Intelligent control system for diamond grain curing and sintering furnace

Abstract

The invention discloses an intelligent control system for a diamond grain curing sintering furnace, belongs to the technical field of industrial furnace control, and aims at solving the problems that existing sintering control depends on experience, process monitoring is not comprehensive, real-time optimization is difficult to achieve, and product quality consistency is difficult to guarantee. The system comprises a sensor module, a multispectral imaging device, a data preprocessing module, a liquid neural network module, an attention fusion module, a user interaction module, a thermal structure coupling module and a real-time feedback control module. A liquid neural network is used for processing time sequence dynamics, multi-dimensional features such as temperature, humidity, spectrum and mechanical force are fused through a multi-head attention mechanism, and in combination with an internal state result of user input and thermal structure coupling analysis, a real-time feedback control module adopts a closed-loop strategy to calculate optimal firing parameters and generate a control signal; and the actuating mechanism is driven to perform self-adaptive adjustment.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial furnace control, and specifically to an intelligent control system for a diamond grain curing and sintering furnace. Background Art

[0002] Due to its excellent properties such as high hardness and high wear resistance, diamond grains are widely used in the manufacture of various tools, abrasives and functional components. Curing and sintering is a key process for preparing high-performance diamond composites. The process is complex and involves multiple physical and chemical changes under high temperature and high pressure. The product quality (such as density, hardness, internal stress, tissue uniformity, etc.) is extremely sensitive to process parameters such as temperature, pressure, holding time, heating and cooling rates during sintering. Traditional sintering furnace control often relies on the operator's experience to set a fixed process curve, or uses simple feedback control based on classical control theories such as PID (Proportional-Integral-Derivative), which is difficult to fully adapt to fluctuations in raw material batches, differences in loading methods, and complex non-linear and time-varying dynamic characteristics during the sintering process, resulting in problems such as low product quality consistency, room for improvement in the finished product rate, and difficulty in optimizing production efficiency and energy consumption. Therefore, it is of great significance to develop a system that can monitor the sintering state in real time and perform intelligent and adaptive control.

[0003] In order to improve the control effect of the sintering process, some explorations have been made on the existing technologies. For example: Chinese Patent No. CN103697686B discloses a diamond grinding block sintering furnace and its sintering process. This system uses multiple temperature and pressure sensors to collect data and processes it through a PID regulator. In particular, it attempts to convert the pressure value into an indirect temperature value using the ideal gas state equation, combines the measured temperature value with the set value for comparison, and calculates the control quantity to adjust the heating circuit. The purpose of this design is to more accurately adjust the furnace temperature and ensure temperature uniformity by multi-point measurement and combining pressure information.

[0004] Chinese Patent No. CN113084718B discloses a forming and sintering process for a metal-bonded diamond grinding head. The focus of this patent is to optimize the preparation process before sintering (segmented forming, fitting assembly) and set a detailed fixed vacuum sintering procedure (heating, holding, cooling in stages and air pressure control) to solve the problems of cracking and holes that are prone to occur after sintering of specific products (grinding heads with a large length-diameter ratio), aiming to improve the yield and reduce costs.

[0005] The above design improves the temperature control accuracy of the sintering process to a certain extent or solves the forming problems of specific products by adopting the method of PID control combined with multi-point temperature and pressure measurement, or by optimizing the pretreatment process and setting a fixed sintering program. However, there are still certain limitations, such as: the perception dimension is single, lacking in-depth insight into the state of the material itself, unable to understand the physical and chemical state changes of the diamond grains in the furnace in real time, such as key information directly affecting the quality of the final product, including composition uniformity, internal stress and strain distribution, microstructure evolution, densification degree, etc. The control strategy is relatively simple, and for complex processes with strong nonlinearity, multivariable coupling, time delay and time-varying characteristics in sintering, the adaptability and control accuracy are limited, lacking real-time adaptive adjustment ability, insufficient process understanding and model fusion, and lacking a mechanism to combine in-depth process understanding such as materials science and physical field simulation with real-time control. Summary of the Invention

[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and propose an intelligent control system for a diamond grain curing and sintering furnace to solve the above problems. The system realizes multi-dimensional real-time perception of the state of diamond grains by integrating sensing technologies such as multi-spectral imaging, uses artificial intelligence models such as liquid neural networks and multi-head attention to process complex time-series and multi-modal data and capture process dynamics, fuses thermal-structure coupling physical models to analyze internal states, and finally realizes precise, adaptive and closed-loop optimization adjustment of sintering parameters through a real-time feedback control module, thereby significantly improving the product quality, consistency, yield and production efficiency of diamond grain sintering.

