An intelligent control system for casting irregularly shaped metal parts

Abstract

This invention relates to the field of intelligent manufacturing technology, and more particularly to an intelligent control system for casting irregularly shaped metal parts. The system includes: an acquisition module, a partitioning module, a data collection module, a judgment module, an identification module, and a control module. This invention extracts geometric feature parameters based on a three-dimensional model of the irregularly shaped part, constructs risk coefficients to divide sensitive areas and determine their types; it deploys sensors based on the differentiated regional types to collect data; it constructs a defect index based on data deviations, and determines risks by combining regional types and material properties; it identifies defect types based on regional types and deviation patterns; and it adjusts casting parameters in a closed loop. This effectively solves the technical problems in existing technologies, such as low defect positioning accuracy, delayed process response, and insufficient reliability in judgment under complex working conditions due to the lack of spatial resolution and online adaptive correction capabilities for irregularly shaped geometric features.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to an intelligent control system for casting irregularly shaped metal parts. Background Technology

[0002] Ferrous metal irregular-shaped components are widely used in high-end equipment manufacturing fields such as aerospace and energy power. Unlike light metals such as aluminum alloys, ferrous metal casting exhibits significant material specificity: high melting point, concentrated latent heat of solidification, extremely high temperature sensitivity near the solidus line, and generally low thermal conductivity, resulting in prominent local thermo-mechanical coupling effects. When these material properties are coupled with the geometric features of irregular structures, such as abrupt changes in wall thickness, abrupt changes in curvature, and concentrated local heat points, defects such as shrinkage cavities, hot cracks, and cold shuts frequently occur in specific geometric regions. Traditional process control methods, mostly based on the assumption of uniform solidification or experience in aluminum-magnesium alloy casting, are insufficient to effectively address the complex working conditions of ferrous metal irregular-shaped components. Therefore, developing adaptive sensing and intelligent control methods for casting processes that are oriented towards the geometric characteristics of ferrous metal irregular-shaped components and fully consider their material specificity has become a key technical problem urgently needing to be solved in this field.

[0003] Chinese Patent Publication No. CN116975770A discloses a defect detection method for cast steel, applied to a defect detection system for cast steel. The system is communicatively connected to a cast steel parts conveying channel, which includes a diversion robot. The method includes: acquiring cast steel model information, optimizing the cast steel process based on the cast steel process type, and generating a baseline of cast steel process parameters; monitoring the timing information of the cast steel execution process parameters, performing a deviation analysis between the cast steel and the baseline of cast steel process parameters, and generating a cast steel deviation coefficient; when the cast steel deviation coefficient is greater than or equal to a cast steel deviation coefficient threshold, extracting abnormal parameter attributes and parameter abnormal time zones; performing correlation analysis based on the abnormal parameter attributes and parameter abnormal time zones to obtain a defect prediction type; activating the diversion robot to divert cast steel parts with a deviation coefficient greater than or equal to the cast steel deviation coefficient threshold to a specific quality inspection channel; and when a cast steel part is located in a preset area of ​​the specific quality inspection channel, performing targeted defect detection based on the defect prediction type.

[0004] However, this existing technology has the following drawbacks: it relies on the comparison of the overall process parameter baseline and the global deviation coefficient, lacks the spatial resolution capability for the geometric features of irregularly shaped components, and is difficult to identify the precise geometric location of defects; at the same time, it relies on an open-loop defect diversion and offline directional detection mechanism, which cannot adjust process parameters in real time during the casting process, and is prone to batch defects due to response lag; in addition, it relies on single-dimensional parameter time-series deviation analysis, and lacks a multi-dimensional dynamic correction mechanism for the specific characteristics of ferrous metal materials, such as sensitivity to solidification temperature difference and low thermal conductivity, which is prone to misjudgment or missed judgment under complex thermo-mechanical coupling conditions; furthermore, its sensor deployment strategy does not consider the advantage of repeatable and accurate sensor positioning in metal molds or precision casting, nor does it provide a degradation solution for the problem of poor sensor accessibility in sand casting, which further limits its applicability and monitoring accuracy in the casting of irregularly shaped components. Summary of the Invention

[0005] To address this, the present invention provides an intelligent control system for casting irregularly shaped metal parts. This system overcomes the technical problems in the prior art, such as low defect location accuracy, delayed process response, and insufficient reliability in complex working conditions, caused by the lack of spatial resolution and online adaptive correction capabilities for irregularly shaped geometric features, through a risk zoning, differentiated monitoring, and closed-loop control mechanism driven by geometric physical analysis.

