Intelligent control method and system for diamond compact laser processing equipment

Abstract

The application relates to the technical field of processing control, and discloses an intelligent control method and system for a diamond compact laser processing device. The method comprises the following steps: scanning a diamond compact by using a spectral reflection sensor and an acoustic emission signal device to obtain a four-dimensional characteristic tensor; inputting the tensor into a double-channel convolution network to identify a grain boundary type; executing a PDC-Lambda algorithm to construct a grain boundary-power response mapping matrix; and establishing a three-level power regulation architecture based on the mapping matrix to realize accurate processing control. The application realizes accurate classification and positioning of different grain boundary types. The problem of lacking a grain boundary-power mapping mechanism in traditional methods is solved, and a corresponding relationship between grain boundary characteristics and optimal processing parameters is established. The problem of insufficient power regulation accuracy is solved, fine parameter control from a macroscopic to a microscopic level is realized through the three-level regulation architecture, and thermal damage and micro-crack formation in the grain boundary region are effectively prevented.

Description

Technical Field

[0001] This application relates to the field of processing control technology, and in particular to an intelligent control method and system for laser processing equipment for diamond composite sheets. Background Technology

[0002] Polycrystalline diamond composites (PDC) are widely used in precision cutting tools, drill bits, and abrasives due to their ultra-high hardness and excellent wear resistance. Currently, laser processing has become the primary method for processing extremely hard PDC materials, and its processing efficiency and precision largely depend on the precise control of laser parameters. Traditional laser processing equipment for diamond composites mainly employs fixed parameters or simple closed-loop control methods, adjusting parameters such as laser power, pulse frequency, and scanning speed as a whole. This method can meet basic requirements when processing homogeneous materials.

[0003] However, in practical applications, the complex grain boundary structure formed during the high-pressure, high-temperature synthesis of PDC materials results in internal inhomogeneities, making traditional control methods difficult to adapt. Existing control methods generally lack the ability to accurately identify the microstructural characteristics of PDC materials, and cannot optimize parameters for different grain boundary types and distributions. Furthermore, during processing, traditional methods fail to adequately predict the accumulation of thermal stress in grain boundary regions, easily leading to microcracks and thermal damage. In addition, due to a lack of in-depth understanding of the relationship between material preparation conditions and processing parameters, existing control systems typically require extensive trial-and-error experiments to determine suitable parameters, resulting in low efficiency and significant material waste.

[0004] More importantly, with the development of high-pressure and high-temperature technology, the hardness and structural complexity of new PDC materials are constantly increasing, highlighting the growing incompatibility between traditional control methods and new materials. In particular, when PDC materials are prepared under high pressure of 14 GPa and high temperature of over 1600°C, various types of grain boundaries are interspersed within them. How to accurately identify grain boundaries, how to establish a mapping relationship between grain boundary characteristics and processing parameters, and how to achieve fine power control at the microscale have become urgent technical challenges to be solved. Summary of the Invention

[0005] This application provides an intelligent control method and system for laser processing equipment of diamond composite sheets, achieving precise classification and positioning of different grain boundary types. It solves the problem of the lack of a grain boundary-power mapping mechanism in traditional methods, establishing a correspondence between grain boundary characteristics and optimal processing parameters. It also addresses the problem of insufficient power control precision, achieving refined parameter control from macroscopic to microscopic levels through a three-level control architecture, effectively preventing thermal damage and microcrack formation in the grain boundary region.

[0006] In a first aspect, this application provides an intelligent control method for a laser processing equipment for diamond composite sheets. The intelligent control method includes: scanning the diamond composite sheet using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectra, and time-frequency characteristics of the acoustic emission signal; inputting the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhancement convolutional network for feature extraction and grain boundary identification to obtain a grain boundary type identification result; executing the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters; and establishing a three-level power control architecture based on the grain boundary-power response mapping matrix, consisting of macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping, to control the processing of the diamond composite sheet.

[0007] Optionally, the scanning of the diamond composite sheet using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum, and time-frequency characteristics of the acoustic emission signal includes:

[0008] The diamond composite sheet is fixed on the processing platform, and the diamond composite sheet is positioned using a three-dimensional positioning system to obtain the positioned diamond composite sheet.

[0009] The diamond composite sheet after positioning was surface scanned with a step spacing of 50 μm, and each sampling point was excited by five laser light sources of different wavelengths to obtain reflectance spectrum data;

[0010] Acoustic emission data is obtained by collecting acoustic signals at each sampling point on the surface of the positioned diamond composite sheet using four symmetrically arranged high-frequency acoustic emission sensors.

[0011] The reflection spectrum data and the acoustic emission data are processed by median filtering, wavelet denoising and frequency domain filtering to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency features of acoustic emission signal.

[0012] Optionally, the step of inputting the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary identification to obtain grain boundary type identification results includes:

[0013] The four-dimensional feature tensor is separated into two parts: spectral information and acoustic emission signal. These parts are then input into the spectral information channel and acoustic emission signal channel of the dual-channel grain boundary feature spectrum enhancement convolutional network, respectively, to obtain the initial feature map.

[0014] The initial feature maps of the spectral information channels are processed through multiple convolutional layers. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function. The last layer uses one-dimensional convolution to integrate the features and obtain the spectral feature extraction results.

[0015] The initial feature map of the acoustic emission signal channel is processed by a dual time-frequency domain processing structure. First, the time-domain signal is converted into a spectrum map through short-time Fourier transform, and then the spectrum features are extracted through a multi-layer convolutional network. At the same time, the original time-domain signal is extracted with time-domain features through a one-dimensional convolutional network. The two parts of features are fused through an attention mechanism to obtain the acoustic emission feature extraction result.

