments of this application, the sampling frequency of the high-frequency acoustic emission signal detection module is not less than twice the maximum value of the preset frequency.

[0013] In the embodiments of this application, the PDC composite sheet includes a PDC cutting part and a PDC base disposed on one side of the PDC cutting part. The PDC cutting part is used to impact cut the rock sample. An adhesive area is formed on the axial sidewall of the PDC base away from the PDC cutting part. A high-frequency acoustic emission signal detection module is connected to the adhesive area. The roughness of the adhesive area is in the range of 0.8μm-1.6μm.

[0014] In the embodiments of this application, the shape of the bonding area is a square with a side length of not less than 8mm or a circle with a diameter of not less than 8mm.

[0015] In the embodiments of this application, the high-frequency acoustic emission signal detection module and the PDC composite sheet are bonded together by a coupling agent. The coupling agent has an acoustic impedance that matches the preset frequency to transmit the acoustic emission signal of the preset frequency.

[0016] In the embodiments of this application, the impact force detection device further includes a slide rail and a clamping base. The clamping base is movably disposed above the slide rail and is used to clamp the rock sample. The clamping base is made of a material with a relative damping coefficient of 20×10⁻⁶. -4 -100×10 -4 The material is HT300 gray cast iron.

[0017] As can be seen from the above technical solution, the impact force detection device includes: a PDC composite sheet clamping module for clamping the PDC composite sheet located above the rock sample; a cutting force detection module connected to the PDC composite sheet clamping module for detecting the cutting force generated when the PDC composite sheet impacts and cuts the rock sample; a high-frequency acoustic emission signal detection module connected to the sidewall of the axial side of the PDC composite sheet for detecting the acoustic emission signal of a preset frequency generated when the PDC composite sheet impacts and cuts the rock sample, wherein the preset frequency range is 100kHz-1MHz; and a processor communicatively connected to the cutting force detection module and the high-frequency acoustic emission signal detection module and configured to: acquire the cutting force and acoustic emission signal; perform noise reduction processing on the acoustic emission signal to obtain noise-reduced emission signal data; perform feature extraction processing on the noise-reduced emission signal data to obtain a feature vector of the noise-reduced emission signal data; acquire an impact force model; and input the cutting force and feature vector into the impact force model to calculate the impact force generated when the PDC composite sheet impacts and cuts the rock sample. The high-frequency acoustic emission signal detection module of the impact force detection device is directly connected to one side of the PDC composite sheet, which reduces the attenuation of high-frequency stress waves and ensures that real high-frequency acoustic emission signals can be collected. Furthermore, the processor can obtain real impact force data containing high-frequency stress wave information, providing direct and accurate quantitative basis for evaluating the impact resistance performance of the PDC composite sheet, optimizing drilling parameters, and providing early warning of abnormal working conditions.

[0018] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0019] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without any inventive effort. In the drawings: Figure 1 This is a schematic diagram of the impact force detection device in this embodiment; Figure 2 This is a schematic diagram of the PDC composite sheet and the high-frequency acoustic emission signal detection module in this embodiment; Figure 3 This is a cross-sectional view of the PDC composite sheet and the high-frequency acoustic emission signal detection module in this embodiment; Figure 4 This is a side view of the PDC composite sheet and the high-frequency acoustic emission signal detection module in this embodiment; Figure 5 This is a comparison chart of the impact force result curve of a traditional force gauge and the impact force result curve of the impact force detection device in this embodiment.

[0020] Explanation of reference numerals in the attached figures 1-PDC composite sheet clamping module; 2-support frame; 3-PDC composite sheet; 301-PDC cutting part; 302-PDC base; 4-high frequency acoustic emission signal detection module. Detailed Implementation

[0021] The specific embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this application.