[0006] To achieve the above objectives, the present invention provides an intelligent control system for casting irregularly shaped metal parts, comprising: The acquisition module is used to acquire the curvature of the spherical shape with equal volume of ferrous metal, the hot spot volume of each grid cell in the grid cells divided on it, the rate of curvature change, the actual curvature and the gate distance. The partitioning module is used to determine several encryption zones based on the threshold comparison results of the risk coefficient of each grid cell, and to determine the encryption type of each encryption zone. The risk coefficient is determined based on the hot spot volume, the rate of curvature change, the distortion factor and the gate distance. The distortion factor is determined based on the actual curvature and the curvature of the equal-volume sphere. The acquisition module is used to acquire in real time the pouring monitoring data of each encryption zone during the pouring process of ferrous metal irregular parts based on preset pouring temperature and preset pouring speed, wherein the pouring monitoring data is determined based on each encryption type. The judgment module is used to determine whether there is a defect risk in each encryption zone based on the comparison result between the correction index of each encryption zone and the corresponding preset judgment threshold. The correction index is determined based on the defect index, the encryption type and the solidification temperature difference of the ferrous metal irregular part. The defect index is determined based on the degree of deviation, total deviation and deviation rate of the casting monitoring data of each encryption zone. An identification module is used to determine the defect type of each encryption zone based on the defect risk, the encryption type, and the casting monitoring data. The control module is used to adjust one or more of the following during the pouring process: pouring temperature, pouring speed, holding time, and pouring pressure, based on the defect type and the correction index.

[0007] Furthermore, the acquisition module includes: The meshing submodule is used to discretize the three-dimensional model of the ferrous metal irregular part to obtain several mesh elements. The hot spot analysis submodule is used to count the volume of the grid cell region where the solidification time exceeds the preset solidification threshold, so as to obtain the hot spot volume; wherein, the solidification time is determined based on the preset pouring temperature of the ferrous metal part, the initial temperature of the mold, and the thermophysical parameters of the ferrous metal material. The curvature analysis submodule is used to calculate the actual curvature and the rate of change of curvature of each grid cell; The equal-volume sphere curvature calculation submodule is used to calculate the curvature of a sphere with the same local volume as each grid cell, so as to obtain the curvature of the equal-volume sphere. The gate distance calculation submodule is used to obtain the gate distance based on the flow path length of each grid cell from the preset gate mark.

[0008] Furthermore, the partitioning module includes: The distortion factor calculation submodule is used to calculate the distortion factor of each grid cell based on the actual curvature of each grid cell and the curvature of the equal-volume sphere. The risk calculation submodule is used to perform min-max normalization on the hot spot volume, curvature change rate, gate distance and distortion factor of each grid cell to obtain normalized hot spot volume, normalized curvature change rate, normalized gate distance and normalized distortion factor, and then combine them with the corresponding preset weight coefficients for weighted fusion to obtain the risk coefficient of each grid cell. The type determination submodule is used to determine the encryption type of each grid cell based on the comparison results of the normalized hot spot volume, normalized rate of curvature change, normalized distortion factor, and normalized gate distance of each grid cell with the corresponding preset threshold; wherein the encryption type includes at least hot spot type, stress type, and cold shut type. The region determination submodule is used to determine the continuous grid cell region with a risk coefficient greater than the preset encryption threshold as the encryption zone.

[0009] Furthermore, the distortion factor calculation submodule includes: A curvature difference calculation unit is used to calculate the difference between the actual curvature of the mesh cell and the curvature of the equal-volume sphere to obtain the curvature difference value. A distortion factor calculation unit is used to divide the curvature difference by the curvature of the equal-volume sphere to obtain the distortion factor.

[0010] Furthermore, the type determination submodule includes: The out-of-standard factor determination unit is used to determine the out-of-standard factor based on the normalized hot spot volume, the normalized rate of change of curvature, the normalized distortion factor, the normalized gate distance, and the corresponding preset threshold. A single-condition determination unit is used to determine the encryption type of the corresponding grid cell based on the number of the excess factors when the number of excess factors is 1. A multi-condition arbitration unit is used to determine the encryption type according to a preset priority order when the number of the excess factors is greater than 1.