[0016] The spectral feature extraction results and the acoustic emission feature extraction results are integrated by the feature fusion module. The diamond grain boundary morphology spectrum analysis unit with residual connection structure and void convolution and the phase transformation prediction and evaluation unit based on material phase transformation theory are applied for processing to obtain grain boundary type identification results including straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions and high-density defect regions.

[0017] Optionally, the step of executing the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters includes:

[0018] The grain boundary type identification results are classified into straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions, and high-density defect regions. Each grain boundary type is encoded as the input feature of the PDC-Lambda algorithm according to the feature vector to obtain grain boundary feature encoding data.

[0019] The grain boundary feature encoding data is processed using the iteration coefficient calculation module in the PDC-Lambda algorithm, with three evaluation indicators: surface roughness, heat-affected zone width, and microcrack density. The data is then fitted using a quadratic response surface model to obtain the grain boundary response surface model.

[0020] The PDC-Lambda algorithm is applied to the grain boundary response surface model for matrix mapping transformation. The processing parameters are calculated using an iterative optimization strategy. A comprehensive processing quality index is constructed by weighted combination of surface roughness, heat-affected zone width, and microcrack density, and the initial mapping matrix of grain boundary parameter response is obtained.

[0021] The initial mapping matrix of the grain boundary parameter response is input into the multidimensional optimization module of the PDC-Lambda algorithm. A hybrid optimization method combining the improved particle swarm optimization algorithm and the simulated annealing algorithm is applied to calculate the optimal combination of processing parameters for each grain boundary type at different relative positions. The results are stored as a three-dimensional array structure to obtain the grain boundary-power response mapping matrix.

[0022] Optionally, the three-level power control architecture based on the grain boundary-power response mapping matrix—comprising macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping—for processing control of diamond composite sheets includes:

[0023] The initial power parameters were obtained by adjusting the macroscopic power baseline based on the preparation conditions and overall hardness distribution of the diamond composite sheet.

[0024] The initial power parameters are compensated for by meso-grain boundary transition based on the characteristics of the grain boundary region to obtain the compensated power parameters.

[0025] Microscopic pulse shaping is performed based on the compensated power parameters, and the corresponding pulse waveform is selected according to the grain boundary type to obtain the final processing parameters;

[0026] The final processing parameters are applied to the laser processing of diamond composite sheets, and the parameters are dynamically adjusted by real-time monitoring signals.

[0027] Optionally, the step of adjusting the macroscopic power baseline based on the preparation conditions and overall hardness distribution of the diamond composite sheet to obtain the initial power parameters includes:

[0028] The average Vickers hardness, relative density, preparation temperature, and preparation pressure of the diamond composite sheet were obtained to acquire basic material property data.

[0029] Based on the material's basic property data, the power coefficient is calculated by substituting the average Vickers hardness, relative density, preparation temperature, and preparation pressure into the preset power baseline calculation rules to obtain the power baseline calculation results.

[0030] The power baseline calculation results are corrected by comparing the corresponding relationship between material properties and power demand to obtain the correction coefficient.

[0031] The correction coefficient is calculated in conjunction with the power baseline to obtain the initial power parameters for laser power control.

[0032] Optionally, the step of performing meso-grain boundary transition compensation on the initial power parameters based on the characteristics of the grain boundary region to obtain the compensated power parameters includes:

[0033] A heat conduction model was established for the grain boundary region, and the temperature distribution of the grain boundary region under different powers was calculated to obtain the grain boundary temperature distribution data.

[0034] The thermal stress distribution in the grain boundary region is calculated based on the grain boundary temperature distribution data to obtain the grain boundary thermal stress distribution data.

[0035] Based on the grain boundary thermal stress distribution data, the power compensation coefficient of the grain boundary region is determined, and the power is adjusted by the mapping relationship between the thermal stress level and the compensation coefficient to obtain the power compensation coefficient.

[0036] The power compensation coefficient is combined with the initial power parameters to calculate the power parameters after transition compensation of the grain boundary region.

[0037] Secondly, this application provides an intelligent control system for a laser processing equipment for diamond composite sheets, the intelligent control system for the laser processing equipment for diamond composite sheets comprising:

[0038] The scanning module is used to scan the diamond composite sheet using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency characteristics of acoustic emission signal;

[0039] The extraction module is used to input the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary identification, and to obtain the grain boundary type identification result.

[0040] The construction module is used to execute the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters;

[0041] The control module is used to establish a three-level power control architecture based on the grain boundary-power response mapping matrix, which includes macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping, to control the processing of diamond composite sheets.

[0042] Thirdly, an intelligent control device for a diamond composite laser processing equipment is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the intelligent control device for the diamond composite laser processing equipment to execute the aforementioned intelligent control method for the diamond composite laser processing equipment.

[0043] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the aforementioned intelligent control method for a diamond composite laser processing equipment.

[0044] The technical solution provided in this application employs a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to perform multi-dimensional characteristic scanning on diamond composite sheets. This acquires a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflectance intensity spectra, and time-frequency characteristics of acoustic emission signals. This achieves comprehensive perception of the microstructural characteristics of PDC materials, overcoming the limitations of traditional single-characteristic detection and significantly improving the comprehensiveness and accuracy of data acquisition. The four-dimensional feature tensor is input into a specially designed dual-channel grain boundary feature spectrum enhancement convolutional network. By separating and processing spectral information and acoustic emission signals, high-precision identification of different grain boundary types is achieved. The network innovatively integrates advanced deep learning technologies such as residual connections, dilated convolutions, and attention mechanisms, significantly improving the accuracy of grain boundary identification and solving the problems of traditional methods. The method struggles to identify complex grain boundary structures. Based on the grain boundary type identification results, the PDC-Lambda algorithm is executed. This algorithm is specifically designed for the characteristics of diamond composite sheets and can accurately construct the correspondence between grain boundary types and optimal processing parameters, forming a grain boundary-power response mapping matrix. This effectively solves the dilemma of traditional methods relying on a large number of trial and error experiments for parameter optimization, and significantly improves the scientificity and accuracy of processing parameter settings. Based on the grain boundary-power response mapping matrix, a three-level power control architecture is established, realizing comprehensive and refined control from macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation to microscopic pulse shaping. In particular, different pulse waveforms are used for different grain boundary types, which greatly improves the processing accuracy of grain boundary regions and effectively prevents thermal damage and microcrack formation.