[0007] The purpose of the present invention is achieved by the following technical solutions: An intelligent control system for a diamond grain curing and sintering furnace, comprising: The sensor module collects data through sensors and multi-spectral cameras to form multi-modal raw data; The data preprocessing module uniformly formats, denoises and aligns the time series of the multi-modal raw data, and outputs standardized data; The liquid neural network module is based on the principle of continuous-time dynamic modeling Processes the standardized data, generates hidden state data by simulating the state change of neurons over time, and uses Taylor expansion and matrix exponential approximation methods to quickly solve the state change equation, and outputs hidden state data; Attention fusion module, which processes the hidden state data. This module has at least four parallel attention heads, where: the first head performs a linear transformation on the temperature data to generate query, key, and value vectors, and uses the attention mechanism to extract temperature features; the second head performs a linear transformation on the humidity data to generate query, key, and value vectors, and uses the attention mechanism to extract humidity features; the third head processes the spectral features reflecting material composition, uniformity, defects, and dryness / humidity extracted from the image; the fourth head processes the mechanical force data; the outputs of each head are fused through concatenation and linear mapping to form unified multimodal information; User interaction module, which receives the preset material ratio and pressing parameters input by the operator through the operation terminal, reads the reference process data from the database, and outputs the user preset parameters; Thermal-structural coupling module, which receives the temperature data and image data collected in real time on-site. After extracting the appearance changes of diamond grit through image analysis, it uses a preset physical model and computer simulation to map the external temperature and appearance changes to the internal stress, strain, and temperature distribution, and outputs the analysis results of the internal structure state; Real-time feedback control module, which receives multimodal information, user preset parameters, and the analysis results of the structural state, calculates the optimal firing parameters using a closed-loop control strategy, converts the calculation results into control signals, and feeds them back to the sintering furnace actuator in real time to achieve automatic adjustment and stable control.

[0008] The sensors in the sensor module specifically collect temperature, humidity, pressure, and mechanical force data. The multispectral camera is fixedly installed inside the sintering furnace and is aligned with the diamond grit to be sintered.

[0009] The data preprocessing module also provides some preprocessed data to the user interaction module for comparison with the user input information.

[0010] The liquid neural network module structurally includes a data input unit, a continuous state modeling unit, and a closed-form solution unit. The continuous state modeling unit is responsible for simulating the evolution of the neuron state over time, and the closed-form solution unit is responsible for quickly approximately solving the state change equation.

[0011] Each parallel attention head in the attention fusion module uses an independent linear transformation unit to generate its query, key, and value vectors respectively, and the specific fusion method of the outputs of each head is to concatenate the output vectors of each head, and then generate unified multimodal information through a linear mapping layer.

[0012] The user interaction module is also configured to automatically compare the actual material composition, uniformity, defects, and dryness / humidity information detected and analyzed by the multispectral camera with the preset material ratio input by the operator, and when the deviation between the two exceeds the preset threshold, an alarm signal is sent through the operation terminal.

[0013] The thermo-structural coupling module structurally includes a data conversion unit, a physical modeling unit, and a simulation analysis unit. The data conversion unit is responsible for converting real-time temperature data and extracted appearance change features into an input format suitable for the physical model. The physical modeling unit performs mapping based on a preset coupling model, and the simulation analysis unit is responsible for completing the calculation of the internal state.

[0014] The preset physical model used in the thermo-structural coupling module and computer simulation can dynamically and relatively accurately reflect the distribution of internal stress, strain, and temperature within diamond grits during different stages of the sintering process, providing decision-making support for the real-time feedback control module.

[0015] The real-time feedback control module structurally includes a decision calculation unit and a control signal generation unit. The decision calculation unit is responsible for executing a closed-loop control strategy to calculate the optimal firing parameters. The control signal generation unit is responsible for converting the calculated parameters into control signals compatible with the sintering furnace actuator and implementing adjustment logic including anomaly compensation.

[0016] It also includes an online learning module configured to receive real-time feedback data and control state data generated during the operation of the system, adaptively update the network weights and dynamic parameters in the liquid neural network module and the attention fusion module, and continuously optimize the accuracy and response speed of the control model to adapt to different sintering conditions.

[0017] The beneficial effects of the present invention are as follows: 1. By integrating a variety of traditional sensors (monitoring temperature, pressure, humidity, mechanical force, etc.) and multi-spectral imaging technology, the system can obtain multi-dimensional and high-resolution real-time process information. It can not only monitor macroscopic process parameters but also finely capture subtle changes in the material composition, distribution uniformity, surface defects, dryness and humidity state, spatial temperature distribution, appearance geometric morphology evolution, and optical characteristics related to the crystal structure of diamond grits during the sintering process. Based on the physical principle of thermo-structural coupling modeling and simulation analysis capabilities, it can map and correlate easily measurable external apparent data (such as surface temperature, appearance changes) with key internal states that are difficult to directly measure (such as internal stress field, strain distribution, internal temperature gradient), enabling the system to focus on the internal physical and chemical changes during the sintering process and deepen the real-time understanding of the complex sintering mechanism.

[0018] 2. By adopting a liquid neural network, it can effectively capture and simulate the complex dynamic characteristics of the sintering process itself, which are strongly nonlinear, time-varying continuous, and long-term dependent. Using the closed-form solution approximation to handle the state evolution equation of the neural network ensures that the complex dynamic model can operate at a speed that meets the requirements of real-time control. The attention mechanism with multiple parallel processing paths is integrated, which can intelligently fuse multi-modal information flows from different sensors, multi-spectral image analysis, physical models, and user inputs. It can dynamically evaluate and weight the importance of different information sources according to the current sintering stage and state, ensuring that the most critical and relevant features play a dominant role in the system decision-making at the current moment, and avoiding information redundancy or key information being submerged.

[0019] 3. A real-time feedback closed-loop control loop is constructed, which can continuously compare the comprehensive evaluation results reflecting the actual sintering state with the preset target or dynamically optimized trajectory. The control decision module can calculate the optimal control parameter adjustment amount (such as heating / cooling rate, holding time, target pressure, pressing rate, etc.) under the current working conditions according to the control strategy. The control instruction is real-time converted into an operation signal for the actuators such as the heating system and pressure system of the sintering furnace, realizing precise, continuous or high-frequency periodic dynamic regulation of the sintering process. It can automatically compensate for the small differences between raw material batches, the fluctuations of environmental conditions, and the unexpected disturbances that may occur during the process, keeping the sintering process always on the optimal or near-optimal path.

