A method for dewaxing of wax patterns in investment casting

Abstract

The present application relates to the technical field of investment casting, and particularly relates to a wax pattern dewaxing method for investment casting, which comprises the following steps: if the thickness of a thin-walled area is less than a preset threshold, the heating power of the area is reduced and air flow assistance is increased to generate process parameters for thin-walled area protection; if the volume of a thick part is greater than a preset threshold, the heating time is prolonged and the local pressure is increased to generate process parameters for thick part treatment, thereby obtaining a dynamic parameter regulation scheme; an infrared sensor is used to collect the wax pattern surface temperature field and residual wax distribution in real time, and the dynamic parameter regulation scheme is combined to generate real-time data flow for dewaxing progress monitoring; the temperature gradient and wax melting rate are extracted from the real-time data flow, if the temperature gradient exceeds a preset range, the heating power and air flow speed are adjusted; if the wax melting rate is lower than a preset threshold, the local pressure is increased to generate feedback instructions for process parameter adjustment.

Description

Technical Field

[0001] This invention relates to the field of investment casting technology, and more particularly to a method for dewaxing wax patterns in investment casting. Background Technology

[0002] Investment casting, as one of the core technologies of precision manufacturing, occupies a key position in aerospace, automotive, and medical device fields due to its high precision and high surface quality. However, dewaxing, a crucial step in investment casting, directly affects casting quality and production efficiency, making the stability and accuracy of the process paramount. Traditional dewaxing methods mainly rely on steam or high-temperature baking. While these methods offer advantages in efficiency and cost, they struggle with complex geometries, especially when thin-walled structures and heavy parts coexist, easily leading to shell damage or incomplete dewaxing. Furthermore, existing methods lack the flexibility to differentiate treatment for different areas of the wax pattern, failing to provide precise control based on the specific characteristics of the wax pattern, thus limiting the optimization of dewaxing results.

[0003] The core challenge in dewaxing wax patterns lies in balancing dewaxing efficiency with shell integrity, while also adapting to the complexity of the wax pattern's geometry. Wax patterns exhibit significant geometric variations; thin-walled areas are sensitive to high-strength treatment and prone to cracking, while thicker areas or internal cavities are more difficult to dewax due to greater wax buildup. Current technologies struggle to provide targeted treatment for different areas and lack dynamic control capabilities, resulting in either low efficiency or damage to the shell during dewaxing. Furthermore, precise regional control of complex wax patterns presents another challenge; traditional methods cannot adjust process parameters in real-time based on the dewaxing progress, failing to meet the demands of high-precision casting.

[0004] Therefore, how to develop a dewaxing device and method that can achieve precise regional control based on the geometric characteristics and complexity of the wax model, so as to improve the dewaxing efficiency while ensuring the integrity of the shell, has become a key problem that needs to be solved in this study. Summary of the Invention

[0005] This invention provides a method for dewaxing wax patterns in investment casting, mainly comprising:

[0006] The three-dimensional geometric data of the wax model is obtained by scanning equipment, and a digital model containing thin-walled areas, thick parts and internal cavity structures is constructed. A geometric feature distribution map of the wax model is generated, and the thickness, curvature and volume parameters of each area are determined.

[0007] Based on the geometric feature distribution map of the wax model, a clustering algorithm is used to divide the wax model into thin-walled regions, thick-walled parts, and transition regions, generating a region segmentation mask to obtain the basic data for regional control.

[0008] The thickness parameters of thin-walled regions and the volume parameters of thick-walled regions are extracted from the region segmentation mask. The thermal conductivity coefficient and wax melting threshold of each region are calculated to determine the initial process parameters for dewaxing by region.

[0009] If the thickness of the thin-walled region is less than the preset threshold, the heating power of the region is reduced and airflow assistance is increased to generate process parameters for thin-walled region protection.

[0010] If the volume of the thick part is greater than the preset threshold, the heating time is extended and the local pressure is increased to generate process parameters for the thick part treatment and obtain a dynamic parameter control scheme.

[0011] Infrared sensors are used to collect the surface temperature field and residual wax distribution of the wax model in real time. Combined with a dynamic parameter control scheme, a real-time data stream for monitoring the dewaxing progress is generated.

[0012] The temperature gradient and wax melting rate are extracted from the real-time data stream. If the temperature gradient exceeds the preset range, the heating power and airflow speed are adjusted.

[0013] If the melting rate of the wax material is lower than the preset threshold, the local pressure is increased, and a feedback instruction for adjusting the process parameters is generated.

[0014] By updating the process parameters of the regional control through feedback instructions, the actuators of the dewaxing unit are driven to adjust the heating power, airflow speed and pressure distribution, thereby generating a dynamic dewaxing scheme that adapts to complex structures.

[0015] According to the dynamic dewaxing scheme, a multi-channel control system is used to simultaneously process the dewaxing process of thin-walled areas, thick-walled areas and transition areas, and generate high-precision casting process execution data.

[0016] Extract the dewaxing completion rate and shell stress distribution of each region from the process execution data. If the completion rate reaches the preset standard and the stress is lower than the safety threshold, the dewaxing process is terminated, and the final dewaxing result is obtained.

[0017] Optionally, the step of acquiring three-dimensional geometric data of the wax model through a scanning device, constructing a digital model including thin-walled regions, thick-walled parts, and internal cavity structures, generating a geometric feature distribution map of the wax model, and determining the thickness, curvature, and volume parameters of each region includes:

[0018] The three-dimensional geometric data of the wax model is collected by scanning equipment to generate a raw dataset containing point cloud information and to ensure data integrity.

[0019] If the original dataset contains noise, a denoising algorithm is used to process the original dataset to generate smooth point cloud data and obtain geometric information without noise interference.

[0020] Based on smooth point cloud data, a digital model containing thin-walled regions, thick parts, and internal cavity structures is constructed, a three-dimensional mesh structure is generated, and the topological integrity of the model is determined.

[0021] Geometric features are extracted from the 3D mesh structure, and the thickness, curvature, and volume parameters of each region are calculated to obtain a set of feature parameters.

[0022] A visualization algorithm is used to process the feature parameter set, generate a geometric feature distribution map, and determine the spatial distribution of parameters in each region.

[0023] If the thickness parameter in the distribution map exceeds the preset threshold, the geometric features of the thin-walled region and the thick part are locally optimized and adjusted to generate an updated feature parameter set.

[0024] By regenerating the geometric feature distribution map using the updated feature parameter set, the final distribution of thickness, curvature, and volume parameters is determined.

[0025] Optionally, based on the geometric feature distribution map of the wax model, a clustering algorithm is used to divide the wax model into thin-walled regions, thick-walled regions, and transition regions, generating a region segmentation mask to obtain the basic data for region-based control, including:

[0026] The structural information of the wax model is extracted by geometric feature distribution, and feature vectors are generated by feature extraction algorithm to obtain the geometric description data of the wax model;

[0027] Based on the geometric description data of the wax model, a clustering algorithm was used to group the feature vectors and determine the initial division results of the thin-walled region, the thick region, and the transition region.

[0028] If the region boundaries are discontinuous in the initial partitioning result, the grouping of feature vectors is adjusted through spatial adjacency analysis to obtain optimized region partitioning data;

[0029] Based on the optimized region segmentation data, a region segmentation mask is generated to determine the pixel-level distribution of each region.

[0030] By using a region segmentation mask, the boundary features of thin-walled regions, thick regions, and transition regions are extracted to obtain region boundary description data.

[0031] Based on the regional boundary description data, calculate the geometric characteristics of each region and generate regional control data;

[0032] By mapping the process parameters of the wax mold structure to the regional control data, the regional process control parameters are obtained.

[0033] Optionally, the step of extracting the thickness parameters of thin-walled regions and the volume parameters of thick-walled regions from the region segmentation mask, calculating the thermal conductivity coefficient and wax melting threshold of each region, and determining the initial process parameters for region-specific dewaxing includes:

[0034] The geometric features of thin-walled regions and thick-walled regions are obtained from the region segmentation mask by the segmentation algorithm, and the thin-walled thickness parameters and the volume parameters of the thick-walled regions are obtained.