[0011] Furthermore, the acquisition module includes: The deployment strategy submodule is used to differentiate the sensor type and sampling frequency according to the encryption type; The partition acquisition submodule is used to acquire the casting monitoring data in real time at the sampling frequency corresponding to each sensor, and associate and store the casting monitoring data with the corresponding encrypted zone; the casting monitoring data includes at least the first temperature data of the hot-spot type encrypted zone, the elastic wave signal of the stress type encrypted zone, and the second temperature data and pressure data of the cold-spot type encrypted zone.

[0012] Furthermore, the determination module includes: The deviation calculation submodule is used to calculate the degree of deviation, total deviation and deviation rate of the pouring monitoring data of each sensor in each encrypted zone relative to the preset expected value. The index calculation submodule is used to take the maximum value of the sensor defect indices of all sensors in each encryption zone as the encryption zone defect index of each encryption zone; wherein, the sensor defect index is the maximum value of the deviation degree, the total deviation and the deviation rate; The correction submodule is used to correct the defect index of the encryption area based on the encryption type and the solidification temperature difference of the ferrous metal irregular part, so as to obtain the correction index. The risk assessment submodule is used to compare the correction index with the corresponding preset assessment threshold. When the correction index is greater than the preset assessment threshold, it is determined that there is a defect risk in the corresponding encryption area.

[0013] Furthermore, the deviation calculation submodule includes: A deviation calculation unit is used to calculate the ratio of the absolute value of the difference between the pouring monitoring data and the preset expected value to the preset expected value, so as to obtain the deviation degree; The deviation calculation unit is used to calculate the cumulative value of the deviation degree from the start time of pouring to the current time, so as to obtain the total deviation. The deviation rate calculation unit is used to calculate the quotient of the difference in deviation between two adjacent samples divided by the sampling time interval to obtain the deviation rate.

[0014] Furthermore, the identification module includes: The hot spot identification submodule is used to extract the temporal features of the first temperature data when the encryption type is hot spot type. If it meets the preset hole shrinkage pattern features, the defect type of the corresponding encryption area is identified as a hole shrinkage defect. The stress identification submodule is used to extract the frequency characteristics of the elastic wave signal when the encryption type is stress type. If it meets the preset thermal cracking mode characteristics, the defect type of the corresponding encryption area is identified as a thermal cracking defect. The cold shut identification submodule is used to extract the temporal features of the second temperature data when the encryption type is cold shut type, and if it meets the preset cold shut mode features, then the defect type of the corresponding encryption area is identified as a cold shut type defect; and to extract the waveform features of the pressure data, and if it meets the preset insufficient filling mode features, then the defect type of the corresponding encryption area is identified as an insufficient filling type defect.

[0015] Furthermore, the control module includes: The grading submodule is used to classify the severity of defects as mild, moderate, or severe based on the correction index. The shrinkage cavity control submodule is used to adjust the pouring speed and / or holding time in the pouring parameters when the defect type is a shrinkage cavity and the severity is mild, performing one of reducing the pouring speed and extending the holding time; when the severity is moderate or above, performing both simultaneously. The hot crack control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is hot crack and the severity is mild, performing either reducing the pouring temperature or reducing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The cold shut control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is a cold shut defect and the severity is mild, by increasing the pouring temperature and increasing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The underfill control submodule is used to adjust the pouring speed and / or pouring pressure in the pouring parameters when the defect type is underfill defect and the severity is mild, by increasing the pouring speed and increasing the pouring pressure; when the severity is moderate or above, both are executed simultaneously.

[0016] Compared with existing technologies, the advantages of this invention lie in its six-level progressive intelligent control architecture, which includes geometric feature analysis, risk zoning, differentiated acquisition, multi-dimensional deviation judgment, dual-correction identification, and hierarchical regulation. This architecture enables prior zoning and online closed-loop control of the casting process for ferrous metal irregularly shaped components. First, based on the three-dimensional model of the irregularly shaped component, the volume of hot spots, rate of curvature change, distortion factor, and gate distance are extracted. The distortion factor is then constructed to transform geometric distortion into a dimensionless physical index, enabling prior location of high-risk defect areas. Based on the risk coefficient, the density zone and density type are determined. Through normalized weighted fusion and physical priority arbitration, mesh units are accurately classified into density types such as hot spot type, stress type, or cold shut type. This ensures a one-to-one correspondence between sensor deployment and physical causes. Sensor configuration and data acquisition are performed using the density zone as the unit, avoiding information redundancy from indiscriminate acquisition and reducing sensor deployment density and the complexity of the data acquisition system. This method extracts abnormal features from three dimensions: degree of deviation, total deviation, and deviation rate. It then combines encryption type and solidification temperature difference for dual correction, mapping the original monitoring data into a risk metric that closely approximates physical reality. This overcomes the shortcomings of single-dimensional deviation analysis, which is prone to misjudgment and omission under complex thermo-mechanical coupling conditions. Based on the correction index, the severity of defects is classified and graded control is implemented. Mild defects are eliminated with minimal intervention, while severe defects are directly addressed with strong control, forming a complete closed loop of identification, judgment, control, and iteration. This effectively solves the technical problems of low defect location accuracy, lag in process response, and insufficient reliability in judgment under complex conditions caused by the lack of spatial resolution capabilities and online adaptive correction capabilities for irregular geometric features in existing technologies. Attached Figure Description