[0020] 4. Through process optimization, it can improve the quality consistency of the final sintered diamond products, obtain a more uniform internal structure, higher density, ideal hardness, and fewer internal defects. Automatic detection and verification of the materials before sintering can effectively identify unqualified raw materials and avoid ineffective production. At the same time, the adaptive adjustment ability during the process reduces the waste products caused by process fluctuations, thereby improving the yield and material utilization rate. By controlling the energy input and process time, redundant heating or over-long holding is avoided, optimizing energy consumption, shortening the production cycle, and reducing production costs.

[0021] 5. Improving the automation level of the sintering process reduces the dependence on the experience of operators and the burden of manual monitoring and parameter adjustment. The dynamic neural network and attention module can continuously self-adjust and optimize using real-time feedback data during the operation of the system, enabling the system to long-term adapt to slowly changing working conditions (such as equipment wear, seasonal environmental changes, and fine-tuning of raw material supply), continuously improving the control accuracy and response characteristics, maintaining the optimal performance, and reducing the frequency and cost of offline model maintenance and retraining. Description of the Drawings

[0022] Figure 1 It is the system interaction diagram of the present invention; Figure 2 This is the system architecture diagram of the present invention; Figure 3 This is the timing diagram of the present invention.

[0023] Explanation of the reference numerals in the figure 1. Sensor module; 2. Data preprocessing module; 3. Liquid neural network module; 4. Attention fusion module; 5. User interaction module; 6. Thermal-structure coupling module; 7. Real-time feedback control module; 8. Online learning module; 101. Multimodal raw data; 102. Normalized data; 103. Hidden state data; 104. Multimodal information; 105. User preset parameters; 106. Structural state analysis results; 107. Control signal. Specific implementation manners

[0024] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present invention.

[0025] It should be noted that the orientation concepts of "left", "right", "up", "down", "front", "rear", "inner", and "outer" in the following solutions are all relative directions, and will not be listed one by one here.

[0026] Embodiment 1: As Figures 1 to 3 shown, this embodiment discloses an intelligent control system for a diamond diamond grain curing and sintering furnace, which realizes real-time acquisition, preprocessing, timing modeling, information fusion, and closed-loop control of multimodal data.

[0027] This system first collects various operating data in the sintering furnace through the sensor module 1. The sensor module 1 includes multiple on-site sensors for collecting temperature, humidity, pressure, and mechanical force data, and also includes a multispectral camera fixedly installed inside the sintering furnace, whose field of view is precisely aligned with the diamond diamond grains to be sintered to collect image data covering the visible light and near-infrared bands. All the collected data are integrated by the sensor module 1 to form multimodal raw data 101 and output to the data preprocessing module 2.

[0028] After receiving the multimodal raw data 101 from the sensor module 1, the data preprocessing module 2 first performs a unified format conversion on the data from different sources; secondly, it uses algorithms such as filtering to perform noise reduction processing on the raw data; finally, it precisely aligns the data according to the timestamps of each data. After processing, the standardized data 102 output by the data preprocessing module 2 not only has a regular data format and a high signal-to-noise ratio, but also meets the requirements of the liquid neural network module 3 for the continuity and timeliness of the input data. In addition, some of the preprocessed data will be transmitted to the user interaction module 5 for process parameter comparison.

[0029] The liquid neural network module 3 is used to perform a temporal dynamic modeling on the standardized data 102. This module is internally composed of a data input unit, a continuous state modeling unit, and a closed-form solution unit. The data input unit receives the standardized data 102; the continuous state modeling unit, based on the continuous-time dynamic modeling principle, simulates the process of the neuron state changing over time, so as to capture the complex dynamics during the sintering process; the closed-form solution unit uses the Taylor expansion and matrix exponential approximation methods to quickly approximate and solve the differential equation describing the state change. The hidden state data 103 output after the processing of the liquid neural network module 3 not only records the historical information of the process, but also can reflect the current dynamics in real time, providing high-quality feature data for subsequent processing.

[0030] The dynamic modeling formula of the liquid neural network module 3 is as follows: Dynamic coupling of time constant and gain In the traditional liquid neural network, the time constant τ is a single learnable parameter. Here, we introduce the double dynamic coupling time constant τ(t) and gain γ(t) that both depend on the input x(t): is a learnable weight matrix; is a learnable scalar; State evolution differential equation: Add a quadratic coupling term θ(t) h(t)⊙h(t) to enhance the non-linear expression ability: is a non-linear activation; is the dynamic quadratic coupling coefficient; ⊙ represents element-wise multiplication; Closed-form discrete update: Let the discrete step size be Δt, and using the closed-form approximation of the dynamic coupling time constant, the state update formula can be obtained: (can be approximated by approximate) and Step weights for primary and secondary coupling respectively This formula combines three contributing terms: first-order decay, adaptive gain drive, and secondary coupling, achieving efficient and expressive dynamic modeling.