[0022] One embodiment of this application provides an impact force detection device for PDC composite sheets, such as... Figures 1-4 As shown, the impact force detection device includes: PDC composite sheet clamping module 1 is used to clamp the PDC composite sheet 3 located above the rock sample; The cutting force detection module is connected to the PDC composite sheet clamping module 1 and is used to detect the cutting force generated when the PDC composite sheet 3 impacts and cuts the rock sample. The high-frequency acoustic emission signal detection module 4 is connected to the side wall on one side of the axial direction of the PDC composite sheet 3 and is used to detect the acoustic emission signal of a preset frequency generated when the PDC composite sheet 3 impacts and cuts the rock sample. The preset frequency range is 100kHz-1MHz. The processor is communicatively connected to the cutting force detection module and the high-frequency acoustic emission signal detection module 4 and is configured to perform the following steps: Step S101: Acquire cutting force and acoustic emission signals; Step S102: Perform noise reduction processing on the acoustic emission signal to obtain noise-reduced emission signal data; Step S103: Perform feature extraction processing on the noise-reduced transmitted signal data to obtain a feature vector of the noise-reduced transmitted signal data; Step S104: Obtain the impact force model; Step S105: Input the cutting force and feature vector into the impact force model to calculate the impact force generated when the PDC composite sheet 3 performs impact cutting on the rock sample.

[0023] Specifically, such as Figure 1 As shown, the impact force detection device in this embodiment also includes a support frame 2, which can be a VTL lathe. The PDC composite sheet clamping module 1 is movably mounted on the VTL lathe to ensure that the PDC composite sheet 3 is located at a preset position and can smoothly perform impact cutting on the rock sample. In this embodiment, the PDC composite sheet 3 is cylindrical and located below the PDC composite sheet clamping module 1. The PDC composite sheet 3 has a forward tilt angle of 15°-25° so that the cutting edge of its cylindrical outer edge can have a small contact area with the rock sample below, thereby ensuring that the PDC composite sheet 3 can cut the rock sample. Furthermore, the rock sample in this embodiment is selected as homogeneous dense dolomite (such as dolomite from the deep structure of the Tarim Basin), and the rock sample has been strictly screened to be free of clay components, is pure dense rock, and has a uniaxial compressive strength (UCS) of 180 MPa. Before the cutting experiment begins, the rotation speed of the PDC composite sheet 3 and the cutting depth of the rock sample are set. In this embodiment, the cutting depth is selected as 2.0 mm.

[0024] In this embodiment, the high-frequency acoustic emission signal detection module 4 can be selected as a high-frequency acoustic emission signal sensor. During the impact cutting of the rock sample by the PDC composite sheet 3, the PDC composite sheet 3 causes the intergranular microcracks inside the rock sample to expand rapidly and release a large number of high-frequency stress waves (frequency range of 100kHz-1MHz). These high-frequency stress waves can propagate along the PDC composite sheet. In order to capture the high-frequency stress wave signal, the high-frequency acoustic emission signal detection module 4 in this embodiment is directly connected to the axial sidewall of the PDC composite sheet 3 away from the rock sample. This allows the high-frequency stress waves of the preset frequency generated when the PDC composite sheet 3 impacts and cuts the rock sample to be directly transmitted to the high-frequency acoustic emission sensor and converted into acoustic emission signals of the preset frequency. The propagation path is simple and direct, reducing data loss and energy attenuation of the high-frequency acoustic emission signal. Meanwhile, in order to detect the low-frequency stress waves generated in the cutting experiment (i.e., stress waves with a frequency of less than 100kHz, which can reflect the average level of macroscopic and overall cutting force during the cutting process, and do not include high-frequency stress waves that have been filtered and attenuated), this embodiment also sets a cutting force detection module (optionally a three-dimensional force measuring instrument) above the PDC composite sheet clamping module 1. The cutting force detection module can detect the signal data of low-frequency cutting force at each time step generated when the PDC composite sheet 3 impacts and cuts the rock sample in the cutting experiment.