[0017] Figure 1 This is a structural diagram of the intelligent control system used for casting irregularly shaped metal parts in this embodiment; Figure 2 This is a flowchart illustrating how the type determination submodule in this embodiment determines the encryption type. Figure 3 This is a flowchart illustrating the deployment strategy and data acquisition process of the acquisition module in this embodiment. Figure 4 This is a flowchart of the deviation calculation and double correction process of the determination module in this embodiment. Detailed Implementation

[0018] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0019] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0020] Please see Figure 1 The diagram shown is a structural diagram of the intelligent control system for casting irregularly shaped metal parts according to this embodiment. This embodiment provides an intelligent control system for casting irregularly shaped metal parts, including: The acquisition module is used to acquire the curvature of the spherical shape with equal volume of ferrous metal, the hot spot volume of each grid cell in the grid cells divided on it, the rate of curvature change, the actual curvature and the gate distance. The partitioning module is used to determine several encryption zones based on the threshold comparison results of the risk coefficient of each grid cell, and to determine the encryption type of each encryption zone. The risk coefficient is determined based on the hot spot volume, the rate of curvature change, the distortion factor and the gate distance. The distortion factor is determined based on the actual curvature and the curvature of the equal-volume sphere. The acquisition module is used to acquire in real time the pouring monitoring data of each encryption zone during the pouring process of ferrous metal irregular parts based on preset pouring temperature and preset pouring speed, wherein the pouring monitoring data is determined based on each encryption type. The judgment module is used to determine whether there is a defect risk in each encryption zone based on the comparison result between the correction index of each encryption zone and the corresponding preset judgment threshold. The correction index is determined based on the defect index, the encryption type and the solidification temperature difference of the ferrous metal irregular part. The defect index is determined based on the degree of deviation, total deviation and deviation rate of the casting monitoring data of each encryption zone. An identification module is used to determine the defect type of each encryption zone based on the defect risk, the encryption type, and the casting monitoring data. The control module is used to adjust one or more of the following during the pouring process: pouring temperature, pouring speed, holding time, and pouring pressure, based on the defect type and the correction index.

[0021] In this embodiment, the ferrous metal irregular-shaped parts refer to metallic materials with iron as the base element and a carbon content typically ranging from 0.02% to 4.0%, mainly including cast steel, cast iron, and their alloys. These are parts formed through casting processes, exhibiting irregular, asymmetrical, or complex curved surface features. Specifically, the geometric characteristics of irregular-shaped parts include: drastic variations in wall thickness, with the wall thickness ratio between adjacent areas reaching 3:1 or higher; abrupt changes in surface curvature, with a curvature change rate exceeding 50%; concentrated local heat points, with solidification time exceeding the average solidification time by more than 30%; asymmetrical structure, lacking a plane of symmetry or axis of symmetry; and complex internal cavities, containing multiple flow channels or irregularly shaped cavities. These parts are particularly suitable for casting scenarios with stable mold structures and where repeated sensor positioning is permitted, including but not limited to areas accessible by metal mold casting, precision casting, and sand casting.

[0022] In this embodiment, the 3D model of the ferrous metal irregular part is obtained through parametric modeling using computer-aided design software. Specifically, any 3D CAD software such as SolidWorks, CATIA, NX, or Creo is used. Based on the functional requirements, assembly constraints, and mechanical performance indicators of the irregular part, a 2D contour is defined by sketching, and then a 3D solid model is constructed using operations such as feature extrusion, rotation, lofting, sweeping, and surface modeling. For irregular parts with complex free-form surfaces, such as turbine housing flow channels or blade roots, non-uniform rational B-spline surface modeling technology is used to accurately describe the continuous change of surface curvature. After modeling, the model is exported as a STEP or STL format file. The model is required to be a closed solid, without holes or overlapping surfaces, and the volume deviation from the design value should not exceed 0.5%. This model is generated before the casting process and serves as input data for subsequent mesh generation and geometric feature parameter calculation.