[0031] The hidden state data 103 is passed to the attention fusion module 4 for intelligent feature fusion. There are at least four parallel attention heads designed inside this module: The first attention head performs a linear transformation on the temperature data to generate query, key, and value vectors, and uses the scaled dot-product attention mechanism to extract temperature features; The second attention head performs a linear transformation on the humidity data to generate query, key, and value vectors, and uses the scaled dot-product attention mechanism to extract humidity features; The third attention head extracts spectral features from the image data that reflect material composition, uniformity, defects, and dryness / humidity; The fourth attention head processes the mechanical force data. After the output vectors of each attention head are concatenated in series, they are then fused through a linear mapping layer to form a unified multi-modal information 104. This multi-modal information 104 comprehensively reflects the important features of the four channels of temperature, humidity, spectrum, and mechanical force, and becomes the key basis for subsequent decision-making.

[0032] Multi-head fusion formula of the attention fusion module 4: Head-wise vector generation extracts different subspace features from the hidden state hn for each head i ∈ {1, 2, 3, 4} respectively to generate query, key, and value vectors: Where each group is a learnable parameter Scaled similarity and adaptive weights: Different from traditional softmax, here we first normalize and scale the similarity, and then introduce the dynamic head weight λi: Prevent division by zero; is the head-specific weight vector, used to adaptively adjust the importance of each head according to the current state; Weighted output, the output of each head combines the scaled similarity and the dynamic weight: , [[ID=--44]], means first perform softmax on and then left-multiply by .

[0033] Concatenation and remapping: Finally, concatenate the outputs of each head and obtain the unified multimodal information Mn through the remapping layer: is the learnable linear mapping parameter, represents vector concatenation; The user interaction module 5 serves as the interface between the system and the operator. It receives the preset material ratio and pressing parameters input by the operator through an operation terminal (such as a touch screen or a computer interface), and can also read reference process data from the internal database. After integrating this information, it outputs the user preset parameter 105 and transmits it to the real-time feedback control module 7 for subsequent decision-making.

[0034] The thermal-structural coupling module 6 is used to map the measurable external data to the internal state of the diamond grits. The thermal-structural coupling module 6 first receives the temperature data collected in real time and the image data from the multispectral camera; subsequently, it extracts the appearance change characteristics of the diamond grits during the sintering process through image analysis; finally, using a preset thermal-structural coupling model based on physical principles and computer simulation methods, it maps the external temperature and appearance changes to the internal stress, strain, and temperature distributions. The internal structure state analysis result 106 output by this module provides an important physical basis for control decisions.

[0035] The thermal-structural coupling dynamic formula is as follows: Let the state vector be the strain tensor; be the stress tensor; be the temperature scalar; Then the entire thermal-structural coupling system can be written as: The descriptions of each item are as follows: Elastic + thermo-mechanical coupling (strain rate) Stress evolution Temperature evolution Stress relaxation coefficient Base elastic stiffness Thermal softening coefficient Base thermal conductivity Strain-dependent thermal conductivity amplification coefficient Mechanical dissipation coupling coefficient Reference temperature The real-time feedback control module 7 is the decision-making and execution center of the system. This module includes a decision-making calculation unit and a control signal generation unit. The decision-making calculation unit receives multi-modal information 104 from the attention fusion module 4, user preset parameters 105 from the user interaction module 5, and the internal structure state analysis result 106 from the thermal-structure coupling module 6. It calculates the optimal firing parameters using a closed-loop control strategy (such as model predictive control, PID control, etc.). The control signal generation unit converts the optimal parameters into a control signal 107, which is fed back to the actuators of the sintering furnace (such as heaters and pressure systems) in real time, thus achieving automatic adjustment and stable control.

[0036] Working process After the system starts, it conducts self-check. The operator inputs the preset material ratio and pressing parameters through the user interaction module 5. The user interaction module 5 forms the user preset parameters 105 with this information and simultaneously reads the reference process data.

[0037] After the raw materials are put in, the sensor module 1 continuously collects on-site sensor data and multi-spectral camera images to form multi-modal raw data 101.

[0038] After the multi-modal raw data 101 is unified in format, denoised, and time-sequentially aligned by the data preprocessing module 2, standardized data 102 is output; part of the data is simultaneously transmitted to the user interaction module 5 for parameter comparison.

[0039] The standardized data 102 is input into the liquid neural network module 3, and under the collaborative action of the data input unit, the continuous state modeling unit, and the closed-form solution unit, hidden state data 103 is generated in real time.

[0040] The hidden state data 103 is transmitted to the attention fusion module 4. The four attention heads of the attention fusion module 4 perform feature extraction and weighted fusion from four dimensions: temperature, humidity, image spectrum, and mechanical force, and finally form unified multi-modal information 104.

[0041] At the same time, the thermal-structure coupling module 6 receives real-time temperature data and image data. After extracting the appearance change features through image analysis, through a preset physical model and computer simulation, the data is mapped into internal stress, strain, and temperature distribution, and the internal structure state analysis result 106 is output.

[0042] The real-time feedback control module 7 aggregates the multimodal information 104, user preset parameters 105, and the internal structure state analysis result 106, calculates the optimal firing parameters using a closed-loop control strategy, converts the parameters into a control signal 107 through the control signal generation unit, and sends it to the sintering furnace actuator in real time to adjust the heater, pressure system, etc.

[0043] The system continuously performs data acquisition, preprocessing, modeling, feature fusion, state mapping, and decision control to ensure that it can respond to on-site dynamic changes at any time until the sintering process ends.