[0035] Based on the thin-wall thickness parameters and the volume parameters of the thick-walled parts, the thermal conductivity coefficient of each region is calculated using the finite element analysis method to obtain the thermal conductivity coefficient distribution;

[0036] If the thermal conductivity coefficient distribution exceeds the preset threshold range, the region is re-divided, the updated thin-wall thickness parameters and the volume parameters of the thick parts are obtained, the thermal conductivity coefficient is recalculated, and the final thermal conductivity coefficient distribution is determined.

[0037] By using the thermal conductivity coefficient distribution and material property database, the melting threshold of wax material in each region was calculated, and the melting threshold distribution was obtained;

[0038] Based on the melting threshold distribution, a clustering algorithm is used to classify the regions and obtain a priority sequence for dewaxing in each region;

[0039] The initial process parameters for each region are calculated using priority sequence and thermal conductivity distribution, resulting in the initial parameter configuration for dewaxing in different regions.

[0040] If the initial parameter configuration does not match the preset process constraints, the regional classification weights are adjusted, the initial process parameters are recalculated, and the final initial parameter configuration is determined.

[0041] Optionally, if the thickness of the thin-walled region is less than a preset threshold, the heating power of that region is reduced and airflow assistance is increased to generate process parameters for thin-walled region protection, including:

[0042] The thickness data of the thin-walled region is obtained by the sensor, and it is determined whether the thickness is lower than a preset threshold to obtain the thickness detection result.

[0043] If the thickness detection result is lower than the preset threshold, the heating power reduction ratio is calculated based on the thickness data, and the adjusted heating power value is determined.

[0044] Based on the adjusted heating power value and the location of the thin-walled area, a corresponding power control command is generated;

[0045] The airflow control module obtains the current position of the thin-walled region, calculates the required airflow assist intensity, and determines the airflow control parameters.

[0046] An initial set of process parameters is generated by using a preset process parameter generation model and integrating power control commands and airflow control parameters.

[0047] For the initial set of process parameters, an optimization algorithm is executed to adjust the ratio between parameters, resulting in the final set of process parameters.

[0048] The real-time monitoring module acquires the state of the thin-walled region after the process parameters are executed, determines whether the protection mechanism requirements are met, and generates feedback data.

[0049] Optionally, if the volume of the thick part is greater than a preset threshold, the heating time is extended and the local pressure is increased to generate process parameters for processing the thick part, resulting in a dynamic parameter control scheme, including:

[0050] If the volume data of the thick part exceeds the preset threshold, its three-dimensional geometric features are obtained through a stereomicroscope to generate a structural description of the thick part.

[0051] Based on the structural description, the heat conduction distribution in the thick parts was calculated using finite element analysis to obtain temperature field distribution data.

[0052] If there are cold spot regions in the temperature field distribution data, the heating time parameter is adjusted through an iterative optimization algorithm to generate the optimized time parameter;

[0053] Based on the optimized time parameters, obtain stress distribution data for the thick parts and determine whether the local pressure meets the preset conditions.

[0054] If the local pressure is lower than the preset condition, the pressure parameters are adjusted incrementally through numerical simulation to obtain the updated pressure control parameters;

[0055] By integrating and optimizing the time parameters and pressure control parameters, a set of process parameters for the heavy parts is generated;

[0056] Based on the set of process parameters, a dynamic control scheme is output using a real-time monitoring system.

[0057] Optionally, the method of using infrared sensors to collect the surface temperature field and residual wax distribution of the wax model in real time, combined with a dynamic parameter control scheme, generates a real-time data stream for monitoring the dewaxing progress, including:

[0058] The initial temperature distribution and wax distribution are obtained by collecting surface temperature field data and residual wax distribution data of the wax model using an infrared sensor.

[0059] Based on the initial temperature distribution and wax quantity distribution, a convolutional neural network algorithm is used to analyze the spatial correlation between the temperature field and the wax quantity, and to determine the temperature-wax quantity mapping relationship.

[0060] If there are local high-temperature areas in the temperature-wax quantity mapping relationship, the heating power is adjusted through a dynamic parameter control scheme to obtain the optimized parameter configuration;

[0061] With optimized parameter configuration, the temperature field and residual wax distribution on the wax model surface are updated using a real-time acquisition module, generating updated temperature and wax distributions.

[0062] Based on the updated temperature distribution and wax quantity distribution, the mean square error algorithm is used to compare the current wax quantity with the preset threshold to determine the dewaxing progress status.

[0063] If the dewaxing progress does not reach the preset threshold, the heating power is readjusted through the dynamic parameter control module to obtain a new parameter configuration;

[0064] With new parameter configuration, the data stream generation module integrates temperature distribution, wax quantity distribution, and progress status to generate a real-time data stream.

[0065] Optionally, the step of extracting the temperature gradient and wax melting rate from the real-time data stream, and adjusting the heating power and airflow speed if the temperature gradient exceeds a preset range, includes:

[0066] Temperature gradient and wax melting rate are obtained from real-time data streams, and environmental data is collected using a sensor array to obtain the raw dataset;

[0067] For the original dataset, the mean filtering algorithm is used to process the temperature data to determine the smoothed temperature gradient value;

[0068] If the smoothed temperature gradient value exceeds the preset range, the degree of anomaly is determined by a logistic regression algorithm to obtain the anomaly level.

[0069] Based on the level of abnormality, a proportional control algorithm is used to adjust the heating power and determine a new power setpoint.

[0070] By obtaining the associated airflow velocity parameters through the new power setting, it is determined whether the airflow velocity needs to be adjusted synchronously, and the velocity adjustment value is obtained.

[0071] For the speed adjustment value, digital signal processing technology is used to update the airflow control unit to determine the final airflow speed parameters;

[0072] The final airflow velocity parameters and heating power values ​​are obtained, the system status is monitored in real time, and the updated temperature gradient and melting rate are obtained.

[0073] Optionally, if the melting rate of the wax material is lower than a preset threshold, the local pressure is increased, and a feedback instruction for adjusting process parameters is generated, including:

[0074] If the melting rate of the wax is lower than the preset threshold, the real-time melting rate data is obtained through the sensor, and the deviation between the rate and the threshold is judged by the comparison algorithm to obtain the deviation value.

[0075] Based on the deviation value, the corresponding pressure increment is queried using a preset mapping table to determine the local pressure adjustment amount;

[0076] By adjusting the pressure, a pressure control command is generated and output to the pressure regulating device to obtain the updated pressure status.

[0077] The updated pressure status is obtained, and combined with real-time melting rate data, the support vector machine algorithm is used to determine the stability of process parameters and obtain stability index.

[0078] If the stability index is lower than the preset standard, the process parameters are optimized using the gradient descent algorithm to generate new parameter adjustment instructions;

[0079] Based on the new parameter adjustment instructions, update the process parameter configuration, output to the control system, and obtain the optimized process status;

[0080] By optimizing the process status, real-time monitoring data is obtained to determine whether the melting rate of the wax material has reached the preset threshold, and the final process parameter instructions are generated.

[0081] Optionally, the step of updating the process parameters controlled by different regions through feedback commands, driving the actuator of the dewaxing device to adjust the heating power, airflow speed, and pressure distribution, and generating a dynamic dewaxing scheme adapted to complex structures includes:

[0082] The system acquires real-time feedback command data from sensors, parses the zonal control requirements contained in the commands, and obtains the zonal control parameter set.

[0083] The control parameters of the dewaxing device are updated by updating the partition control parameter set, and the heating power adjustment value is determined by using a preset mapping rule;

[0084] The heating module of the dewaxing device is driven according to the heating power adjustment value. If the heating power exceeds the preset threshold, the power increment is reduced to obtain a stable heating distribution.

[0085] By acquiring airflow velocity sensor data and combining it with the zonal control parameter set, a linear regression algorithm is used to determine the optimal airflow velocity value.

[0086] The blower module of the dewaxing device is adjusted by optimizing the airflow velocity. If the velocity fluctuation exceeds the preset range, the velocity curve is smoothed to determine a uniform airflow field.

[0087] Pressure distribution data is obtained from pressure sensors, and combined with a complex structural geometric model, finite element analysis is used to obtain an optimized pressure distribution scheme.

[0088] Based on the optimization schemes of heating distribution, airflow field and pressure distribution, and by integrating the zoning control parameters, a dynamic dewaxing scheme adapted to complex structures is generated.