[0023] In this embodiment, the preset pouring temperature and preset pouring speed are determined in advance based on the ferrous metal grade, the geometric characteristics of the irregular part, and the mold type, through recommended values ​​in the casting process manual or solidification filling simulation. During the pouring process, the control module can adjust one or more of the pouring temperature, pouring speed, holding time, and pouring pressure in real time according to the identified defect type. The preset pouring temperature and preset pouring speed are only used as a reference for the start time of pouring.

[0024] Through a six-level progressive architecture of geometric information acquisition, risk zoning, differentiated acquisition, defect judgment, defect identification and intelligent control, a complete closed loop from geometric features to process control is achieved. The acquisition module constructs a distortion factor based on the curvature of an equal-volume sphere, transforming geometric distortion into a dimensionless physical index to achieve prior location of high-risk defect areas and determine the densification zone and densification type. The partitioning module constructs a risk coefficient through normalized weighted fusion and classifies grid units into densification types such as thermal knot type, stress type, or cold shut type based on physical priority, providing spatial basis and type labels for the acquisition module. The acquisition module configures sensors according to the densification type and collects data in densification zones as units, avoiding data redundancy from indiscriminate deployment and reducing system complexity. The judgment module extracts abnormal features from three dimensions: deviation degree, total deviation, and deviation rate, and performs double correction based on densification type and solidification temperature difference to obtain a correction index, overcoming the misjudgment and omission defects of single-dimensional deviation analysis. The identification module selects the corresponding feature dimension according to the densification type to achieve targeted identification of four types of defects: shrinkage cavity, thermal crack, cold shut, and insufficient filling. The control module classifies the severity of defects according to the correction index and performs graded control to achieve a balance between defect improvement effect and process stability. Through the collaborative work of the above six modules, the entire process from geometric analysis to closed-loop control is made intelligent, which effectively solves the technical problems of low defect positioning accuracy, lag in process response and insufficient reliability of judgment under complex working conditions caused by the lack of spatial resolution capability and online adaptive correction capability for irregular geometric features in the existing technology.

[0025] Specifically, the acquisition module includes: The meshing submodule is used to discretize the three-dimensional model of the ferrous metal irregular part to obtain several mesh elements. The hot spot analysis submodule is used to count the volume of the grid cell region where the solidification time exceeds the preset solidification threshold, so as to obtain the hot spot volume; wherein, the solidification time is determined based on the preset pouring temperature of the ferrous metal part, the initial temperature of the mold, and the thermophysical parameters of the ferrous metal material. The curvature analysis submodule is used to calculate the actual curvature and the rate of change of curvature of each grid cell; The equal-volume sphere curvature calculation submodule is used to calculate the curvature of a sphere with the same local volume as each grid cell, so as to obtain the curvature of the equal-volume sphere. The gate distance calculation submodule is used to obtain the gate distance based on the flow path length of each grid cell from the preset gate mark.

[0026] The mesh generation submodule discretizes the acquired 3D model of the ferrous metal irregular part. This embodiment employs an adaptive tetrahedral mesh generation algorithm to divide the model into several finite volume mesh elements. The mesh element size is dynamically adjusted according to the geometric complexity of the irregular part: in areas with gentle curvature changes, such as large planes, the target size is set to 5mm; in areas with drastic curvature changes, such as sharp corners and thin-walled transitions, the target size is set to 0.5mm. Each mesh element stores its node coordinates, volume, and relationships with adjacent elements.

[0027] The hot spot analysis submodule is used to identify the hot spot regions with the slowest solidification. The specific method is as follows: Based on the finite element method, the transient heat conduction equation is solved, and the preset pouring temperature of the ferrous metal irregular part, the initial mold temperature, and the thermophysical parameters of the ferrous metal material are input. Starting from the moment of completion of pouring, the cooling curve of each mesh element is calculated, and the moment when the temperature drops to the solidus temperature is recorded as the solidification time. The volume of all mesh element regions whose solidification time exceeds the preset solidification threshold is counted to obtain the hot spot volume.