[0044] The sensor module 1 collects temperature, humidity, pressure, mechanical force, and multispectral image data, enabling the system to comprehensively and real-time obtain the multimodal information in the sintering furnace. The multimodal raw data 101 can truly reflect the operating state inside the furnace.

[0045] The data preprocessing module 2 performs unified formatting, noise reduction, and time series alignment on the multimodal raw data 101. The output standardized data 102 ensures data continuity and time series accuracy, provides high-quality input for the liquid neural network module 3, and further improves the reliability of the hidden state data 103.

[0046] The liquid neural network module 3 uses continuous-time dynamic modeling and closed-form solution techniques, can capture the complex dynamics during the sintering process in real time with low memory occupancy and high-speed response characteristics, lays a solid foundation for subsequent feature fusion, and the output hidden state data 103 reflects the process history and current dynamics.

[0047] The multi-head attention mechanism in the attention fusion module 4 can independently weight the temperature, humidity, image spectrum, and mechanical force data respectively, and finally form a unified multimodal information 104 through concatenation and linear mapping. This enables the system to intelligently and dynamically adjust the information weights of each channel, thereby achieving a more accurate state description.

[0048] The real-time feedback control module 7 synthesizes the multimodal information 104 output by the attention fusion module 4, the user preset parameters 105 of the user interaction module 5, and the internal structure state analysis result 106 of the thermal structure coupling module 6, calculates the optimal firing parameters using a closed-loop control strategy, and converts the result into a control signal 107 to achieve precise instructions for the sintering furnace actuator, thereby achieving the purpose of automatic adjustment and abnormal compensation, and ensuring process stability and product quality consistency.

[0049] This embodiment realizes the integrated integration from data acquisition, preprocessing, state modeling, feature fusion to closed-loop control. The systematic automatic control can reduce human intervention and improve production efficiency and process stability.

[0050] In summary, in Embodiment 1, through the collaborative work of the sensor module 1, data preprocessing module 2, liquid neural network module 3, attention fusion module 4, user interaction module 5, thermal-structure coupling module 6, and real-time feedback control module 7, the real-time acquisition, processing, modeling, fusion, and closed-loop control of multimodal data in the sintering furnace are realized, significantly improving the monitoring accuracy and control effect of the sintering process, and having an obvious promoting effect on product quality and process stability.

[0051] Embodiment 2: As Figures 1 to 3 shown, this embodiment adopts the core intelligent control system architecture in Embodiment 1, namely the sensor module 1, data preprocessing module 2, liquid neural network module 3, attention fusion module 4, user interaction module 5, thermal-structure coupling module 6, and real-time feedback control module 7. At the same time, the functions of the user interaction module 5 and the thermal-structure coupling module 6 are further enhanced and refined.

[0052] In this embodiment, the user interaction module 5 not only receives the preset material ratio and pressing parameters input by the operator through the operation terminal to form the user preset parameter 105. At the same time, this module also has an automatic material verification and alarm function, and a comparison logic unit is configured inside for the following work: Receive part of the preprocessed data output from the data preprocessing module 2 (for example, the initially detected material characteristic data); Receive the actual material information collected by the multispectral camera through the sensor module 1 and preprocessed by the data preprocessing module 2, including data such as the material composition, distribution uniformity, surface defects, and dryness of the diamond grits; Automatically compare the actual material information with the material ratio preset by the operator, and judge the deviation according to the preset safety or process threshold; If the comparison result shows that the deviation exceeds the threshold, trigger the alarm function, and inform the operator to confirm or process through the display interface or voice prompt; At the same time, output the verified user preset parameter 105 to the real-time feedback control module 7.

[0053] In this embodiment, the thermal-structure coupling module 6 is further refined into three cooperating sub-units on the basis of Embodiment 1 to enhance the analysis ability of the internal physical state during the sintering process: The data conversion unit receives the temperature data collected in real time on-site from the sensor module 1 and the appearance change characteristics of the diamond grits extracted by image analysis (such as size shrinkage rate, surface roughness change, etc.), and converts these data into a standardized data format that matches the requirements of physical modeling.

[0054] The physical modeling unit incorporates a preset model based on the thermal-structural coupling principle, which describes the interaction between diamond grains, grain boundary migration, deformation diffusion, and interaction with the binder under high temperature and high pressure environments. This unit uses the standardized data 102 output by the data conversion unit as the model input to calculate the internal stress distribution, strain field, temperature field, etc. within the diamond grains.

[0055] The simulation analysis unit uses the preliminary results mapped by the physical modeling unit and performs numerical calculations and simulations through finite element or finite difference methods to obtain a more detailed and accurate analysis result 106 of the internal structure state, which can dynamically reflect the changes in the internal state at each stage of the sintering process.

[0056] Except for the above enhanced parts, the structures and functions of the sensor module 1, data preprocessing module 2, liquid neural network module 3, attention fusion module 4, and real-time feedback control module 7 are the same as those in Embodiment 1. However, they can now fully utilize the richer and more accurate information provided by the enhanced user interaction module 5 and thermal structure coupling module 6 to form higher-quality multimodal information 104 and analysis result 106 of the internal structure state, thereby supporting more precise control decisions.

[0057] Working process Based on Embodiment 1, the working process of this embodiment adds the links of material verification and enhanced physical state analysis. The specific steps are as follows: In the initialization stage, after the system self-check is completed, the operator inputs the preset material ratio and pressing parameters through the user interaction module 5 on the operation terminal to form the user preset parameters 105.