[0089] Optionally, according to the dynamic dewaxing scheme, a multi-channel control system is used to simultaneously process the dewaxing process of thin-walled areas, thick-walled areas, and transition areas, generating high-precision casting process execution data, including:

[0090] Geometric feature data of thin-walled regions, thick-walled parts, and transition regions are acquired through a multi-channel control system to determine the dewaxing parameters for each region.

[0091] The thermal conductivity characteristics of each region are extracted from the geometric feature data to obtain the heat distribution model of the region division;

[0092] Based on the heat distribution model, a synchronous processing algorithm is used to adjust the dewaxing rate of the multi-channel control system and generate the process execution sequence for each region.

[0093] If the rate deviation of the process execution sequence exceeds the preset threshold, the control system parameters are adjusted through an iterative optimization algorithm to obtain a stable dewaxing rate.

[0094] Time series data are extracted from the stable dewaxing rate to generate high-precision casting process data;

[0095] For the process data, a data verification algorithm is used to detect outliers to obtain the final process execution data.

[0096] Optionally, the step of extracting the dewaxing completion rate and shell stress distribution of each region from the process execution data, and terminating the dewaxing process if the completion rate reaches a preset standard and the stress is below a safety threshold, to obtain the final dewaxing result, includes:

[0097] Acquire process data, extract the dewaxing completion rate and shell stress distribution of each region, and obtain the initial analysis dataset;

[0098] By analyzing the initial dataset, the mean and standard deviation of dewaxing completion in each region are calculated to obtain a quantitative indicator of completion.

[0099] The completion rate is measured by a quantitative indicator. If the completion rate of each region reaches the preset threshold, a completion rate qualification mark is generated.

[0100] The stress distribution of the shell is extracted from the initial analysis dataset, and the maximum stress value and regional differences are calculated to obtain the stress distribution characteristics.

[0101] Based on the stress distribution characteristics, if the maximum stress is lower than the safety threshold and the regional difference is less than a preset range, a stress safety indicator is generated.

[0102] By using the completion qualification mark and the stress safety mark, a logical AND operation is used to determine whether both are true, and a termination signal is generated.

[0103] Based on the termination signal, the control system stops the dewaxing process and obtains the final dewaxing result dataset.

[0104] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0105] This invention discloses a method for dewaxing wax patterns in investment casting, specifically an intelligent dewaxing method for complex structure wax patterns. The method involves scanning to acquire three-dimensional geometric data of the wax pattern, constructing a digital model, and generating a geometric feature distribution map. A clustering algorithm is used to divide the wax pattern into thin-walled regions, thick-walled areas, and transition regions, and parameters for each region are extracted to determine initial process parameters. Heating power, airflow, and pressure are dynamically adjusted based on region characteristics, and the temperature field and residual wax distribution are monitored in real time. Temperature gradients and melting rates are extracted for feedback adjustments. A multi-channel control system synchronously processes the dewaxing process in each region, achieving precise dewaxing of complex structure wax patterns. This invention can adapt to different structural features, dynamically control process parameters, effectively solve the problems of thin-walled protection and thick-walled area handling in the dewaxing process of complex wax patterns, improve dewaxing accuracy and efficiency, and provide technical support for high-quality casting. Attached Figure Description

[0106] Figure 1 This is a flowchart of a wax pattern dewaxing method for investment casting according to the present invention.

[0107] Figure 2 This is a schematic diagram of a wax pattern dewaxing method for investment casting according to the present invention.

[0108] Figure 3 This is another schematic diagram of a wax pattern dewaxing method for investment casting according to the present invention. Detailed Implementation

[0109] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0110] like Figure 1-3 This embodiment of a method for dewaxing wax patterns in investment casting may specifically include:

[0111] S101. Obtain three-dimensional geometric data of the wax model through a scanning device, construct a digital model including thin-walled areas, thick parts and internal cavity structures, generate a geometric feature distribution map of the wax model, and determine the thickness, curvature and volume parameters of each area.

[0112] Three-dimensional geometric data of wax models are acquired using scanning equipment to generate an original dataset containing point cloud information, ensuring data integrity. If the original dataset contains noise, a denoising algorithm is used to process it, generating smooth point cloud data to obtain noise-free geometric information. Based on the smooth point cloud data, a digital model containing thin-walled regions, thick parts, and internal cavity structures is constructed, generating a three-dimensional mesh structure to ensure the model's topological integrity. Geometric features are extracted from the three-dimensional mesh structure, and the thickness, curvature, and volume parameters of each region are calculated to obtain a feature parameter set. A visualization algorithm is used to process the feature parameter set, generating a geometric feature distribution map to determine the spatial distribution of parameters in each region. If the thickness parameter in the distribution map exceeds a preset threshold, the geometric features of the thin-walled regions and thick parts are locally optimized and adjusted to generate an updated feature parameter set. Using the updated feature parameter set, a new geometric feature distribution map is generated to determine the final distribution of thickness, curvature, and volume parameters.

[0113] For example, point cloud data of the wax model surface is acquired at a resolution of 0.02 mm using a high-precision laser scanner. Point cloud registration is performed using the ICP iterative nearest-point algorithm, with the registration error controlled within 0.05 mm. A triangular mesh model is generated based on the Poisson reconstruction algorithm, and the octree depth is set to 10 levels to achieve a surface fitting accuracy of 0.1 mm. An input point set is obtained from the point cloud data, and the normal vector of each point is calculated using the Poisson reconstruction algorithm to obtain the normal vector distribution. An octree structure is constructed based on the normal vector distribution, depth parameters are set, space is divided, and spatial mesh units are determined. For each spatial mesh unit, the Poisson reconstruction algorithm is applied to generate an initial triangular mesh model, obtaining a preliminary surface. Based on the preliminary surface, surface smoothing is applied, and the positions of the mesh vertices are adjusted to obtain a smooth triangular mesh. Surface features are extracted from the smooth triangular mesh. If the feature deviation exceeds a preset threshold, the mesh is iteratively optimized to determine the optimized mesh. The surface fitting accuracy is calculated using the optimized mesh to determine if the accuracy meets the requirements, resulting in the final triangular mesh model. The final triangular mesh model is obtained, and a high-precision surface is output, completing the model generation. In the model segmentation stage, the region growing algorithm is used to mark regions with a wall thickness of less than 1.5 mm as thin-walled structures and planar regions with a radius of curvature greater than 50 mm as thick parts. The internal cavity structure is extracted by the AlphaShape algorithm, and the α parameter is set to 3.0 to accurately identify cavities with a diameter of more than 2 mm.

[0114]

[0115] Where T represents the wall thickness function, x and y represent spatial coordinates, σ represents the regional diffusion coefficient, x_0 and y_0 represent the coordinates of the initial seed point, and 1.5 represents the wall thickness threshold.

[0116]

[0117] Where R represents the radius of curvature function, p represents a point on the surface, n_i represents the normal vector of the i-th adjacent point, v represents the normal vector of the center point, n represents the number of adjacent points, and 50 represents the radius of curvature threshold.

[0118] A(S)={∪B α (p i )|p i ∈S,diameter(B α )≥2}

[0119] Where A represents the AlphaShape function, S represents the point set, B_α represents a sphere with radius α, p_i represents the sampling point, diameter represents the cavity diameter, α = 3.0 is the sphere radius parameter, and 2 represents the cavity diameter threshold.

[0120] Thickness analysis was performed using the projection ray method, emitting 1000 rays per mm along the surface normal. 2 Thickness measurements were performed, and K-means clustering was used to categorize the thickness into three levels: thin-walled region (0.8-1.5 mm), transition region (1.5-3.0 mm), and thick-walled region (3.0-5.0 mm). Curvature calculation was based on fitting a local surface using the moving least squares method, with a neighborhood radius of 2 mm. A threshold was set when generating the Gaussian curvature distribution map: high curvature (>0.8 mm). -1 ), medium curvature (0.3-0.8mm) -1 and low curvature (<0.3mm) -1 Volume parameters were calculated using voxelization with 0.2 mm cubes. A flooding algorithm was used to count the number of voxels in each region. Thin-walled areas accounted for approximately 15±3% of the volume, thick-walled areas for 40±5%, and internal structures for 25±2%. Finally, a feature fusion algorithm mapped the thickness, curvature, and volume parameters to a UV coordinate system, generating a geometric feature distribution map containing RGB channels. The R channel encodes thickness (0-255 corresponds to 0.5-5.0 mm), and the G channel represents curvature (0-255 corresponds to 0-1.0 mm). -1 The B channel reflects the volume weight (0-255 corresponds to 0-50% proportion).