[0045] The dual-channel grain boundary feature spectrum enhanced convolutional network of this invention is specifically designed for the spectral and acoustic properties of PDC materials. Its network structure employs a dual-channel processing and feature fusion mechanism to specifically extract grain boundary features. The PDC-Lambda algorithm considers the differences in the properties of diamond materials under different pressure and temperature conditions, combining materials science theory with optimization algorithms to achieve a precise mapping from grain boundary features to processing parameters. The three-level power control architecture adaptively adjusts parameter strategies according to the processing requirements at different scales. These algorithmic features directly solve the core problems in PDC material processing, significantly improving processing accuracy and efficiency, making the precision processing of high-hardness PDC materials possible, and providing an innovative solution for the field of superhard material manufacturing. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a schematic diagram of an embodiment of the intelligent control method for a laser processing equipment for diamond composite sheets in this application.

[0048] Figure 2 This is a schematic diagram of an embodiment of the intelligent control system for a diamond composite sheet laser processing equipment in this application.

[0049] Figure 3 This is a schematic block diagram of the intelligent control device for a diamond composite sheet laser processing equipment in an embodiment of the present invention. Detailed Implementation

[0050] This application provides an intelligent control method and system for a laser processing equipment for diamond composite sheets. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, 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 described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0051] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent control method for laser processing equipment of diamond composite sheets in this application includes:

[0052] Step S101: The diamond composite sheet is scanned using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency characteristics of acoustic emission signal;

[0053] Step S102: Input the four-dimensional feature tensor into the dual-channel grain boundary feature spectrum enhancement convolutional network for feature extraction and grain boundary recognition to obtain the grain boundary type recognition result;

[0054] Step S103: Execute the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary type and processing parameters;

[0055] Step S104: Based on the grain boundary-power response mapping matrix, establish a three-level power control architecture of macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation and microscopic pulse shaping to control the processing of diamond composite sheets.

[0056] It is understood that the executing entity of this application can be an intelligent control system for laser processing equipment of diamond composite sheets, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiments use a server as an example for illustration.

[0057] Specifically, a multi-wavelength spectral reflectance sensor and a high-frequency acoustic emission sensor were used to comprehensively scan the diamond composite sheet, acquiring a complete dataset including its spatial location, spectral reflectance characteristics, and acoustic properties. The raw data was denoised using various filtering techniques to construct a four-dimensional feature tensor, which contains comprehensive feature information for each sampling point on the diamond composite sheet surface. Next, the four-dimensional feature tensor was input into a specially designed dual-channel grain boundary feature spectrum enhancement convolutional network. This network separates spectral information and acoustic emission signals for parallel processing, extracting material features from different dimensions. These features were then integrated through a feature fusion module to accurately identify eight different types of grain boundary structures. Third, based on the grain boundary type identification results, the PDC-Lambda algorithm was executed. This algorithm encodes the grain boundary features and iteratively calculates a mapping relationship between grain boundary characteristics and processing parameters, constructing a grain boundary-power response mapping matrix. This matrix records the optimal combination of processing parameters for different grain boundary types at different locations, providing precise guidance for subsequent processing. Finally, a three-level power control architecture is used to implement precise processing control. Macroscopic power baseline adjustment sets basic parameters based on the overall material characteristics, mesoscopic grain boundary transition compensation compensates for the special characteristics of grain boundary regions, and microscopic pulse shaping selects different waveforms for finer control based on grain boundary type. Through real-time monitoring and dynamic adjustment mechanisms, high-precision control of the laser processing of diamond composite sheets is achieved. For example, when processing a PDC composite sheet prepared at 1900°C under 14 GPa pressure, the system first acquires the four-dimensional feature tensor of the sample. The BE-CNN network identifies multiple grain boundary types on the sample surface, especially multiple intersecting grain boundaries in the central region. The system calculates the optimal power for this region as 58 W and the scanning speed as 120 mm / s using the PDC-Lambda algorithm, and selects a bimodal pulse waveform for processing. Through the synergistic effect of the three-level power control architecture, the final processing accuracy of the sample reaches ±0.8 μm, effectively solving the problem of insufficient accuracy in grain boundary region processing by traditional control methods.

[0058] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0059] The diamond composite sheet is fixed on the processing platform, and a three-dimensional positioning system is used to position the diamond composite sheet to obtain the positioned diamond composite sheet.

[0060] The diamond composite sheet after positioning was surface scanned with a step spacing of 50 μm. Each sampling point was excited by five laser sources of different wavelengths to obtain reflectance spectrum data.

[0061] Acoustic emission data is obtained by collecting acoustic signals at each sampling point on the surface of the diamond composite sheet after positioning using four symmetrically arranged high-frequency acoustic emission sensors.

[0062] Median filtering, wavelet denoising, and frequency domain filtering are performed on the reflection spectrum data and acoustic emission data to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum, and time-frequency characteristics of the acoustic emission signal.