[0058] Meanwhile, the multispectral camera captures the initial material image, which is processed by the data preprocessing module 2 to generate the standardized data 102, and part of the information about the material characteristics in the preprocessed data is transmitted to the user interaction module 5.

[0059] The comparison logic unit in the user interaction module 5 automatically compares the material ratio preset by the operator with the actually detected material composition, uniformity, defects, and dryness. If the deviation between the detection result and the preset information exceeds the safety threshold, the alarm function is automatically triggered, and the operator is confirmed or adjusted through the display interface or voice prompt to ensure that the input raw materials meet the requirements.

[0060] The links of data acquisition, preprocessing, liquid neural network processing, and attention fusion operate according to the working process of Embodiment 1 to generate unified multimodal information 104.

[0061] The thermal-structural coupling module 6 simultaneously receives on-site real-time temperature data and image data, and the data conversion unit converts this data into a format suitable for physical modeling; the physical modeling unit performs state mapping based on a preset thermal-structural coupling model; the simulation analysis unit calculates the detailed distributions of internal stress, strain, and temperature fields, and outputs the internal structure state analysis result 106.

[0062] The dynamic simulation results of the thermal-structural coupling module 6 can reflect the internal state changes during different stages of the sintering process, such as heating, holding, pressurizing, pressure holding, and cooling, in real time.

[0063] The real-time feedback control module 7 receives the multi-modal information 104 from the attention fusion module 4, the user preset parameters 105 from the user interaction module 5, and the internal structure state analysis result 106 from the thermal-structural coupling module 6.

[0064] The control module calculates the optimal firing parameters for the current or next stage through a closed-loop control strategy and converts them into a control signal 107.

[0065] The control signal 107 is fed back to the sintering furnace actuator (including the heater controller, pressure controller, etc.) in real time, thereby realizing the automatic adjustment and stable control of the sintering process.

[0066] The system continuously and circularly executes data acquisition, state modeling, physical analysis, and control decision-making until the sintering process ends.

[0067] Through the enhanced user interaction module 5, the system automatically compares the preset material ratio input by the operator with the actual material detection data before sintering. If the deviation exceeds the standard, an alarm signal is sent in a timely manner, thereby identifying raw material abnormalities in advance and reducing production risks and quality degradation caused by unqualified raw materials.

[0068] The thermal-structural coupling module 6 is refined into a data conversion unit, a physical modeling unit, and a simulation analysis unit. The high-precision thermal-structural coupling model combines computer simulation technology to dynamically reflect the internal stress, strain, and temperature distributions inside diamond grains during different sintering stages, fundamentally revealing the physical evolution mechanism during the sintering process and providing a more scientific basis for control decision-making.

[0069] The real-time feedback control module 7 performs more accurate closed-loop calculations on the control strategy by leveraging the enhanced internal structure state analysis result 106 and multi-dimensional fusion multi-modal information 104, thereby quickly responding to raw material and environmental changes during the sintering process, effectively suppressing the occurrence of process abnormalities, and improving the consistency of finished products and the stability of quality.

[0070] This embodiment realizes the automatic integration of the whole process from material verification, in-depth internal state analysis to control decision-making, reduces the need for manual intervention, improves production efficiency and system security, and at the same time provides rich analysis basis for process monitoring and fault diagnosis.

[0071] In summary, through the enhancement of the functions of the user interaction module 5 and the thermal structure coupling module 6 in Embodiment 2, not only the raw material quality control and warning capabilities are improved, but also more accurate information is provided for real-time feedback control through in-depth physical state analysis, enabling the entire sintering process to achieve automatic closed-loop control with higher precision and stability, which has a significant improvement effect on improving product quality and process reliability.

[0072] Embodiment 3: As Figures 1 to 3 shown, in this embodiment, the online learning module 8 is further integrated on the basis of the core intelligent control system of Embodiment 1 or Embodiment 2, enabling the system to have the ability of continuous self-adaptation and dynamic optimization.

[0073] The sensor module 1 collects temperature, humidity, pressure and mechanical force data from on-site sensors, and at the same time collects image data through a multi-spectral camera fixed in the sintering furnace and aligned with the diamond grains to be sintered, forming multi-modal raw data 101.

[0074] The data preprocessing module 2 performs unified formatting, noise reduction and time series alignment processing on the multi-modal raw data 101 output by the sensor module 1, and outputs standardized data 102. Part of the data is also transmitted to the user interaction module 5 as a comparison basis.

[0075] The liquid neural network module 3 processes the standardized data 102 using the principle of continuous-time dynamic modeling. The liquid neural network module 3 includes a data input unit, a continuous state modeling unit and a closed-form solution unit. Among them, the continuous state modeling unit simulates the change of neuron state over time, and the closed-form solution unit uses the Taylor expansion and matrix exponential approximation method to quickly solve the state change equation, and outputs hidden state data 103.

[0076] The attention fusion module 4 has at least four parallel attention heads, which process the temperature, humidity, spectral features extracted from the image (reflecting material composition, uniformity, defects and dryness) and mechanical force data respectively. Each head generates query, key and value vectors using independent linear transformations. After calculating their respective features through the attention mechanism, they are fused through a concatenation and linear mapping layer to form unified multi-modal information 104.