[0025] First, the cutting force (i.e., low-frequency data) and acoustic emission signal (i.e., high-frequency data) at each time step are obtained using the above method. Then, wavelet transform is applied to the acoustic emission signal (which is a voltage signal) data for noise reduction, removing environmental noise and other components to obtain the noise-reduced emission signal data. The processor performs time-domain and frequency-domain analysis on the noise-reduced emission signal data to obtain the feature vectors for each time step of the noise-reduced emission signal data. These feature vectors are denoted as X = [x1, x2, ..., x...]. n That is, each feature vector contains n-dimensional features; the processor calls the impact force model (which is pre-stored in the processor's accessible memory), inputs the cutting force and feature vectors of each time step into the impact force model, and finally outputs the impact force result data of each time step.

[0026] like Figure 5 As shown, the time series curve of impact force data measured by a traditional force gauge (such as a triaxial force gauge) presents a continuous, smooth, and low-ripple shape with almost no sharp peaks or troughs. This indicates that the high-frequency components in the stress wave detected by the traditional force gauge have been attenuated, and therefore only impact force data containing low-frequency stress wave information is detected. In contrast, the time series curve of impact force data at each time step in this embodiment is no longer smooth, but contains multiple steep and significant pulse peaks, presenting an overall sawtooth shape. This shape corresponds to the information of high-frequency stress waves released when the rock sample undergoes brittle fracture.

[0027] Therefore, the impact force detection device in this embodiment can reflect the real impact load curve during the cutting process, providing data support that traditional smooth curves cannot provide for the impact resistance performance evaluation and life prediction of PDC composite sheet 3. At the same time, it provides a reliable basis for the optimized design of PDC composite sheet 3. In addition, the high-frequency acoustic emission signal detection module 4 in this embodiment is set on the side wall on one side of the axial direction of PDC composite sheet 3, which shortens the propagation path of high-frequency stress wave to the greatest extent and reduces the attenuation of high-frequency stress wave, thereby enabling the capture of real high-frequency acoustic emission signal.

[0028] In one embodiment of this application, the feature vector includes the mean of the noise-reduced transmitted signal data, the variance of the noise-reduced transmitted signal data, the root mean square of the noise-reduced transmitted signal data, the peak value of the noise-reduced transmitted signal data, the skewness of the noise-reduced transmitted signal data, the kurtosis of the noise-reduced transmitted signal data, the waveform factor of the noise-reduced transmitted signal data, the impulse factor of the noise-reduced transmitted signal data, the dominant frequency of the noise-reduced transmitted signal data, the centroid frequency of the noise-reduced transmitted signal data, the mean square frequency of the noise-reduced transmitted signal data, and the energy proportion of a specific high-frequency band of the noise-reduced transmitted signal data.

[0029] Specifically, the feature vector in this embodiment includes 12 dimensions: 8-dimensional time-domain features of the noise-reduced transmitted signal data: mean (x1), variance (x2), root mean square (x3), peak value (x4), skewness (x5), kurtosis (x6), waveform factor (x7), and impulse factor (x8); and 4-dimensional frequency-domain features of the noise-reduced transmitted signal data: dominant frequency (x9), centroid frequency (x1), and peak value (x2). 10 ), mean square frequency (denoted as x) 11 ) and the energy percentage of a specific high-frequency band (denoted as x) 12 The eigenvectors are denoted as X = [x1, x2, ..., x]. 12 ].

[0030] The 8-dimensional time-domain features primarily map the energy accumulation and statistical anomalies of shock waves from a macroscopic perspective. Among them, the root mean square and mean reflect the average energy level of the noise-reduced transmitted signal data within the time step; variance describes the fluctuation and dispersion of the noise-reduced transmitted signal data; peak value captures the instantaneous maximum amplitude of the noise-reduced transmitted signal data; kurtosis and impulse factor are extremely sensitive to the abrupt and short-lived peak impacts when rock samples undergo brittle fracture, and can effectively distinguish between normal cutting and sudden fracture events; skewness characterizes the waveform asymmetry of the noise-reduced transmitted signal data and can indicate the rising / falling edge characteristics of the shock wave; and waveform factor reflects the changes in the overall waveform shape of the noise-reduced transmitted signal data.