[0063] Specifically, the diamond composite sheet is fixed on a processing platform with a three-axis precision adjustment mechanism. A three-dimensional positioning system combining laser interferometry and image recognition technology is used to precisely position the sample, achieving a positioning accuracy of ±0.5μm, ensuring the spatial accuracy of subsequent data acquisition. After positioning, a high-precision scanning device driven by a stepper motor performs a grid scan on the sample surface with a uniform step interval of 50μm. At each sampling point, five laser sources with wavelengths of 532nm, 635nm, 785nm, 980nm, and 1064nm are triggered sequentially. The reflection spectrum information at each wavelength is collected by a photodetector, forming a three-dimensional data matrix containing spatial position and multi-wavelength reflection intensity. Simultaneously, four high-frequency acoustic emission sensors with a frequency response range of 100kHz to 1MHz are symmetrically arranged around the sample. When the laser irradiates the sample surface, the generated acoustic emission signal is recorded at a sampling rate of 10MS / s to obtain complete acoustic emission time-frequency characteristics. The acquired raw data underwent a three-step processing procedure: first, median filtering was used to eliminate burst noise within a 5×5 window; then, wavelet denoising with a 5-level decomposition was performed using the db4 wavelet basis function to remove background noise; finally, FFT transformation was performed on the signal followed by frequency domain filtering to remove high-frequency random noise components. The processed data was reconstructed into a four-dimensional feature tensor, containing the (x,y) coordinates of each sampling point, the reflection intensity values ​​of five wavelengths, and the acoustic emission spectrum characteristics of four sensors. Taking the fabrication of a PDC composite sheet prepared under 14 GPa pressure and 1900°C temperature as an example, the positioning system locked the sample position within 0.3 μm, and the surface scanning formed a 100×100 sampling point grid, acquiring five wavelength reflection data and four sensor acoustic emission data at each point. Through multi-level filtering, the grain boundary distribution characteristics of the sample surface were clearly revealed, especially the tiny grain boundary structures that were originally obscured by noise, which could be accurately identified.

[0064] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0065] The four-dimensional feature tensor is separated into two parts: spectral information and acoustic emission signal. These parts are then input into the spectral information channel and acoustic emission signal channel of the dual-channel grain boundary feature spectrum enhancement convolutional network, respectively, to obtain the initial feature map.

[0066] The initial feature maps of the spectral information channels are processed through multiple convolutional layers. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function. The last layer uses one-dimensional convolution to integrate the features and obtain the spectral feature extraction results.

[0067] The initial feature map of the acoustic emission signal channel is processed by a dual time-frequency domain processing structure. First, the time-domain signal is converted into a spectrum map through short-time Fourier transform, and then the spectral features are extracted through a multi-layer convolutional network. At the same time, the original time-domain signal is extracted with time-domain features through a one-dimensional convolutional network. The two parts of features are fused through an attention mechanism to obtain the acoustic emission feature extraction result.

[0068] The spectral feature extraction results and acoustic emission feature extraction results are integrated by the feature fusion module. The diamond grain boundary morphology spectrum analysis unit with residual connection structure and void convolution and the phase transformation prediction and evaluation unit based on material phase transformation theory are applied for processing to obtain grain boundary type identification results including straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions and high-density defect regions.

[0069] Specifically, the core of the dual-channel grain boundary feature spectrum enhanced convolutional network (BE-CNN) lies in separating and processing the spectral and acoustic features of diamond composite sheets, and then achieving accurate grain boundary identification through multi-level feature fusion. First, feature separation is performed, separating the multi-wavelength reflection intensity spectrum data and the acoustic emission signal time-frequency feature data from the four-dimensional feature tensor to form two independent three-dimensional data blocks, which are input into the network's spectral information channel and acoustic emission signal channel, respectively. The spectral information channel constructs a convolutional network structure with a depth of 4. The first layer uses 32 3×3×5 convolutional kernels to process multi-wavelength information. Subsequent convolutional layers gradually reduce the kernel size and increase the depth, using 64 3×3×3 convolutional kernels, 128 2×2×3 convolutional kernels, and 256 2×2×1 convolutional kernels, respectively. After each convolutional operation, a batch normalization layer is connected to eliminate internal covariate bias, and nonlinear characteristics are introduced through the ReLU activation function. Finally, 1×1 convolution is used for cross-channel integration to extract a high-dimensional feature map reflecting the spectral response characteristics of the grain boundaries. The acoustic emission signal channel employs a dual time-frequency domain processing structure. First, the original time-domain signal is converted into a spectrum using a short-time Fourier transform (256-point window, 75% overlap), and frequency domain features are extracted using a three-layer convolutional network. Simultaneously, the original time-domain signal is directly processed by a one-dimensional convolutional network (kernel widths of 9, 7, and 5) to extract time-domain features. These two feature sets are fused using an attention mechanism, with attention weights dynamically calculated using a softmax function to highlight key frequencies and time periods. The feature fusion module receives the processing results from both channels. First, it is processed by a Diamond Grain Boundary Morphology Spectrum Analysis (DMFS) unit, which uses a residual connection structure and three layers of void convolution (expansion rates of 1, 2, and 4) to enhance the perception of multi-scale grain boundary morphology. Then, it is further processed by a Phase Transformation Prediction and Evaluation (PTPE) unit, which integrates materials science theory to establish a material phase transformation sensitivity feature extractor. This extractor includes a three-layer fully connected network and a self-attention layer, capable of accurately capturing the microstructural features of PDC materials under different pressure and temperature conditions. The final network output layer uses a softmax classifier to map features to eight basic grain boundary types.

[0070] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0071] The grain boundary type identification results are classified into straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions, and high-density defect regions. Each grain boundary type is encoded as the input feature of the PDC-Lambda algorithm according to the feature vector, and the grain boundary feature encoding data is obtained.

[0072] The grain boundary feature encoding data is processed using the iteration coefficient calculation module in the PDC-Lambda algorithm, with three evaluation indicators: surface roughness, heat-affected zone width, and microcrack density. The data is then fitted using a quadratic response surface model to obtain the grain boundary response surface model.

[0073] The PDC-Lambda algorithm is applied to the grain boundary response surface model for matrix mapping transformation. The processing parameters are calculated using an iterative optimization strategy. A comprehensive processing quality index is constructed by weighted combination of surface roughness, heat-affected zone width, and microcrack density, and the initial mapping matrix of grain boundary parameter response is obtained.