[0077] The user interaction module 5 allows the user to input preset material ratios and pressing parameters through an operation terminal. At the same time, the user interaction module 5 can read reference process data from the database. The user interaction module 5 integrates and outputs the user's preset parameters 105, and automatically compares them with the preset parameters based on part of the initial detection data from the data preprocessing module 2. When the preset threshold is reached, an alarm is triggered.

[0078] The thermal-structural coupling module 6 receives the real-time temperature data and image data from the sensor module 1, extracts the appearance change characteristics of diamond grits (such as size shrinkage, surface roughness, etc.) through image analysis, and maps the external data to the internal stress, strain, and temperature distribution through a preset thermal-structural coupling model and computer simulation, and outputs the internal structure state analysis result 106.

[0079] The real-time feedback control module 7 synthesizes the multi-modal information 104 of the attention fusion module 4, the user's preset parameters 105 of the user interaction module 5, and the internal structure state analysis result 106 of the thermal-structural coupling module 6, calculates the optimal firing parameters using a closed-loop control strategy, and converts them into a control signal 107, which is fed back to the sintering furnace actuator in real time to achieve automatic control and anomaly compensation.

[0080] The online learning module 8 is the core enhancement component of this embodiment. The online learning module 8 can receive real-time feedback data and status data from various parts of the system, mainly including: Feedback information from the real-time feedback control module 7 (for example, the deviation between the target firing parameters and the actual execution parameters, the dynamic change of the control signal 107); Intermediate state and output data from the liquid neural network module 3 and the attention fusion module 4 (for example, the comparison information between the hidden state data 103 and the actual measurement value); Key operating condition data from the sensor module 1 or the data preprocessing module 2, and (if detectable) the final product quality evaluation data.

[0081] The online learning module 8 internally integrates online learning algorithms such as online gradient descent, Kalman filtering, or reinforcement learning, calculates the adjustment amount of the weights and dynamic parameters in the liquid neural network module 3 and the attention fusion module 4 using a preset objective function (such as minimizing the control error, improving product quality), and after being processed by the online learning module 8, updates the parameters and feeds them back to the liquid neural network module 3 and the attention fusion module 4 in real time, thereby dynamically adjusting the response characteristics of these core models.

[0082] Working process The specific steps of the working process of this embodiment are as follows: The sensor module 1 continuously collects on-site temperature, humidity, pressure, mechanical force, and image data to form multimodal raw data 101, which is processed by the data preprocessing module 2 into standardized data 102.

[0083] The standardized data 102 is processed in real time by the liquid neural network module 3 to generate hidden state data 103; the hidden state data 103 enters the attention fusion module 4, and each attention head extracts temperature, humidity, image spectrum, and mechanical force features and then fuses them into multimodal information 104.

[0084] Meanwhile, the user interaction module 5 receives the operator's input and outputs user preset parameters 105; the thermal-structural coupling module 6 outputs the internal structure state analysis result 106 based on the real-time temperature and image data; the real-time feedback control module 7 synthesizes the multimodal information 104, user preset parameters 105, and structure state analysis result 106, calculates the optimal firing parameters through a closed-loop control strategy, and generates a control signal 107 to feedback to the actuator to achieve real-time adjustment of the sintering process.

[0085] The online learning module 8 regularly or in real time collects the control error information of the real-time feedback control module 7, hidden state data 103, attention fusion intermediate output, and key environmental data while the main control loop is running.

[0086] The online learning module 8 calculates the adjustment amounts of the weights and dynamic parameters in the liquid neural network module 3 and the attention fusion module 4 according to the collected data and the preset learning objectives through an online learning algorithm.

[0087] The calculated update amounts are fed back to the liquid neural network module 3 and the attention fusion module 4 in real time or periodically, enabling these core modules to have better dynamic response capabilities when processing input data.

[0088] This online learning process runs in parallel with the main control loop to achieve continuous self-adaptation and optimization of the model.

[0089] The online learning module 8 enables the liquid neural network module 3 and the attention fusion module 4 to dynamically adjust the core parameters according to the real-time feedback, so as to adapt to the working condition changes caused by raw material batches, environmental changes, or equipment aging, and maintain the best control performance of the system under various conditions.

[0090] The system continuously learns from the real-time feedback data to achieve online updates of the model weights and dynamic parameters. As the running time increases, the control accuracy, response speed, and robustness of the system are gradually improved, achieving the effect of "getting better with use".

[0091] The online learning mechanism reduces the need for a large amount of offline training data and periodic manual intervention, enabling the system to automatically adapt to environmental changes and reducing the costs of long-term model maintenance and update.

[0092] Benefiting from more optimized liquid neural network modeling and more dynamic multimodal information fusion, the real-time feedback control module 7 can generate more accurate control signals 107, ensuring that the sintering process parameters are maintained at the optimal state and improving product consistency and quality.

[0093] The online learning module 8 can also provide long-term data accumulation for the system, helping to identify and diagnose abnormal working conditions, timely feedback potential problems, and further ensuring production safety and process stability.

[0094] In summary, based on Embodiment 1 and Embodiment 2, Embodiment 3 realizes the continuous self-learning of the system and the real-time optimization of the model through the online learning module 8, overcomes the shortcoming of the limited adaptability of traditional pre-trained models to environmental changes, makes the entire intelligent control system have higher self-adaptability, continuous optimization ability and control accuracy, and has significant advantages in improving the stability of the sintering process and product quality.