[0031] The 4-dimensional frequency domain characteristics strictly correspond to the micro-scale intergranular microfracture mechanism. The frequency level is directly related to the fracture scale of dense dolomite: a surge in the energy proportion of a specific high-frequency band often corresponds to intergranular cleavage fracture at a very small scale, i.e., brittle fracture is dominant; the shift of the dominant frequency to lower frequencies indicates the convergence and connection of macroscopic large cracks, and the fracture mode tends to be ductile or large-scale spalling; the centroid frequency and mean square frequency quantify the overall distribution range and dispersion of spectral energy.

[0032] Using fewer than 12 features in the feature vector leads to the loss of high-frequency specificity of transient impacts and fails to capture the scale differences of microscopic fractures. Introducing more redundant features (such as 20 or more dimensions) introduces strongly correlated noise, resulting in computational redundancy and a high susceptibility to overfitting. Therefore, a 12-dimensional feature vector achieves the most efficient compression while ensuring information integrity, providing a sensitive yet robust input for the subsequent impact force model, thus ensuring the accurate reproduction of transient impact force curves that traditional force gauges cannot capture.

[0033] In one embodiment of this application, the impact force model is as follows:

[0034] in, It is impact force; It is an eigenvector; It is a local feature extraction layer; It is an output layer that fuses hidden state vectors; It is the attention mechanism layer.

[0035] Specifically, in this embodiment, the local feature extraction layer can be selected as a one-dimensional convolutional neural network, the fusion hidden state vector output layer can be selected as a long short-term memory network, and the attention mechanism layer can be selected as an attention model. The feature vectors X = [x1, x2, ..., x] at each time step are... 12 As a time series The input is processed by a one-dimensional convolutional neural network. The output of the one-dimensional convolutional neural network and the cutting force are then input into a long short-term memory network for further calculation. Finally, the output of the long short-term memory network is input into... The attention model is used to perform calculations, and the impact data at each time step is finally output.

[0036] In one embodiment of this application, step S105, which inputs the cutting force and feature vector into the impact force model to calculate the impact force generated when the PDC composite sheet 3 performs impact cutting on the rock sample, includes steps S201-S203, wherein: Step S201: Input the feature vector into the local feature extraction layer of the impact force model to obtain the local features of the noise-reduced transmission signal data; Step S202: Input the cutting force and local features into the fusion hidden state vector output layer of the impact force model to obtain the fusion hidden state vector at each time step, wherein the fusion hidden state vector is formed by fusing local features and cutting force; Step S203: Input the fused hidden state vectors of each time step into the attention mechanism layer of the impact force model to obtain the impact force of each time step when the PDC composite sheet 3 impacts and cuts the rock sample.

[0037] Specifically, firstly, the 12-dimensional feature vectors at each time step are used as the time series X(t) and input into a one-dimensional convolutional neural network (denoted as...). This one-dimensional convolutional neural network slides along the time axis through multiple learnable convolutional kernels to automatically extract local combination patterns between 12-dimensional features in each time step, and calculates and outputs local features of the noise-reduced transmitted signal data in each time step. The processor synchronously acquires low-frequency cutting force signals (collected by a triaxial force gauge, in Newtons). These cutting force signals are strictly aligned with the acoustic emission signals in terms of acquisition time; for example, the cutting force signals at 0:00:00, 0:00:01, and 0:00:02 are aligned with the acoustic emission signals. However, within the time intervals between these moments, the amount of data from high-frequency acoustic emission signals is always greater than that from low-frequency cutting force signals (for example, within a 1-second time interval, there might be 1000 high-frequency acoustic emission signal data points, while there might only be 2 low-frequency cutting force signal data points). Therefore, it is neither necessary nor possible to align each cutting force signal with each acoustic emission signal; instead, it is sufficient to align the cutting force signals at each moment with the acoustic emission signals. Following this alignment method, the cutting force values ​​at the same time step are fused with the local features obtained in the previous step to form a fused feature vector. Then, the fused feature vectors from all time steps are input into a Long Short-Term Memory (LSTM) network in chronological order (denoted as ). This Long Short-Term Memory (LSTM) network recursively processes the input at each time step along the time axis, outputting a fused hidden state vector sequence. Each hidden state vector in this sequence simultaneously incorporates local features of the denoised transmitted signal data from the current and historical time steps, as well as low-frequency cutting force information, reflecting the dynamic coupling relationship between macroscopic loads and microscopic high-frequency stress waves.