[0074] The initial mapping matrix of grain boundary parameter response is input into the multidimensional optimization module of the PDC-Lambda algorithm. A hybrid optimization method combining the improved particle swarm optimization algorithm and the simulated annealing algorithm is applied to calculate the optimal combination of processing parameters for each grain boundary type at different relative positions. The results are stored as a three-dimensional array structure to obtain the grain boundary-power response mapping matrix.

[0075] Specifically, the grain boundary types identified by the dual-channel grain boundary feature spectrum enhanced convolutional network are classified and encoded according to their geometric morphology and material properties. Straight grain boundaries are encoded using an eight-dimensional feature vector, including grain boundary length, width, tilt angle, interfacial energy, atomic arrangement regularity, thermal conductivity difference, stress concentration factor, and interfacial bonding strength. Serrated grain boundaries are encoded by adding two more feature parameters, serration depth and serration density, forming a ten-dimensional vector. Curved grain boundaries are encoded by incorporating curvature radius and tortuosity parameters. Embedded grain boundary feature vectors include embedding depth and embedding angle. Multiple-cross grain boundaries encode the number of cross points and the distribution of cross angles. Stepped grain boundaries record step height and step density. Amorphous regions encode the degree of amorphization and diffusion range. High-density defect regions encode defect density and defect type distribution. The iterative coefficient calculation module processes data by establishing mathematical relationships between surface roughness Ra, heat-affected zone width HAZ, microcrack density Dc, and grain boundary feature vectors. Surface roughness is calculated based on the ternary function relationship of laser power P, scanning speed V, and pulse frequency F. The width of the heat-affected zone is calculated using the heat conduction equation to determine the diffusion range of laser energy at the grain boundaries. The microcrack density is calculated based on the stress concentration theory to determine the probability of crack initiation at the grain boundaries. The quadratic response surface model uses a polynomial fitting method to establish the nonlinear relationship between the three evaluation indicators and the processing parameters. During the fitting process, the coefficients are determined using the least squares method to form a three-dimensional response surface.

[0076] During the matrix mapping transformation, the grain boundary response surface model is converted into a matrix form, with each grain boundary type corresponding to a response matrix. The matrix elements represent the quality evaluation index values ​​under a specific combination of processing parameters. The iterative optimization strategy uses the gradient descent method to calculate the optimal parameter combination. The comprehensive processing quality index is constructed through a weighted average, where surface roughness has a weight of 0.4, heat-affected zone width has a weight of 0.35, and microcrack density has a weight of 0.25. The weight allocation is determined based on the degree of influence of each index on the processing quality.

[0077] The multidimensional optimization module employs a hybrid optimization method combining an improved particle swarm optimization (PSO) algorithm and a simulated annealing algorithm to process the initial mapping matrix of grain boundary parameter responses. The PSO algorithm performs a global search of the optimal parameter space, with each particle representing a set of processing parameter combinations. Particles move within the parameter space using velocity and position update formulas to find the optimal solution. The simulated annealing algorithm performs a local fine-grained search, controlling the search range through a temperature parameter. The search range is gradually reduced with each iteration until convergence to the optimal solution. During the optimization process, the differences in parameter requirements for each grain boundary type at different relative positions are considered. The results are stored as a three-dimensional array structure: the first dimension represents the grain boundary type, the second dimension represents the relative position coordinates, and the third dimension represents the corresponding optimal processing parameter combination.

[0078] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0079] The initial power parameters were obtained by adjusting the macroscopic power baseline based on the preparation conditions and overall hardness distribution of the diamond composite sheet.

[0080] The initial power parameters are compensated for by meso-grain boundary transition based on the characteristics of the grain boundary region, and the compensated power parameters are obtained.

[0081] Microscopic pulse shaping is performed based on the compensated power parameters, and the appropriate pulse waveform is selected according to the grain boundary type to obtain the final processing parameters;

[0082] The final processing parameters are applied to the laser processing of diamond composite sheets, and the parameters are dynamically adjusted by real-time monitoring signals.

[0083] Specifically, the three-level power control architecture achieves precise laser power control from macro to micro levels. First, in the macro power baseline adjustment stage, four key material characteristic parameters of the diamond composite sheet are obtained: average Vickers hardness (GPa), relative density (%), preparation temperature (°C), and preparation pressure (GPa). These parameters are input into the power baseline calculation rule, which multiplies the material hardness by 0.187, the relative density by 0.092, the preparation temperature by 0.0024, and the preparation pressure by 0.153, plus a constant term of 21.5 to calculate the baseline power value. Subsequently, correction coefficients are applied to different hardness regions. By consulting a pre-established material characteristic-power requirement correspondence table, the power baseline correction coefficient is determined. Multiplying the correction coefficient by the baseline power yields the initial power parameter. Next, in the meso-grain boundary transition compensation stage, a local heat conduction model is established for each grain boundary type. The temperature and thermal stress distributions in the grain boundary region are calculated under the initial power parameters. The thermal stress distribution calculation needs to consider the product of the material's thermal expansion coefficient and the temperature gradient. Subsequently, the thermal stress level is mapped to the power compensation coefficient range [-0.3, 0.5]. The larger the thermal stress, the smaller the power compensation coefficient to prevent thermal damage. The power compensation coefficient is multiplied by the initial power parameters to obtain the compensated power parameters. Then, in the micro-pulse shaping stage, different pulse waveforms are selected according to the grain boundary type: trapezoidal pulses are used for straight and curved grain boundaries; bimodal pulses are used for serrated and embedded grain boundaries; ramp pulses are used for multiple intersecting grain boundaries and ladder-like grain boundaries; and rectangular pulses are used for amorphous regions and high-density defect regions. Each waveform also has its own parameter adjustment mechanism, such as rise time, fall time, peak time, and peak power. These parameters are determined by querying the grain boundary-power response mapping matrix, ultimately forming a complete processing parameter configuration. During actual processing, the real-time monitoring system continues to use spectral reflectance and acoustic emission signals, with the sampling frequency increased to 100kHz. It calculates the processing stability index, thermal damage risk index, and surface quality prediction index. When the index exceeds the preset threshold, it triggers dynamic parameter adjustment. The adjustment amount is controlled by the parameter change smoothing control algorithm to avoid processing instability caused by sudden changes.