[0095] The above is only the preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be changed within the scope of the concept described herein through the above teachings or the technology or knowledge in related fields. As long as the changes and variations made by those skilled in the art do not depart from the spirit and scope of the present invention, they should all be within the protection scope of the appended claims of the present invention.

Claims

1. An intelligent control system for a diamond diamond particle solidification and sintering furnace, characterized in that, Including: The sensor module (1) collects data through sensors and a multi-spectral camera to form multi-modal raw data (101); The data preprocessing module (2) uniformly formats, denoises, and aligns the multi-modal raw data (101) in time sequence, and outputs standardized data (102); The liquid neural network module (3) uses the principle of continuous-time dynamic modeling to process the standardized data (102), generates hidden state data by simulating the change of neuron states over time, and adopts the Taylor expansion and matrix exponential approximation methods to quickly solve the state change equation, and outputs hidden state data (103); The attention fusion module (4) processes the hidden state data (103). The module is provided with at least four parallel attention heads, where: the first head performs a linear transformation on the temperature data to generate query, key, and value vectors, and uses the attention mechanism to extract temperature features; the second head performs a linear transformation on the humidity data to generate query, key, and value vectors, and uses the attention mechanism to extract humidity features; the third head processes the spectral features reflecting material composition, uniformity, defects, and dryness / wetness extracted from the image; the fourth head processes the mechanical force data; the outputs of each head are fused through concatenation and linear mapping to form unified multi-modal information (104); The user interaction module (5) receives the preset material ratio and pressing parameters input by the operator through the operation terminal, reads the reference process data from the database, and outputs user preset parameters (105); The thermal-structure coupling module (6) receives the temperature data and image data collected in real time on-site. After extracting the appearance changes of the diamond grits through image analysis, it uses a preset physical model and computer simulation to map the external temperature and appearance changes to the internal stress, strain, and temperature distribution, and outputs the internal structure state analysis result (106); The real-time feedback control module (7) receives the multi-modal information (104), user preset parameters (105), and the structure state analysis result (106), calculates the optimal firing parameters using a closed-loop control strategy, converts the calculation result into a control signal (107), and feeds it back to the sintering furnace actuator in real time to achieve automatic adjustment and stable control.

2. The intelligent control system of a diamond diamond particle curing and sintering furnace according to claim 1, characterized in that: The sensors in the sensor module (1) specifically collect temperature, humidity, pressure, and mechanical force data, and the multi-spectral camera is fixedly installed inside the sintering furnace and aligned with the diamond grits to be sintered.

3. An intelligent control system for a diamond diamond particle curing and sintering furnace according to claim 1, characterized in that: The data preprocessing module (2) also provides some preprocessed data to the user interaction module (5) for comparison with the user input information.

4. An intelligent control system for a diamond diamond grain solidification sintering furnace according to claim 1, characterized in that: The liquid neural network module (3) structurally includes a data input unit, a continuous state modeling unit, and a closed-form solution unit. The continuous state modeling unit is responsible for simulating the evolution of neuron states over time, and the closed-form solution unit is responsible for quickly approximating the solution of the state change equation.

5. An intelligent control system for a diamond diamond particle curing and sintering furnace according to claim 1, characterized in that: Each parallel attention head in the attention fusion module (4) generates its query, key, and value vectors using independent linear transformation units respectively, and the fusion method of the outputs of each head is specifically to concatenate the output vectors of each head, and then generate the unified multi-modal information (104) through a linear mapping layer.

6. The intelligent control system of a diamond and diamond particle curing and sintering furnace according to claim 1, characterized in that: The user interaction module (5) is further configured to automatically compare the actual material composition, uniformity, defects, and dryness information detected and analyzed by the multispectral camera with the preset material ratio input by the operator, and when the deviation between the two exceeds a preset threshold, an alarm signal is sent through the operation terminal.

7. The intelligent control system of a diamond diamond particle curing and sintering furnace according to claim 1, characterized in that: The thermal-structure coupling module (6) structurally includes a data conversion unit, a physical modeling unit, and a simulation analysis unit. The data conversion unit is responsible for converting the real-time temperature data and the extracted appearance change features into an input format suitable for the physical model. The physical modeling unit performs mapping based on a preset coupling model, and the simulation analysis unit is responsible for completing the calculation of the internal state.

8. An intelligent control system for a diamond diamond particle curing and sintering furnace according to claim 1, characterized in that: The preset physical model and computer simulation used in the thermal-structure coupling module (6) can dynamically and relatively accurately reflect the distribution of internal stress, strain, and temperature in the diamond grit during different stages of the sintering process, providing decision support for the real-time feedback control module (7).

9. The intelligent control system of a diamond and diamond particle solidification and sintering furnace according to claim 1, characterized in that: The real-time feedback control module (7) structurally includes a decision calculation unit and a control signal generation unit. The decision calculation unit is responsible for executing the closed-loop control strategy to calculate the optimal firing parameters. The control signal generation unit is responsible for converting the calculated parameters into control signals (107) compatible with the sintering furnace actuator and implementing adjustment logic including anomaly compensation.

10. An intelligent control system for a diamond diamond particle solidification and sintering furnace according to claim 1, characterized in that: It further includes an online learning module (8). The online learning module (8) is configured to receive the real-time feedback data and control status data generated during the operation of the system, adaptively update the network weights and dynamic parameters in the liquid neural network module (3) and the attention fusion module (4), and continuously optimize the accuracy and response speed of the control model to adapt to different sintering working conditions.

Images