[0038] The processor acquires a pre-built attention model (denoted as...). The model works as follows: For each current time step, using the fused hidden state vector of the current time step as the standard, it calculates the correlation between the fused hidden state vectors of all historical time steps and the query vector, and obtains weight coefficients through normalization. Then, it performs a weighted summation of the hidden state vectors of each historical time step based on the weight coefficients to generate a weighted feature vector (also called a context vector). This weighted feature vector highlights the moments in history most relevant to the current impact. By inputting the weighted feature vector into a fully connected layer for mapping, the impact force at the current time step can be obtained. By repeating the above process for all time steps, the impact force can be obtained. The time series.

[0039] The impact force model in this embodiment overcomes the bottleneck of insufficient frequency in traditional force gauges, outputting impact force curves with high temporal resolution (i.e., high frequency), clearly presenting the peak pulses of each brittle fracture of the rock sample; this impact force model significantly improves the accuracy of impact force peak prediction (correlation coefficient (R)) through multi-source information fusion and attention weighting. 2 (reaching 0.96 or higher); This impact force model is also easy to adapt to different lithologies and working conditions, providing quantitative basis that traditional methods cannot obtain for drill bit (PDC composite 3) life assessment, drilling parameter optimization and abnormal impact early warning.

[0040] In one embodiment of this application, the impact force model is constructed in the following manner: Step S301: Obtain the initial model of the impact force model; Step S302: Obtain the cutting force training sample time series and the acoustic emission signal training sample time series, wherein the cutting force training sample time series includes a set of cutting force training sample values ​​for N time steps, and the acoustic emission signal training sample time series includes a set of acoustic emission signal training sample values ​​for N time steps. Step S303: Construct the target loss function for the impact force model, wherein the target loss function is determined according to the following formula:

[0041] in, It is the calculated value of the target loss function, which is the mean square error loss value of the impact force; It is the number of sample values; It is the first The smooth predictive impact of a sample; It is the first The cutting force of each sample; Step S304: Train the initial model of the impact force model based on the cutting force training sample time series, the acoustic emission signal training sample time series, and the target loss function, and obtain the training model of the impact force model; Step S305: If the target loss function satisfies the preset convergence condition, stop training the impact force model and obtain the impact force model.

[0042] Specifically, before the model training computation begins, an untrained initial impact force model is obtained, which contains a one-dimensional convolutional network, a long short-term memory network, and an attention model, but its parameters are set to random initial values. During the cutting experiment, the time series of the cutting force and the time series of the acoustic emission signal are detected and obtained. The time series of the cutting force is the cutting force training sample time series (i.e., containing the time series of each time step). The time series of the acoustic emission signal is the time series of the acoustic emission signal training samples, and each of the two time series includes M training sample values. During model training, the time series of acoustic emission signal training samples are first denoised to obtain a denoised emission signal training sample time series. Then, the time series of local features of the denoised emission signal training sample time series are obtained and input into the initial model for calculation to obtain the initial predicted impact force at each time step (denoted as ). (i.e., the initial predicted impact force of the j-th time step); then, the time series of the initial predicted impact force is smoothed (smoothing means eliminating high-frequency information in the initial predicted impact force, thereby eliminating sharp peaks and troughs in the curve of the initial predicted impact force), to obtain the smoothed predicted impact force at time step. (The smoothed predicted impact no longer contains high-frequency information, and its curve appears as follows) Figure 5 The smooth shape of the impact force curve of the traditional force gauge), and then M, , Substitute each value into the objective loss function and verify whether it converges. The preset convergence condition is... <1. If the preset convergence condition is not met, adjust the parameters in the initial model and recalculate according to the above steps. The training continues until the preset convergence condition is met; at this point, the model is considered a successfully trained impact force model.