[0084] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0085] The average Vickers hardness, relative density, preparation temperature and preparation pressure of the diamond composite sheet were obtained to acquire basic material property data.

[0086] The power coefficient is calculated based on the basic material properties data. The average Vickers hardness, relative density, preparation temperature and preparation pressure are substituted into the preset power baseline calculation rules to obtain the power baseline calculation results.

[0087] The power baseline calculation results are corrected by comparing the corresponding relationship between material properties and power demand to obtain the correction coefficient.

[0088] The correction coefficients are calculated together with the power baseline to obtain the initial power parameters for laser power control.

[0089] Specifically, macroscopic power baseline adjustment first requires obtaining basic characteristic data of the diamond composite sheet, which directly affects the energy requirements and parameter settings of laser processing. The average Vickers hardness is obtained by indentation testing at multiple test points evenly distributed on the sample surface using a microhardness tester. A specific load is applied to each point and held for a certain time, and the diagonal length of the indentation is measured to calculate the Vickers hardness value. The average value is taken as the overall hardness characteristic. The relative density is determined using the Archimedes displacement method. After immersing the sample in distilled water, its mass in air and water is measured, and the ratio of actual density to theoretical density is calculated. The preparation temperature and preparation pressure are directly obtained from the sample preparation process record. After obtaining these four data points, power coefficients are calculated. A linear weighted model is used to correlate each parameter with the laser power requirement. The specific calculation rule is to multiply the average Vickers hardness by the first weighting coefficient, the relative density by the second weighting coefficient, the preparation temperature by the third weighting coefficient, and the preparation pressure by the fourth weighting coefficient. Then, these four items are added together and a basic power constant is added to obtain the power baseline calculation result. The optimal weighting coefficients were obtained by fitting a large amount of experimental data. Hardness had the largest weight, followed by preparation pressure, then relative density, and preparation temperature had the smallest weight. After calculating the power baseline, a correction process was required because the linear model could not fully capture the complex nonlinear relationship between material properties and power requirements. The correction process was achieved through table lookup and comparison. First, the basic material property data was compared with records in a pre-established empirical database to find the five most similar records. The nearest neighbor weighted average method was used to calculate the correction coefficient, that is, the correction coefficients of the five records were weighted according to their similarity to the properties of the current sample, with higher similarity resulting in higher weights. Finally, the correction coefficients were multiplied by the power baseline calculation result to obtain the initial power parameters used for laser power control. Taking a PDC composite sheet prepared under high pressure and high temperature conditions as an example, its basic property data was measured and then entered into the power baseline calculation rules to obtain the power baseline value. In the correction stage, similar records were found by querying the empirical database, the correction coefficients were calculated, and finally multiplied by the power baseline to obtain the initial power parameters. This macroscopic power baseline adjustment method solves the problem of insufficient consideration of the material characteristics of diamond composite sheets in traditional control methods, and provides a reasonable starting point for power parameters of PDC materials under different preparation conditions, effectively reducing the trial and error process in processing.

[0090] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0091] A heat conduction model was established for the grain boundary region, and the temperature distribution of the grain boundary region under different powers was calculated to obtain the grain boundary temperature distribution data.

[0092] The thermal stress distribution in the grain boundary region is calculated based on the grain boundary temperature distribution data to obtain the grain boundary thermal stress distribution data.

[0093] The power compensation coefficient of the grain boundary region is determined based on the data of grain boundary thermal stress distribution. The power is adjusted by the mapping relationship between the thermal stress level and the compensation coefficient to obtain the power compensation coefficient.

[0094] The power compensation coefficient is combined with the initial power parameters to calculate the power parameters after transition compensation of the grain boundary region.

[0095] Specifically, a heat conduction model was established for the grain boundary region. This model employs a two-dimensional finite difference method, dividing the grain boundary region into 100×100 microgrid cells, each cell measuring 5μm×5μm, to calculate the temperature distribution of the grain boundary region under initial power parameters. The calculation considers thermophysical parameters of diamond material, including thermal conductivity (approximately 1000-2200 W / m·K, varying with temperature), specific heat capacity (approximately 520 J / kg·K), and density (approximately 3.52 g / cm³), as well as laser energy absorptivity and thermal diffusion characteristics. The heat conduction model is iteratively calculated with each time step of 1μs, for a total calculation time of 10ms, obtaining complete grain boundary temperature distribution data. The data records the temperature values ​​of each grid cell at different times, forming a spatiotemporal distribution matrix of the temperature field. Subsequently, the thermal stress distribution is calculated based on the grain boundary temperature distribution data. The thermal stress calculation is based on thermoelastic theory, considering the thermal expansion coefficient of diamond (approximately 1×10⁻⁻⁻⁴). 6The parameters are: K, Young's modulus (approximately 1050 GPa), and Poisson's ratio (approximately 0.2). During calculation, the temperature gradient is converted into thermal stress. For each mesh element, the temperature difference between it and surrounding elements is calculated. Combining the material's thermal expansion coefficient and elastic constant, the thermal stress value of that element is obtained. The thermal stress distribution data also forms a two-dimensional matrix, recording the stress state at each location in the grain boundary region. Based on the thermal stress distribution data, the next step is to determine the power compensation coefficient. A nonlinear mapping relationship is used to convert the thermal stress level into a power adjustment range. First, a thermal stress safety threshold is determined. For PDC materials, this threshold is typically 3.5 GPa. When the thermal stress is below 1 GPa, the area is considered a safe zone, and the compensation coefficient is positive (0.1 to 0.5), appropriately increasing the power to improve processing efficiency. When the thermal stress is in the range of 1-3 GPa, it is a warning zone, and the compensation coefficient is a small negative value (-0.1 to -0.2), slightly reducing the power. When the thermal stress exceeds 3 GPa and approaches the threshold, it is a danger zone, and the compensation coefficient is a large negative value (-0.2 to -0.3), significantly reducing the power to avoid damage. The mapping relationship employs a piecewise linear function, achieving rapid conversion through table lookup. After determining the power compensation coefficient, it is combined with the initial power parameters for calculation: compensated power = initial power × (1 + compensation coefficient). For example, when processing a PDC sample containing multiple grain boundaries, the thermal conduction model calculation shows severe heat accumulation at the grain boundary intersections, with temperatures reaching 1400°C, far exceeding the 800°C of the surrounding area, resulting in thermal stress of 3.2 GPa in this region, approaching the damage threshold. Based on the mapping relationship, this region is assigned a power compensation coefficient of -0.25, while the initial power is 55W. After combined calculation, the compensated power at the grain boundary intersections is 41.25W, significantly reducing the power to avoid the risk of thermal damage. Simultaneously, in the grain boundary transition region, the thermal stress is lower, approximately 1.5 GPa, with a power compensation coefficient of -0.1, resulting in a compensated power of 49.5W. This achieves a smooth power transition from the matrix region to the grain boundary region, effectively solving the problem of microcracks easily generated in the grain boundary region in traditional control methods.