[0121]

[0122] Where T(u,v) represents the thickness feature mapping function, h_i represents the thickness value of the sampling point, max(h) represents the maximum thickness value of 5.0mm, α is the normalization coefficient, and n is the number of sampling points.

[0123]

[0124] Where C(u,v) represents the curvature feature mapping function, S(x,y) represents the surface function, Ω represents the integration region, and β is the curvature coefficient. Represents the gradient operator, This represents the Laplace operator.

[0125]

[0126] Where V(u,v) represents the volume feature mapping function, v_k represents the local volume, V_total represents the total volume, γ is the volume weight coefficient, and m is the number of voxels.

[0127] S102. Based on the geometric feature distribution map of the wax model, a clustering algorithm is used to divide the wax model into thin-walled regions, thick-walled parts, and transition regions, generating a region segmentation mask to obtain the basic data for regional control.

[0128] Structural information of the wax model is extracted through geometric feature distribution. Feature vectors are generated using a feature extraction algorithm to obtain geometric description data of the wax model. Based on the geometric description data, a clustering algorithm is used to group the feature vectors, determining the initial division results for thin-walled regions, thick-walled regions, and transition regions. If the region boundaries in the initial division results are discontinuous, spatial adjacency analysis is used to adjust the grouping of feature vectors to obtain optimized region division data. Based on the optimized region division data, a region segmentation mask is generated to determine the pixel-level distribution of each region. Boundary features of the thin-walled regions, thick-walled regions, and transition regions are extracted using the region segmentation mask to obtain region boundary description data. Based on the region boundary description data, the geometric characteristics of each region are calculated to generate sub-regional control data. The process parameters of the wax model structure are mapped using the sub-regional control data to obtain regional process control parameters.

[0129] For example, point cloud data of the wax model was first acquired using a 3D scanner with a sampling precision of 0.1 mm, generating a geometric feature distribution map containing 500,000 data points. The density-based DBSCAN clustering algorithm was used for region division, with a neighborhood radius ε of 2 mm and a minimum number of points MinPts of 50. The local curvature (range 0-0.5 mm^-1) and the angle between the normal vectors (threshold 15 degrees) of each point were calculated as feature vectors. Regions with curvature greater than 0.3 mm^-1 and a point density higher than 120 points / m³ were marked as thin-walled regions, identifying a total of 8 independent thin-walled regions with an average thickness of 1.

[0130] 2 ± 0.3 mm. Thick regions were identified using k-means clustering (k = 3), with features including an angle less than 30 degrees between the point cloud normal vector and the gravity direction and a curvature less than 0.1 mm^-1. Three main thick regions were identified, with an average thickness of 12.5 mm. Transition regions were defined by calculating the Hausdorff distance (threshold 5 mm) between the thin-walled and thick regions, and then using morphological dilation (structural element radius 3 mm) to expand the boundaries, ultimately generating a binary mask image (resolution 2048 × 2048) containing the three regions. Based on the region segmentation results, a cooling rate control parameter of 3.2℃ / s was set for the thin-walled region, a gradient cooling strategy of 0.8℃ / s was adopted for the thick regions, and a linear transition control of 1.5℃ / s was implemented for the transition regions. All parameters were verified through finite element thermodynamic simulation, and the maximum temperature deviation was controlled within...

[0131] Within ±2.1℃.

[0132] S103. Extract the thickness parameters of the thin-walled region and the volume parameters of the thick part from the region segmentation mask, calculate the thermal conductivity coefficient and wax melting threshold of each region, and determine the initial process parameters for dewaxing by region.

[0133] Geometric features of thin-walled and thick-walled regions are obtained from a region segmentation mask using a segmentation algorithm, yielding thin-walled thickness parameters and thick-walled volume parameters. Based on these parameters, the thermal conductivity coefficient of each region is calculated using finite element analysis, resulting in a thermal conductivity coefficient distribution. If the thermal conductivity coefficient distribution exceeds a preset threshold range, the region is re-divided, and updated thin-walled thickness and thick-walled volume parameters are obtained. The thermal conductivity coefficient is then recalculated to determine the final thermal conductivity coefficient distribution. Using the thermal conductivity coefficient distribution and a material property database, the melting threshold of the wax material in each region is calculated, resulting in a melting threshold distribution. Based on the melting threshold distribution, a clustering algorithm is used to classify the regions, obtaining a priority sequence for dewaxing by region. Using the priority sequence and thermal conductivity coefficient distribution, the initial process parameters for each region are calculated, resulting in the initial parameter configuration for dewaxing by region. If the initial parameter configuration does not match the preset process constraints, the region classification weights are adjusted, the initial process parameters are recalculated, and the final initial parameter configuration is determined.

[0134] For example, when extracting thickness parameters of thin-walled regions from a region segmentation mask, a distance transformation method based on morphological algorithms can be used. For instance, Euclidean distance transformation can be performed on the mask image to extract the skeleton lines, and then the local thickness can be calculated by doubling the distance value on both sides of the skeleton lines. Assuming the pixel width of a thin-walled region is 5 to 8 pixels, corresponding to an actual thickness of 0.5 to 0.8 mm, regions with a thickness less than 1 mm can be marked as thin-walled regions through threshold segmentation. For calculating the volume parameters of thicker areas, connected component analysis combined with voxel statistics is used. For example, if a thicker region contains 5000 voxels, and each voxel is 0.1 mm × 0.1 mm × 0.1 mm, then the volume is 5 cubic millimeters. The thermal conductivity coefficient is calculated based on material properties and regional geometric features. For thin-walled regions, Fourier's law is used for derivation. Assuming the thermal conductivity of wax is 0.25 W / (m·K), the thermal resistance is 2.4 K / W when the thickness is 0.6 mm. For the thicker regions, volumetric heat capacity is considered, with a specific heat capacity of 2.1 J / (g·K) and a density of 0.95 g / cm³. 3 Calculations show that a 5 cubic millimeter area requires 6.3 joules of heat to reach its melting point. The melting threshold of the wax material was set using differential scanning calorimetry data, for example, an initial melting temperature of 52℃ and a complete melting temperature of 58℃. A regional heating strategy was generated based on the regional thermal conductivity: a gradient heating method was used for thin-walled areas, with an initial temperature of 55℃ and a rate of increase of 2℃ / min; a step heating method was used for thick areas, with an initial temperature of 70℃ maintained for 30 seconds to ensure heat penetration. The uniformity of the thermal field distribution was verified through finite element simulation. After iterative optimization, a combination of process parameters was determined: a heating time of 90 seconds for thin-walled areas and a heating time of 150 seconds for thick areas.

[0135] S104. If the thickness of the thin-walled region is less than the preset threshold, reduce the heating power of the region and increase airflow assistance to generate process parameters for thin-walled region protection.

[0136] The thickness data of the thin-walled region is acquired by sensors to determine if the thickness is below a preset threshold, thus obtaining a thickness detection result. If the thickness detection result is below the preset threshold, the heating power reduction ratio is calculated based on the thickness data to determine the adjusted heating power value. Based on the adjusted heating power value and the location of the thin-walled region, a corresponding power control command is generated. The current position of the thin-walled region is acquired through the airflow control module, the required airflow assistance intensity is calculated, and the airflow control parameters are determined. A preset process parameter generation model is used to fuse the power control command and airflow control parameters to generate an initial process parameter set. For the initial process parameter set, an optimization algorithm is executed to adjust the ratio between parameters to obtain the final process parameter set. The state of the thin-walled region after the process parameters are executed is acquired through a real-time monitoring module to determine whether the protection mechanism requirements are met and to generate feedback data.