[0043] The smoothed predicted impact force is obtained as follows: During the training of the impact force model, a time series of training predicted impact force values ​​is obtained; a time-impact force data curve is plotted based on the time series of training predicted impact force values; the time-impact force data curve is smoothed to obtain the smoothed predicted impact force value at each time step. When smoothing the time series of the initial predicted impact force, the sampling period of the low-frequency cutting force (T) is used. L =1 / f L T L That is, the sampling period, f L Using the sampling frequency as the length of the sliding time window, the dynamic characteristic value of the high-frequency impact force within each time window is calculated to obtain the equivalent impact force at each smoothed time step. The dynamic eigenvalues ​​are calculated using either the root mean square (RMS) algorithm or the window arithmetic mean algorithm.

[0044] In one embodiment of this application, the sampling frequency of the high-frequency acoustic emission signal detection module 4 is not less than twice the maximum value of the preset frequency. The selection of this parameter is based on the Nyquist-Shannon sampling theorem, that is, the sampling frequency should be at least twice the highest frequency of the acquired signal. In this embodiment, the highest frequency of the acquired signal is 1MHz, and the preferred sampling frequency in this embodiment is 5MS / s (the minimum sampling rate required for the highest frequency of the acquired signal of 1MHz is 2×1MHz=2MS / s). This sampling frequency can ensure that the microsecond-level transient pulse waveform generated when the rock sample undergoes brittle fracture can be accurately captured, without causing waveform distortion, peak loss or false frequency components. At the same time, it avoids the data explosion and storage pressure caused by excessively high sampling frequencies, and provides a reliable original data basis for subsequent calculations.

[0045] In one embodiment of this application, the PDC composite sheet 3 includes a PDC cutting part 301 and a PDC base 302 disposed on one side of the PDC cutting part 301. The PDC cutting part 301 is used to impact cut the rock sample. An adhesive area is formed on the axial sidewall of the PDC base 302 away from the PDC cutting part 301. The high-frequency acoustic emission signal detection module 4 is connected to the adhesive area. The roughness range of the adhesive area is 0.8μm-1.6μm.

[0046] Specifically, such as Figure 2 As shown, the cutting edge of the cylindrical outer edge of the PDC cutting part 301 is used to contact and cut the rock sample. The PDC base 302 is directly held by the PDC composite sheet clamping module. If the bonding area on the axial sidewall of the PDC base 302 away from the PDC cutting part 301 is too smooth, high-frequency stress waves are easily reflected when they are transmitted to this point, which makes it impossible for the high-frequency acoustic emission signal detection module 4 to detect the real high-frequency acoustic emission signal. The roughness setting of the bonding area in this embodiment can ensure the smooth propagation of high-frequency stress waves and improve the acquisition accuracy and precision of high-frequency acoustic emission signals.

[0047] In one embodiment of this application, the shape of the bonding area is a square with a side length of not less than 8 mm or a circle with a diameter of not less than 8 mm. This ensures that there is a sufficient effective contact area between the high-frequency acoustic emission signal detection module 4 and the PDC composite sheet 3, avoiding stress concentration caused by insufficient contact area, thereby ensuring the smooth transmission of stress waves. In addition, the regular shape of the square or circle is conducive to standardized processing and assembly, avoiding stress concentration or residual stress generated during grinding or bonding of irregular edges, thereby ensuring the long-term stability of the bonding area and ensuring that the high-frequency stress waves are transmitted to the high-frequency acoustic emission signal detection module 4 in a consistent manner over a long experimental period.