[0096] The intelligent control method for laser processing equipment of diamond composite sheets in the embodiments of this application has been described above. The intelligent control system for laser processing equipment of diamond composite sheets in the embodiments of this application is described below. Please refer to [link / reference]. Figure 2 One embodiment of the intelligent control system for laser processing equipment of diamond composite sheets in this application includes:

[0097] The scanning module 201 is used to scan the diamond composite sheet using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency characteristics of acoustic emission signal;

[0098] Extraction module 202 is used to input the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary recognition, and obtain grain boundary type recognition results;

[0099] Module 203 is used to execute the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary type and processing parameters;

[0100] The control module 204 is used to establish a three-level power control architecture based on the grain boundary-power response mapping matrix, which includes macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping, to control the processing of diamond composite sheets.

[0101] above Figure 2 The intelligent control system for the laser processing equipment of diamond composite sheets in this embodiment of the invention is described in detail from the perspective of modular functional entities. The intelligent control device for the laser processing equipment of diamond composite sheets in this embodiment of the invention is described in detail from the perspective of hardware processing.

[0102] Reference Figure 3 This invention also provides an intelligent control device for a diamond composite sheet laser processing equipment. This intelligent control device can be a server, and its internal structure can be as follows: Figure 3 As shown, the intelligent control device for diamond composite laser processing equipment includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor, designed as a computer, provides computing and control capabilities. The memory of the intelligent control device for diamond composite laser processing equipment includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the intelligent control device for diamond composite laser processing equipment stores the data corresponding to this embodiment. The network interface of the intelligent control device for diamond composite laser processing equipment is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0103] Those skilled in the art will understand that Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the intelligent control device for the diamond composite laser processing equipment to which the present invention is applied.

[0104] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the intelligent control method for a diamond composite laser processing equipment.

[0105] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an intelligent control device (which may be a personal computer, server, or network device, etc.) for laser processing equipment for diamond composite sheets to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0107] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent control method for a diamond compact laser processing apparatus, characterized by, The method includes: A diamond composite sheet was scanned using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum, and time-frequency characteristics of acoustic emission signal. The four-dimensional feature tensor is input into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary identification, and the grain boundary type identification result is obtained. Based on the grain boundary type identification results, the PDC-Lambda algorithm is executed to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters; Based on the aforementioned grain boundary-power response mapping matrix, a three-level power control architecture is established, consisting of macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping, to control the processing of diamond composite sheets.

2. The intelligent control method for the diamond compact laser processing apparatus according to claim 1, wherein, The method employs a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to scan the diamond composite sheet, obtaining a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectra, and time-frequency characteristics of the acoustic emission signal, including: The diamond composite sheet is fixed on the processing platform, and the diamond composite sheet is positioned using a three-dimensional positioning system to obtain the positioned diamond composite sheet. The diamond composite sheet after positioning was surface scanned with a step spacing of 50 μm, and each sampling point was excited by five laser light sources of different wavelengths to obtain reflectance spectrum data; Acoustic emission data is obtained by collecting acoustic signals at each sampling point on the surface of the positioned diamond composite sheet using four symmetrically arranged high-frequency acoustic emission sensors. The reflection spectrum data and the acoustic emission data are processed by median filtering, wavelet denoising and frequency domain filtering to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency features of acoustic emission signal.

3. The intelligent control method for the diamond compact laser processing apparatus according to claim 1, wherein, The step of inputting the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary identification, to obtain grain boundary type identification results, includes: The four-dimensional feature tensor is separated into two parts: spectral information and acoustic emission signal. These parts are then input into the spectral information channel and acoustic emission signal channel of the dual-channel grain boundary feature spectrum enhancement convolutional network, respectively, to obtain the initial feature map. The initial feature maps of the spectral information channels are processed through multiple convolutional layers. Each convolutional layer is followed by a batch normalization layer and a ReLU activation function. The last layer uses one-dimensional convolution to integrate the features and obtain the spectral feature extraction results. The initial feature map of the acoustic emission signal channel is processed by a dual time-frequency domain processing structure. First, the time-domain signal is converted into a spectrum map through short-time Fourier transform, and then the spectrum features are extracted through a multi-layer convolutional network. At the same time, the original time-domain signal is extracted with time-domain features through a one-dimensional convolutional network. The two parts of features are fused through an attention mechanism to obtain the acoustic emission feature extraction result. The spectral feature extraction results and the acoustic emission feature extraction results are integrated by the feature fusion module. The diamond grain boundary morphology spectrum analysis unit with residual connection structure and void convolution and the phase transformation prediction and evaluation unit based on material phase transformation theory are applied for processing to obtain grain boundary type identification results including straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions and high-density defect regions.