[0137] For example, in the process of generating process parameters for thin-walled region protection, the wall thickness features are first extracted using a 3D model slicing algorithm. The thickness distribution of each region is then calculated using a voxel-based spatial partitioning method. When the thickness of a region is detected to be below a preset threshold (e.g., 0.8 mm), a power adjustment module is triggered. This module uses a gradient descent algorithm to dynamically calculate the power correction, for example, reducing the original heating power from 2000W to 1500W proportionally to the thickness. Simultaneously, computational fluid dynamics (CFD) simulation is used to determine the optimal airflow parameters, increasing the auxiliary airflow velocity from 5 m / s to 8 m / s to enhance heat dissipation. The gradient descent algorithm is used to calculate the power correction, obtain thickness ratio data, and determine the adjusted heating power. Computational fluid dynamics simulation is used to obtain the heat distribution characteristics from the adjusted heating power data, yielding the initial airflow parameters. If the heat distribution characteristics exceed the preset threshold, an iterative optimization algorithm is used to adjust the airflow velocity to determine the optimized airflow parameters. Based on the optimized airflow parameters, an auxiliary airflow control signal is obtained, resulting in an airflow velocity adjustment command. This airflow velocity adjustment command dynamically adjusts the auxiliary airflow drive module to obtain real-time heat dissipation effect data. The system extracts temperature change trends from real-time heat dissipation data. If the temperature change trend does not reach the preset target, a genetic algorithm is used to fine-tune airflow parameters to determine the final heat dissipation scheme. Based on the final heat dissipation scheme, a set of control parameters is obtained to generate equipment operation commands. The system establishes a finite element model of heat conduction and iteratively analyzes the temperature field distribution in 0.1-second increments. When the predicted peak temperature exceeds the material's critical value (e.g., 240℃ for ABS plastic), a secondary correction algorithm is automatically activated. A thickness compensation coefficient (t is the actual thickness, T is the threshold) is added to the power adjustment formula P_new=P_original×[1-0.5×(1-t / T)], and the airflow control system is linked to compensate for the wind speed according to a linear relationship of ΔV=2.5×(Tt). The entire process is adjusted in real time by a PID controller with a proportional coefficient of 0.6, an integral time constant of 2 seconds, and a derivative time constant of 0.5 seconds to ensure that temperature fluctuations are controlled within ±3℃. For complex geometries, an adaptive mesh subdivision technique is employed, refining the mesh size from 0.5 mm to 0.2 mm in areas with a curvature radius less than 3 mm to improve thickness detection accuracy to ±0.05 mm. All parameters are stored in a JSON-formatted process file, including a power mapping matrix and an airflow parameter lookup table, for use in subsequent processing stages.

[0138] S105. If the volume of the thick part is greater than the preset threshold, extend the heating time and increase the local pressure to generate process parameters for the thick part treatment and obtain a dynamic parameter control scheme.

[0139] If the volume data of the thick section exceeds a preset threshold, its three-dimensional geometric features are obtained using a stereomicroscope to generate a structural description of the thick section. Based on the structural description, finite element analysis is used to calculate the heat conduction distribution of the thick section, obtaining temperature field distribution data. If cold spot regions exist in the temperature field distribution data, the heating time parameter is adjusted using an iterative optimization algorithm to generate optimized time parameters. Based on the optimized time parameters, stress distribution data of the thick section is obtained to determine whether the local pressure meets preset conditions. If the local pressure is lower than the preset conditions, the pressure parameter is incrementally adjusted through numerical simulation to obtain updated pressure control parameters. By integrating the optimized time parameters and pressure control parameters, a set of process parameters for the thick section is generated. Based on the set of process parameters, a dynamic control scheme is output using a real-time monitoring system.

[0140] For example, in the dynamic parameter control scheme, the workpiece geometric data is first obtained through 3D scanning, and the workpiece is divided into 1mm segments using a voxelization algorithm. 3 For each voxel unit, calculate the distance from each voxel unit to the surface, when the volume of a continuous voxel exceeds 50 cm³. 3 The area was identified as a thick, heavy section. For the identified thick, heavy sections, based on the heat conduction equation... Finite element analysis was performed, with the initial heating temperature set at 200℃. A control mechanism was triggered when the temperature difference Δt > 15℃. Pressure regulation employed a PID control algorithm with a proportional gain Kp = 0.8, integral time Ti = 120s, and derivative time Td = 30s. The target pressure value was calculated based on the thickness h using the formula P = 0.2h + 0.5MPa, with a maximum not exceeding 3MPa. The heating time extension Δt = 0.05V (V is the excess volume in cm³). 3 Simultaneously, a temperature gradient compensation coefficient k = 1 + 0.01ΔT is introduced to ensure uniformity. The system collects thermocouple data in real time, updating parameters every 5 seconds. Steady state is determined when the standard deviation of temperature fluctuation σ < 2℃. All parameters are transmitted to the PLC via the OPC-UA protocol. The actuator positioning accuracy reaches ±0.1mm, and the pressure control error is ±0.05MPa, forming a closed-loop control. Historical data is stored in a time-series database, and an LSTM network is used to predict the parameter adjustment amount for the next batch, with a prediction period set to 30 minutes.

[0141] S106. An infrared sensor is used to collect the surface temperature field and residual wax distribution of the wax model in real time. Combined with a dynamic parameter control scheme, a real-time data stream for monitoring the dewaxing progress is generated.

[0142] Infrared sensors are used to collect surface temperature field data and residual wax distribution data of the wax model, obtaining the initial temperature distribution and wax quantity distribution. Based on the initial temperature and wax quantity distributions, a convolutional neural network algorithm is used to analyze the spatial correlation between the temperature field and wax quantity, determining the temperature-wax quantity mapping relationship. If there are local high-temperature regions in the temperature-wax quantity mapping relationship, the heating power is adjusted through a dynamic parameter control scheme to obtain an optimized parameter configuration. Using the optimized parameter configuration, a real-time acquisition module updates the surface temperature field and residual wax quantity distribution of the wax model, generating updated temperature and wax quantity distributions. Based on the updated temperature and wax quantity distributions, a mean squared error algorithm is used to compare the current wax quantity with a preset threshold to determine the dewaxing progress status. If the dewaxing progress status has not reached the preset threshold, the heating power is readjusted through the dynamic parameter control module to obtain a new parameter configuration. Using the new parameter configuration, a data stream generation module integrates the temperature distribution, wax quantity distribution, and progress status to generate a real-time data stream.

[0143] For example, during the dewaxing process of the wax model, an infrared sensor collects real-time temperature field data of the wax model surface at a frequency of 10 times per second, with a temperature range of 20℃ to 80℃ and an accuracy of ±0.5℃. Simultaneously, a weight sensor monitors the residual wax distribution of the wax model with an accuracy of 0.1 grams. The collected temperature data undergoes noise removal using a Kalman filter algorithm to ensure data accuracy. The residual wax data is fitted using the least squares method to generate a wax distribution curve. Combining the temperature field and wax distribution data, the system employs a dynamic parameter control scheme based on a PID control algorithm to adjust the heating power and ventilation volume of the dewaxing furnace in real time.

[0144] For example, when the temperature in a certain area is detected to be below 50°C and the residual wax content is above 30 grams, the system will automatically increase the heating power of that area to 2000 watts and increase the ventilation to 50 cubic meters per hour. In this way, the system can generate a real-time data stream monitoring the dewaxing progress and update the dewaxing progress percentage every 5 seconds. The data analysis module uses a support vector machine algorithm to predict the dewaxing progress, with the prediction error controlled within ±2%. Finally, the system displays the real-time data stream and prediction results through a visual interface for operators' reference and decision-making.

[0145] S107. Extract the temperature gradient and wax melting rate from the real-time data stream. If the temperature gradient exceeds the preset range, adjust the heating power and airflow speed.

[0146] Temperature gradient and wax melting rate are acquired from real-time data streams. Environmental data is collected using a sensor array to obtain the raw dataset. For this raw dataset, a mean filtering algorithm is used to process the temperature data, determining a smoothed temperature gradient value. If the smoothed temperature gradient value exceeds a preset range, a logistic regression algorithm is used to determine the degree of anomaly and assign an anomaly level. Based on the anomaly level, a proportional control algorithm is used to adjust the heating power, determining a new power setpoint. Using this new power setpoint, the associated airflow velocity parameters are acquired to determine if synchronous airflow velocity adjustment is necessary, resulting in a velocity adjustment value. For this velocity adjustment value, digital signal processing technology is used to update the airflow control unit, determining the final airflow velocity parameters. The final airflow velocity parameters and heating power value are obtained, and the system status is monitored in real-time to obtain the updated temperature gradient and melting rate.