[0048] In one embodiment of this application, the high-frequency acoustic emission signal detection module 4 and the PDC composite sheet 3 are bonded together with a coupling agent. The coupling agent has an acoustic impedance that matches the preset frequency to transmit the acoustic emission signal at the preset frequency. In this embodiment, the coupling agent is preferably an epoxy resin-based coupling agent. If the acoustic impedance of the coupling agent cannot match the preset frequency, interface reflection will occur when the high-frequency stress wave is transmitted to the coupling agent, and the accuracy of the acquired high-frequency acoustic emission signal data will be greatly reduced. By using a coupling agent whose acoustic impedance matches the preset frequency, the high-frequency stress wave can pass through the connection interface between the high-frequency acoustic emission signal detection module 4 and the PDC composite sheet 3 with the highest transmittance, thereby obtaining a high-frequency acoustic emission signal with full amplitude and high waveform fidelity, providing high-quality raw data for subsequent noise reduction, feature extraction, and impact force calculation processes.

[0049] In one embodiment of this application, the impact force detection device further includes a slide rail and a clamping base. The clamping base is movably disposed above the slide rail and is used to clamp the rock sample. The clamping base is made of a material with a relative damping coefficient of 20×10⁻⁶. -4 -100×10 -4 The material is HT300 gray cast iron.

[0050] Specifically, such as Figure 1 As shown, a slide rail portion is formed on the upper surface of the slide rail seat, and a structure that mates with the slide rail portion is formed on the side of the clamping base near the slide rail seat, allowing the clamping base to slide on the slide rail seat, thereby adjusting the position of the rock sample. During the cutting experiment, the stress wave generated by the cutting process propagates not only towards the PDC composite sheet 3, but also towards the clamping base below, with a relative damping coefficient of 20×10⁻⁶. -4 -100×10 -4 The HT300 gray cast iron material can effectively absorb and dissipate the energy propagated to the clamping base, avoiding interference with the collected high-frequency acoustic emission signals. At the same time, the clamping base made of this material has sufficient structural strength to ensure stable clamping of the rock sample, ensuring that the collected acoustic emission signals mainly reflect the real impact process between the PDC composite sheet 3 and the rock sample, reducing the influence of mechanical boundary conditions on the experimental data.

[0051] In the description of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0052] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between components; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0053] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0054] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. An impact force detection device for PDC composite sheets, characterized in that, The impact force detection device includes: PDC composite sheet clamping module (1) is used to clamp the PDC composite sheet (3) located above the rock sample. The cutting force detection module is connected to the PDC composite sheet clamping module (1) and is used to detect the cutting force generated when the PDC composite sheet (3) impacts and cuts the rock sample; The high-frequency acoustic emission signal detection module (4) is connected to the side wall on one side of the axial direction of the PDC composite sheet (3) and is used to detect the acoustic emission signal of a preset frequency generated when the PDC composite sheet (3) impacts and cuts the rock sample, wherein the preset frequency ranges from 100kHz to 1MHz. The processor is communicatively connected to the cutting force detection module and the high-frequency acoustic emission signal detection module (4) and is configured as follows: Obtain the cutting force and the acoustic emission signal; The acoustic emission signal is subjected to noise reduction processing to obtain noise-reduced emission signal data; The noise-reduced transmitted signal data is subjected to feature extraction processing to obtain a feature vector of the noise-reduced transmitted signal data; Obtain the impact force model; The cutting force and the feature vector are input into the impact force model to calculate the impact force generated when the PDC composite sheet (3) performs impact cutting on the rock sample.

2. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The feature vector includes the mean of the noise-reduced transmitted signal data, the variance of the noise-reduced transmitted signal data, the root mean square of the noise-reduced transmitted signal data, the peak value of the noise-reduced transmitted signal data, the skewness of the noise-reduced transmitted signal data, the kurtosis of the noise-reduced transmitted signal data, the waveform factor of the noise-reduced transmitted signal data, the impulse factor of the noise-reduced transmitted signal data, the dominant frequency of the noise-reduced transmitted signal data, the centroid frequency of the noise-reduced transmitted signal data, the mean square frequency of the noise-reduced transmitted signal data, and the energy proportion of a specific high-frequency band of the noise-reduced transmitted signal data.

3. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The impact force model is as follows: in, It is the impact force mentioned; It is the feature vector; It is a local feature extraction layer; It is an output layer that fuses hidden state vectors; It is the attention mechanism layer.

4. The impact force detection device for PDC composite sheets according to claim 3, characterized in that, The step of inputting the cutting force and the feature vector into the impact force model to calculate the impact force generated when the PDC composite sheet (3) performs impact cutting on the rock sample includes: The feature vector is input into the local feature extraction layer of the impact force model to obtain the local features of the noise-reduced transmission signal data; The cutting force and the local features are input into the fusion hidden state vector output layer of the impact force model to obtain the fusion hidden state vector at each time step, wherein the fusion hidden state vector is formed by fusing the local features and the cutting force. The fused hidden state vectors at each time step are input into the attention mechanism layer of the impact force model to obtain the impact force at each time step when the PDC composite sheet (3) performs impact cutting on the rock sample.

5. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The impact force model is constructed in the following manner: Obtain the initial model of the impact force model; Obtain the cutting force training sample time series and the acoustic emission signal training sample time series, wherein the cutting force training sample time series includes a set of cutting force training sample values ​​for N time steps, and the acoustic emission signal training sample time series includes a set of acoustic emission signal training sample values ​​for N time steps. The target loss function of the impact force model is constructed, wherein the target loss function is determined according to the following formula: in, It is the calculated value of the target loss function, where the calculated value of the target loss function is the mean square error loss value of the impact force; It is the number of sample values; It is the first The smooth predictive impact of a sample; It is the first The cutting force of each sample; The initial model of the impact force model is trained based on the cutting force training sample time series, the acoustic emission signal training sample time series, and the target loss function, and the trained model of the impact force model is obtained. If the target loss function satisfies the preset convergence condition, the training model of the impact force model is stopped and the impact force model is obtained.

6. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The sampling frequency of the high-frequency acoustic emission signal detection module (4) is not less than twice the maximum value of the preset frequency.

7. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The PDC composite sheet (3) includes a PDC cutting part (301) and a PDC base (302) disposed on one side of the PDC cutting part (301). The PDC cutting part (301) is used to impact cut the rock sample. An adhesive area is formed on the axial sidewall of the PDC base (302) away from the PDC cutting part (301). The high-frequency acoustic emission signal detection module (4) is connected to the adhesive area. The roughness range of the adhesive area is 0.8μm-1.6μm.

8. The impact force detection device for PDC composite sheets according to claim 7, characterized in that, The shape of the bonding area is a square with a side length of not less than 8mm or a circle with a diameter of not less than 8mm.

9. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The high-frequency acoustic emission signal detection module (4) and the PDC composite sheet (3) are bonded together by a coupling agent, which has an acoustic impedance that matches the preset frequency to transmit the acoustic emission signal of the preset frequency.

10. The impact force detection device for PDC composite sheets according to claim 1, characterized in that, The impact force detection device also includes a slide rail and a clamping base. The clamping base is movably disposed above the slide rail and is used to clamp the rock sample. The clamping base is made of a material with a relative damping coefficient of 20×10⁻⁶. -4 -100×10 -4 The material is HT300 gray cast iron.

Images