4. The intelligent control method for the diamond compact laser processing apparatus according to claim 1, wherein, The step of executing the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters includes: The grain boundary type identification results are classified into straight grain boundaries, serrated grain boundaries, curved grain boundaries, embedded grain boundaries, multiple intersecting grain boundaries, ladder grain boundaries, amorphous regions, and high-density defect regions. Each grain boundary type is encoded as the input feature of the PDC-Lambda algorithm according to the feature vector to obtain grain boundary feature encoding data. The grain boundary feature encoding data is processed using the iteration coefficient calculation module in the PDC-Lambda algorithm, with three evaluation indicators: surface roughness, heat-affected zone width, and microcrack density. The data is then fitted using a quadratic response surface model to obtain the grain boundary response surface model. The PDC-Lambda algorithm is applied to the grain boundary response surface model for matrix mapping transformation. The processing parameters are calculated using an iterative optimization strategy. A comprehensive processing quality index is constructed by weighted combination of surface roughness, heat-affected zone width, and microcrack density, and the initial mapping matrix of grain boundary parameter response is obtained. The initial mapping matrix of the grain boundary parameter response is input into the multidimensional optimization module of the PDC-Lambda algorithm. A hybrid optimization method combining the improved particle swarm optimization algorithm and the simulated annealing algorithm is applied to calculate the optimal combination of processing parameters for each grain boundary type at different relative positions. The results are stored as a three-dimensional array structure to obtain the grain boundary-power response mapping matrix.

5. The intelligent control method for the diamond compact laser processing apparatus according to claim 1, wherein, The three-level power control architecture based on the grain boundary-power response mapping matrix—macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping—is used to control the processing of diamond composite sheets, including: The initial power parameters were obtained by adjusting the macroscopic power baseline based on the preparation conditions and overall hardness distribution of the diamond composite sheet. The initial power parameters are compensated for by meso-grain boundary transition based on the characteristics of the grain boundary region to obtain the compensated power parameters. Microscopic pulse shaping is performed based on the compensated power parameters, and the corresponding pulse waveform is selected according to the grain boundary type to obtain the final processing parameters; The final processing parameters are applied to the laser processing of diamond composite sheets, and the parameters are dynamically adjusted by real-time monitoring signals.

6. The intelligent control method for the diamond compact laser processing apparatus according to claim 5, wherein, The initial power parameters are obtained by adjusting the macroscopic power baseline based on the preparation conditions and overall hardness distribution of the diamond composite sheet, including: The average Vickers hardness, relative density, preparation temperature, and preparation pressure of the diamond composite sheet were obtained to acquire basic material property data. Based on the material's basic property data, the power coefficient is calculated by substituting the average Vickers hardness, relative density, preparation temperature, and preparation pressure into the preset power baseline calculation rules to obtain the power baseline calculation results. The power baseline calculation results are corrected by comparing the corresponding relationship between material properties and power demand to obtain the correction coefficient. The correction coefficient is calculated in conjunction with the power baseline to obtain the initial power parameters for laser power control.

7. The intelligent control method for the diamond compact laser processing apparatus according to claim 5, wherein, The step of performing meso-grain boundary transition compensation on the initial power parameters based on the characteristics of the grain boundary region to obtain the compensated power parameters includes: A heat conduction model was established for the grain boundary region, and the temperature distribution of the grain boundary region under different powers was calculated to obtain the grain boundary temperature distribution data. The thermal stress distribution in the grain boundary region is calculated based on the grain boundary temperature distribution data to obtain the grain boundary thermal stress distribution data. Based on the grain boundary thermal stress distribution data, the power compensation coefficient of the grain boundary region is determined, and the power is adjusted by the mapping relationship between the thermal stress level and the compensation coefficient to obtain the power compensation coefficient. The power compensation coefficient is combined with the initial power parameters to calculate the power parameters after transition compensation of the grain boundary region.

8. An intelligent control system for laser processing equipment of diamond composite sheets, characterized in that, For implementing the intelligent control method for a diamond composite laser processing equipment as described in any one of claims 1-7, the intelligent control system for the diamond composite laser processing equipment comprises: The scanning module is used to scan the diamond composite sheet using a spectral reflectance sensor assembly and an acoustic emission signal acquisition device to obtain a four-dimensional feature tensor containing spatial coordinate information, multi-wavelength reflection intensity spectrum and time-frequency characteristics of acoustic emission signal; The extraction module is used to input the four-dimensional feature tensor into a dual-channel grain boundary feature spectrum enhanced convolutional network for feature extraction and grain boundary identification, and to obtain the grain boundary type identification result. The construction module is used to execute the PDC-Lambda algorithm based on the grain boundary type identification result to construct a grain boundary-power response mapping matrix containing the correspondence between grain boundary types and processing parameters; The control module is used to establish a three-level power control architecture based on the grain boundary-power response mapping matrix, which includes macroscopic power baseline adjustment, mesoscopic grain boundary transition compensation, and microscopic pulse shaping, to control the processing of diamond composite sheets.

9. An intelligent control device for a diamond compact laser processing apparatus, characterized by, The device includes a memory and a processor, the memory storing a computer program that can run on the processor, and the processor executing the computer program to implement the intelligent control method for a diamond composite laser processing equipment as described in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is run by the processor, it causes the processor to execute the intelligent control method for a diamond composite laser processing equipment as described in any one of claims 1 to 7.

Images