[0147] For example, in real-time data stream processing, temperature data is first collected 10 times per second using a thermocouple sensor, with a temperature range of 50℃ to 300℃ and an accuracy of ±0.5℃. A sliding window algorithm (window size of 5 seconds) is used to calculate the temperature gradient. For instance, if the current temperature sequence within the window is [120℃, 125℃, 130℃, 135℃, 140℃], then the gradient ΔT / Δx is (140-120) / 5 = 4℃ / s. A preset gradient safety threshold is 2-6℃ / s; when a gradient value of 7℃ / s is detected, a control mechanism is triggered. At this point, the heating power is adjusted based on the PID control algorithm. The setpoint P = Kp × e(t) + Ki × ∫e(t)dt + Kd × de(t) / dt, where Kp = 0.8, Ki = 0.2, Kd = 0.1, and e(t) is the deviation value (7-6 = 1). The calculated output power decreases from 1500W to 1350W. Simultaneously, a linear regression model y = 0.3x + 2 (where x is the temperature gradient and y is the airflow velocity in m / s) is used to adjust the airflow velocity. When x = 7, y = 4.1 m / s, increasing the original velocity from 3 m / s to 4.1 m / s. The melting rate of the wax material is monitored by a laser displacement sensor with a sampling interval of 200ms. When a sudden change in the melting rate from 0.5mm / s to 0.8mm / s is detected, combined with a temperature gradient exceeding the limit event, it is determined to be local overheating. The system automatically divides the heating zone into a 5×5 grid and implements targeted cooling for the overheated area (such as the grid at coordinate (3,2)). The power is reduced by 20% while the airflow velocity is increased by 15%, ensuring that the gradient returns to within 5℃ / s. All control parameters are written to the PLC controller in real time via the MQTT protocol, with the response delay controlled within 50ms.

[0148] S108. If the melting rate of the wax material is lower than the preset threshold, increase the local pressure and generate a feedback instruction to adjust the process parameters.

[0149] If the wax melting rate is lower than a preset threshold, real-time melting rate data is acquired via sensors. A comparison algorithm is used to determine the deviation between the rate and the threshold, obtaining the deviation value. Based on the deviation value, a preset mapping table is used to look up the corresponding pressure increment to determine the local pressure adjustment amount. A pressure control command is generated based on the pressure adjustment amount and output to the pressure regulating device to obtain the updated pressure status. The updated pressure status is then combined with real-time melting rate data, and a support vector machine algorithm is used to determine the stability of the process parameters, obtaining a stability index. If the stability index is lower than a preset standard, the process parameters are optimized using a gradient descent algorithm, generating a new parameter adjustment command. Based on the new parameter adjustment command, the process parameter configuration is updated and output to the control system to obtain the optimized process status. Real-time monitoring data is acquired based on the optimized process status to determine whether the wax melting rate has reached the preset threshold, generating the final process parameter command.

[0150] For example, in the wax melting process, when the real-time monitoring system detects that the melting rate in a certain area drops to 80 grams per minute (the preset threshold is 100 grams per minute), the control system calculates the pressure compensation value based on the PID algorithm. The specific formula is ΔP = Kp × e(t) + Ki × ∫e(t)dt + Kd × de(t) / dt, where the error e(t) = 100 - 80 = 20 grams per minute, the proportional coefficient Kp is set to 0.5, the integral time Ti = 2 minutes, and the derivative time Td = 0.1 minutes. After real-time calculation, the output pressure adjustment ΔP = 12 MPa is applied to the local pressurization module through the hydraulic actuator, increasing the pressure from the initial value of 15 MPa to 27 MPa. Simultaneously, the process parameter library cross-validates historical data models (including a three-dimensional mapping table of temperature-pressure-rate). When the pressure increases to 27 MPa, the model predicts the melting rate should rebound to 105 g / min. This prediction is verified by a laser velocimeter, with a measured value of 103 g / min, within the allowable ±3% range. During the process, the temperature sensor collects molten pool temperature data at 0.5-second intervals. When a temperature fluctuation exceeding ±5℃ is detected, pressure fine-tuning is automatically triggered. The fine-tuning amplitude is calculated using the formula ΔP' = 0.2 × ΔT. For example, if a temperature drop of 7℃ is detected, an additional 1.4 MPa of pressure compensation is added. All adjustment commands are transmitted via the industrial bus using the Modbus-RTU protocol, with a response latency controlled within 50 milliseconds to ensure the real-time performance of the control loop.

[0151] S109. By updating the process parameters of the regional control through feedback instructions, the actuator of the dewaxing device is driven to adjust the heating power, airflow speed and pressure distribution, thereby generating a dynamic dewaxing scheme that adapts to complex structures.

[0152] Real-time feedback command data is acquired from sensors, and the zoning control requirements contained in the commands are parsed to obtain a zoning control parameter set. The control parameters of the dewaxing device are updated using this parameter set, and a preset mapping rule is used to determine the heating power adjustment value. The heating module of the dewaxing device is driven according to the heating power adjustment value. If the heating power exceeds a preset threshold, the power increment is reduced to achieve a stable heating distribution. Airflow velocity sensor data is acquired and, combined with the zoning control parameter set, a linear regression algorithm is used to determine the optimal airflow velocity value. The fan module of the dewaxing device is adjusted based on the optimized airflow velocity value. If the velocity fluctuation exceeds a preset range, the velocity curve is smoothed to determine a uniform airflow field. Pressure distribution data is acquired from pressure sensors and, combined with a complex structural geometric model, finite element analysis is used to obtain an optimized pressure distribution scheme. Based on the heating distribution, airflow field, and pressure distribution optimization scheme, the zoning control parameters are integrated to generate a dynamic dewaxing scheme adapted to complex structures.

[0153] For example, during the control process of the dewaxing device, the temperature, airflow speed and pressure data of each area are first collected in real time by sensors. For example, the temperature range is 60°C to 80°C and the airflow speed is 0.

[0154] The speed ranges from 5 m / s to 1.5 m / s, with a pressure distribution of 0.1 MPa to 0.3 MPa. This data is transmitted to the central control system via the Internet of Things (IoT). The system employs a feedback control model based on the PID algorithm to analyze and calculate the collected data.

[0155] For example, when the temperature in a certain area is below the set value of 65℃, the system uses a PID algorithm to calculate the adjustment amount of the heating power, increasing it from 500W to 600W to ensure the temperature remains stable within the target range. Simultaneously, based on real-time airflow velocity data, the system uses a fuzzy control algorithm to adjust the fan speed, optimizing the airflow velocity from 0.8m / s to 1.2m / s to improve dewaxing efficiency. Regarding pressure distribution, the system uses a finite element analysis model, combined with the characteristics of the area's structure, to dynamically adjust the opening of the pressure valve, adjusting the pressure from 0.2MPa to 0.25MPa to ensure uniform wax removal. Furthermore, the system uses machine learning algorithms to train on historical data, generating dynamic dewaxing schemes adapted to complex structures. For example, for workpieces of different shapes, it automatically adjusts the combination parameters of heating power, airflow velocity, and pressure distribution to ensure optimal dewaxing results. The entire process is automated, requiring no manual intervention, ensuring precise adjustment of process parameters and efficient operation of the dewaxing device.

[0156] S1010. According to the dynamic dewaxing scheme, a multi-channel control system is used to simultaneously process the dewaxing process of thin-walled areas, thick-walled parts and transition areas, and generate high-precision casting process execution data.

[0157] Geometric feature data of thin-walled regions, thick-walled areas, and transition regions are acquired through a multi-channel control system to determine dewaxing parameters for each region. Thermal conductivity characteristics of each region are extracted from the geometric feature data to obtain a thermal distribution model for the region division. Based on the thermal distribution model, a synchronous processing algorithm is used to adjust the dewaxing rate of the multi-channel control system, generating a process execution sequence for each region. If the rate deviation of the process execution sequence exceeds a preset threshold, the control system parameters are adjusted through an iterative optimization algorithm to obtain a stable dewaxing rate. Time-series data is extracted from the stable dewaxing rate to generate high-precision casting process data. For the process data, a data verification algorithm is used to detect outliers, obtaining the final process execution data.

[0158] For example, in a dynamic dewaxing scheme, a multi-channel control system ensures that the dewaxing process in thin-walled areas, thick-walled areas, and transition areas is carried out synchronously by monitoring and adjusting temperature, pressure, and time parameters in real time.

[0159] For example, for thin-walled areas, the system employs a rapid heating strategy, increasing the temperature from room temperature to 80°C at a rate of 5°C per minute and maintaining this temperature for 10 minutes to ensure complete wax melting. For thicker areas, the system uses a gradual heating strategy, increasing the temperature to 120°C at a rate of 2°C per minute and maintaining this temperature for 20 minutes to ensure complete wax removal. The dewaxing process in transitional areas is optimized through an algorithm, employing a segmented heating strategy: increasing the temperature to 90°C at a rate of 3°C per minute for the first 5 minutes, then increasing it to 110°C at a rate of 1°C per minute and maintaining this temperature for 15 minutes to achieve uniform dewaxing. The system dynamically adjusts the heating power using a PID control algorithm by collecting temperature data from each area in real time, combined with a preset dewaxing curve, to ensure temperature fluctuations are controlled within ±1°C. Simultaneously, the system monitors pressure changes during the dewaxing process using pressure sensors. When the pressure exceeds a set threshold, it automatically adjusts the opening of the exhaust valve to ensure the pressure remains stable between 0.1MPa and 0.15MPa. Through precise control of the multi-channel control system, the generated high-precision casting process execution data includes temperature, pressure, and time curves for each region, as well as real-time status information during the dewaxing process, providing data support for the optimization of subsequent casting processes.

[0160] S1011. Extract the dewaxing completion rate and shell stress distribution of each region from the process execution data. If the completion rate reaches the preset standard and the stress is lower than the safety threshold, terminate the dewaxing process and obtain the final dewaxing result.

[0161] Acquire process data and extract the dewaxing completion rate and shell stress distribution for each region to obtain an initial analysis dataset. Using this dataset, calculate the mean and standard deviation of the dewaxing completion rate for each region to obtain a completion metric. Using this metric, if the completion rate of each region reaches a preset threshold, a completion pass indicator is generated. Extract the shell stress distribution from the initial analysis dataset, calculate the maximum stress value and regional differences to obtain stress distribution characteristics. Based on these characteristics, if the maximum stress is below a safety threshold and the regional differences are less than a preset range, a stress safety indicator is generated. Using the completion pass indicator and the stress safety indicator, perform a logical AND operation; if both are true, a termination signal is generated. Based on the termination signal, control the process system to stop the dewaxing process, obtaining the final dewaxing result dataset.

[0162] For example, during the dewaxing process, temperature, pressure and time data of each area are first collected by sensors, and the degree of dewaxing is calculated using these data.

[0163] For example, if the temperature in a certain area reaches 120℃ and the pressure is maintained at 0.5MPa for 30 minutes, the dewaxing completion rate for that area is calculated to be 95% using the formula "Completion Rate = (Actual Temperature / Preset Temperature) × (Actual Pressure / Preset Pressure) × (Actual Time / Preset Time)". Next, the stress distribution of the shell is simulated using the Finite Element Analysis (FEA) algorithm. Assuming the stress distribution of the shell is σ(x,y) = 50MPa × e^(-0.1x) × sin(0.2y), where x and y are the coordinates of the shell, the analysis reveals that the maximum stress on the shell is 45MPa, which is below the safety threshold of 50MPa. Combining the dewaxing completion rate and stress distribution data, the system automatically determines that the dewaxing process in that area has reached the preset standard and the stress is below the safety threshold. Therefore, the dewaxing process is terminated, yielding the final dewaxing result. The entire process is automated, requiring no manual intervention, ensuring the accuracy and consistency of the process.

[0164] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for dewaxing wax patterns in investment casting, characterized in that, The method includes: The three-dimensional geometric data of the wax model is obtained by scanning equipment, and a digital model containing thin-walled areas, thick parts and internal cavity structures is constructed. A geometric feature distribution map of the wax model is generated, and the thickness, curvature and volume parameters of each area are determined. Based on the geometric feature distribution map of the wax model, a clustering algorithm is used to divide the wax model into thin-walled regions, thick-walled parts, and transition regions, generating a region segmentation mask to obtain the basic data for regional control. The thickness parameters of thin-walled regions and the volume parameters of thick-walled regions are extracted from the region segmentation mask. The thermal conductivity coefficient and wax melting threshold of each region are calculated to determine the initial process parameters for dewaxing by region. If the thickness of the thin-walled region is less than the preset threshold, the heating power of the region is reduced and airflow assistance is increased to generate process parameters for thin-walled region protection. If the volume of the thick part is greater than the preset threshold, the heating time is extended and the local pressure is increased to generate process parameters for the thick part treatment and obtain a dynamic parameter control scheme. Infrared sensors are used to collect the surface temperature field and residual wax distribution of the wax model in real time. Combined with a dynamic parameter control scheme, a real-time data stream for monitoring the dewaxing progress is generated. The temperature gradient and wax melting rate are extracted from the real-time data stream. If the temperature gradient exceeds the preset range, the heating power and airflow speed are adjusted. If the melting rate of the wax material is lower than the preset threshold, the local pressure is increased, and a feedback instruction for adjusting the process parameters is generated. By updating the process parameters of the regional control through feedback instructions, the actuators of the dewaxing unit are driven to adjust the heating power, airflow speed and pressure distribution, thereby generating a dynamic dewaxing scheme that adapts to complex structures. According to the dynamic dewaxing scheme, a multi-channel control system is used to simultaneously process the dewaxing process of thin-walled areas, thick-walled areas and transition areas, and generate high-precision casting process execution data. Extract the dewaxing completion rate and shell stress distribution of each region from the process execution data. If the completion rate reaches the preset standard and the stress is lower than the safety threshold, the dewaxing process is terminated, and the final dewaxing result is obtained.

2. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, The process involves acquiring three-dimensional geometric data of the wax model using a scanning device, constructing a digital model that includes thin-walled regions, thick-walled areas, and internal cavity structures, generating a geometric feature distribution map of the wax model, and determining the thickness, curvature, and volume parameters of each region, including: The three-dimensional geometric data of the wax model is collected by scanning equipment to generate a raw dataset containing point cloud information and to ensure data integrity. If the original dataset contains noise, a denoising algorithm is used to process the original dataset to generate smooth point cloud data and obtain geometric information without noise interference. Based on smooth point cloud data, a digital model containing thin-walled regions, thick parts, and internal cavity structures is constructed, a three-dimensional mesh structure is generated, and the topological integrity of the model is determined. Geometric features are extracted from the 3D mesh structure, and the thickness, curvature, and volume parameters of each region are calculated to obtain a set of feature parameters. A visualization algorithm is used to process the feature parameter set, generate a geometric feature distribution map, and determine the spatial distribution of parameters in each region. If the thickness parameter in the distribution map exceeds the preset threshold, the geometric features of the thin-walled region and the thick part are locally optimized and adjusted to generate an updated feature parameter set. By regenerating the geometric feature distribution map using the updated feature parameter set, the final distribution of thickness, curvature, and volume parameters is determined.

3. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, Based on the geometric feature distribution map of the wax model, a clustering algorithm is used to divide the wax model into thin-walled regions, thick-walled areas, and transition regions, generating a region segmentation mask to obtain the basic data for region-based control, including: The structural information of the wax model is extracted by geometric feature distribution, and feature vectors are generated by feature extraction algorithm to obtain the geometric description data of the wax model; Based on the geometric description data of the wax model, a clustering algorithm was used to group the feature vectors and determine the initial division results of the thin-walled region, the thick region, and the transition region. If the region boundaries are discontinuous in the initial partitioning result, the grouping of feature vectors is adjusted through spatial adjacency analysis to obtain optimized region partitioning data; Based on the optimized region segmentation data, a region segmentation mask is generated to determine the pixel-level distribution of each region. By using a region segmentation mask, the boundary features of thin-walled regions, thick regions, and transition regions are extracted to obtain region boundary description data. Based on the regional boundary description data, calculate the geometric characteristics of each region and generate regional control data; By mapping the process parameters of the wax mold structure to the regional control data, the regional process control parameters are obtained.

4. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, The process of extracting thickness parameters of thin-walled regions and volume parameters of thick-walled regions from the region segmentation mask, calculating the thermal conductivity coefficient and wax melting threshold of each region, and determining the initial process parameters for region-specific dewaxing includes: The geometric features of thin-walled regions and thick-walled regions are obtained from the region segmentation mask by the segmentation algorithm, and the thin-walled thickness parameters and the volume parameters of the thick-walled regions are obtained. Based on the thin-wall thickness parameters and the volume parameters of the thick-walled parts, the thermal conductivity coefficient of each region is calculated using the finite element analysis method to obtain the thermal conductivity coefficient distribution; If the thermal conductivity coefficient distribution exceeds the preset threshold range, the region is re-divided, the updated thin-wall thickness parameters and the volume parameters of the thick parts are obtained, the thermal conductivity coefficient is recalculated, and the final thermal conductivity coefficient distribution is determined. By using the thermal conductivity coefficient distribution and material property database, the melting threshold of wax material in each region was calculated, and the melting threshold distribution was obtained; Based on the melting threshold distribution, a clustering algorithm is used to classify the regions and obtain a priority sequence for dewaxing in each region; The initial process parameters for each region are calculated using priority sequence and thermal conductivity coefficient distribution, resulting in the initial parameter configuration for dewaxing in different regions. If the initial parameter configuration does not match the preset process constraints, the regional classification weights are adjusted, the initial process parameters are recalculated, and the final initial parameter configuration is determined.

5. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, If the thickness of the thin-walled region is less than a preset threshold, the heating power of that region is reduced and airflow assistance is increased to generate process parameters for thin-walled region protection, including: The thickness data of the thin-walled region is obtained by the sensor, and it is determined whether the thickness is lower than a preset threshold to obtain the thickness detection result. If the thickness detection result is lower than the preset threshold, the heating power reduction ratio is calculated based on the thickness data, and the adjusted heating power value is determined. Based on the adjusted heating power value and the location of the thin-walled area, a corresponding power control command is generated; The airflow control module obtains the current position of the thin-walled region, calculates the required airflow assist intensity, and determines the airflow control parameters. An initial set of process parameters is generated by using a preset process parameter generation model and integrating power control commands and airflow control parameters. For the initial set of process parameters, an optimization algorithm is executed to adjust the ratio between parameters, resulting in the final set of process parameters. The real-time monitoring module acquires the state of the thin-walled region after the process parameters are executed, determines whether the protection mechanism requirements are met, and generates feedback data.

6. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, If the volume of the thick part exceeds a preset threshold, the heating time is extended and the local pressure is increased to generate process parameters for the thick part treatment, resulting in a dynamic parameter control scheme, including: If the volume data of the thick part exceeds the preset threshold, its three-dimensional geometric features are obtained through a stereomicroscope to generate a structural description of the thick part. Based on the structural description, the heat conduction distribution in the thick parts was calculated using finite element analysis to obtain temperature field distribution data. If there are cold spot regions in the temperature field distribution data, the heating time parameter is adjusted through an iterative optimization algorithm to generate the optimized time parameter; Based on the optimized time parameters, obtain stress distribution data for the thick parts and determine whether the local pressure meets the preset conditions. If the local pressure is lower than the preset condition, the pressure parameters are adjusted incrementally through numerical simulation to obtain the updated pressure control parameters; By integrating and optimizing the time parameters and pressure control parameters, a set of process parameters for the heavy parts is generated; Based on the set of process parameters, a dynamic control scheme is output using a real-time monitoring system.

7. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, The method employs infrared sensors to collect real-time temperature fields and residual wax distribution on the wax mold surface, and combines this with a dynamic parameter control scheme to generate a real-time data stream for monitoring the dewaxing progress, including: The initial temperature distribution and wax distribution are obtained by collecting surface temperature field data and residual wax distribution data of the wax model using an infrared sensor. Based on the initial temperature distribution and wax quantity distribution, a convolutional neural network algorithm is used to analyze the spatial correlation between the temperature field and the wax quantity, and to determine the temperature-wax quantity mapping relationship. If there are local high-temperature areas in the temperature-wax quantity mapping relationship, the heating power is adjusted through a dynamic parameter control scheme to obtain the optimized parameter configuration; With optimized parameter configuration, the temperature field and residual wax distribution on the wax model surface are updated using a real-time acquisition module, generating updated temperature and wax distributions. Based on the updated temperature distribution and wax quantity distribution, the mean square error algorithm is used to compare the current wax quantity with the preset threshold to determine the dewaxing progress status. If the dewaxing progress does not reach the preset threshold, the heating power is readjusted through the dynamic parameter control module to obtain a new parameter configuration; With new parameter configuration, the data stream generation module integrates temperature distribution, wax quantity distribution, and progress status to generate a real-time data stream.

8. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, The step of extracting temperature gradient and wax melting rate from real-time data stream, and adjusting heating power and airflow speed if the temperature gradient exceeds a preset range, includes: Temperature gradient and wax melting rate are obtained from real-time data streams, and environmental data is collected using a sensor array to obtain the raw dataset; For the original dataset, the mean filtering algorithm is used to process the temperature data to determine the smoothed temperature gradient value; If the smoothed temperature gradient value exceeds the preset range, the degree of anomaly is determined by a logistic regression algorithm to obtain the anomaly level. Based on the level of abnormality, a proportional control algorithm is used to adjust the heating power and determine a new power setpoint. By obtaining the associated airflow velocity parameters through the new power setting, it is determined whether the airflow velocity needs to be adjusted synchronously, and the velocity adjustment value is obtained. For the speed adjustment value, digital signal processing technology is used to update the airflow control unit to determine the final airflow speed parameters; The final airflow velocity parameters and heating power values ​​are obtained, the system status is monitored in real time, and the updated temperature gradient and melting rate are obtained.

9. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, If the melting rate of the wax material is lower than a preset threshold, the local pressure is increased, and a feedback instruction for adjusting process parameters is generated, including: If the melting rate of the wax is lower than the preset threshold, the real-time melting rate data is obtained through the sensor, and the deviation between the rate and the threshold is judged by the comparison algorithm to obtain the deviation value. Based on the deviation value, the corresponding pressure increment is queried using a preset mapping table to determine the local pressure adjustment amount; By adjusting the pressure, a pressure control command is generated and output to the pressure regulating device to obtain the updated pressure status. The updated pressure status is obtained, and combined with real-time melting rate data, the support vector machine algorithm is used to determine the stability of process parameters and obtain stability index. If the stability index is lower than the preset standard, the process parameters are optimized using the gradient descent algorithm to generate new parameter adjustment instructions; Based on the new parameter adjustment instructions, update the process parameter configuration, output to the control system, and obtain the optimized process status; By optimizing the process status, real-time monitoring data is obtained to determine whether the melting rate of the wax material has reached the preset threshold, and the final process parameter instructions are generated.

10. The method for dewaxing wax patterns in investment casting according to claim 1, characterized in that, The process parameters controlled by regional adjustments are updated via feedback commands, driving the actuators of the dewaxing device to adjust heating power, airflow velocity, and pressure distribution, thereby generating a dynamic dewaxing scheme adapted to complex structures. This includes: The system acquires real-time feedback command data from sensors, parses the zonal control requirements contained in the commands, and obtains the zonal control parameter set. The control parameters of the dewaxing device are updated by updating the partition control parameter set, and the heating power adjustment value is determined by using a preset mapping rule; The heating module of the dewaxing device is driven according to the heating power adjustment value. If the heating power exceeds the preset threshold, the power increment is reduced to obtain a stable heating distribution. By acquiring airflow velocity sensor data and combining it with the zonal control parameter set, a linear regression algorithm is used to determine the optimal airflow velocity value. The blower module of the dewaxing device is adjusted by optimizing the airflow velocity. If the velocity fluctuation exceeds the preset range, the velocity curve is smoothed to determine a uniform airflow field. Pressure distribution data is obtained from pressure sensors, and combined with a complex structural geometric model, finite element analysis is used to obtain an optimized pressure distribution scheme. Based on the optimization schemes of heating distribution, airflow field and pressure distribution, and by integrating the zoning control parameters, a dynamic dewaxing scheme adapted to complex structures is generated.

Images