[0028] In this embodiment, the preset solidification threshold is determined through the following steps: First, the solidification process of the ferrous metal irregular part is simulated based on the finite element method to obtain the cooling curve of each grid cell above the solidus temperature; second, the cooling rate of each grid cell is calculated, and the target grid cell region with a cooling rate lower than the preset critical cooling rate is extracted; finally, in the target grid cell region, the minimum solidification time is selected as the preset solidification threshold to accurately locate the core hot spot region in the casting with the slowest heat dissipation and the most severe heat accumulation, ensuring that the calculation result of the hot spot volume truly reflects the physical solidification risk. The preset critical cooling rate refers to the cooling rate threshold that can distinguish between normal solidification and the high-risk area of ​​shrinkage cavity. It depends on the specific ferrous metal grade and the casting wall thickness, and is usually in the range of 0.3℃ / s to 1.0℃ / s. In this embodiment, the critical cooling rate is set to 0.5℃ / s, which can accurately identify the grid cells in the solid-liquid coexisting mushy region at this value, ensuring that the calculation result of the hot spot volume truly reflects the risk of shrinkage cavity formation.

[0029] In this embodiment, the preset pouring temperature refers to the initial temperature at which the molten metal is poured into the mold, and it depends on the specific grade of the ferrous metal part used and its liquidus temperature. The selection of the pouring temperature should ensure that the molten metal has good fluidity to complete the filling process, while avoiding excessive superheating that could lead to gas absorption, oxidation, and coarse grains. Typically, the pouring temperature is selected within the range of 50°C to 80°C above the liquidus temperature. For ductile iron parts, the preset pouring temperature is usually set to 1380°C to 1450°C.

[0030] In this embodiment, the initial mold temperature refers to the temperature of the mold before pouring begins, which depends on the mold material and the mold preheating process. For sand casting, the initial mold temperature is typically 20℃~40℃ (without preheating) or 150℃~250℃ (with preheating); for metal mold casting, it is typically preheated to 180℃~300℃ to mitigate the chilling effect. Depending on the mold type and the complexity of the irregular part, the initial mold temperature in this embodiment is set to 20℃~300℃. Specifically, for simple irregular parts with uniform wall thickness, a sand mold without preheating is used, with a temperature of 20℃~40℃; for irregular parts with thin walls or complex flow channels, to prevent cold shut defects, a preheated sand mold or metal mold is used, with a temperature of 150℃~300℃. This temperature serves as the initial boundary condition for the heat conduction equation, setting it as a uniform temperature field on the inner surface of the mold. This makes the solidification time calculation closer to the actual process, especially significant for predicting rapid solidification in thin-walled areas, avoiding misjudgments of hot spot regions due to deviations in temperature boundary conditions.

[0031] In this embodiment, the thermophysical parameters of ferrous metal materials refer to key physical quantities that describe the material's behavior during thermophysical processes, including thermal conductivity, specific heat capacity, density, solidus temperature, liquidus temperature, and latent heat of solidification. The values ​​of these parameters depend on the specific grade of the selected ferrous metal part and can be obtained from material property databases such as JMatPro, Thermo-Calc, or ASM Handbooks, or calibrated using differential scanning calorimetry or thermodynamic calculation software. In this embodiment, taking cast steel grade ZG270-500 as an example, the following parameter values ​​are used: thermal conductivity of 45 W / (m·K) (solid phase), specific heat capacity of 550 J / (kg·K), density of 7800 kg / m³ (room temperature), solidus temperature of 1440℃, liquidus temperature of 1480℃, and latent heat of solidification of 290 kJ / kg; for ductile iron, the solidus temperature is 1150℃, the liquidus temperature is 1180℃, and the latent heat of solidification is 270 kJ / kg. These parameters are used as inputs to the transient heat conduction equation to solve for the cooling curves and solidification times of each grid element. Calculations based on measured or database parameters of specific material grades allow the prediction error of solidification time to be controlled within ±10%, and the identification accuracy of hot spot volume is significantly higher than that of traditional methods using fixed empirical values. This provides a reliable physical basis for subsequent risk zoning and densification zone determination. At the same time, the parameters are searchable, measurable, and adjustable, ensuring that the technical solution has good adaptability and reproducibility under different ferrous metal materials and casting conditions.

[0032] The curvature analysis submodule is used to calculate the actual curvature and rate of change of curvature for each mesh cell. For each mesh cell, the surface nodes and all nodes in its neighborhood are first extracted. The neighborhood radius is taken as 2 to 3 times the average side length of the mesh cell to ensure sufficient data points for surface fitting. Then, the moving least squares method is used to fit a local quadratic surface function in the neighborhood. A Gaussian weighting function is used during fitting, with nodes closer to the center point having a larger weight. Based on the fitted surface, the first and second fundamental quantities of the surface are calculated, and then the two principal curvatures are solved. The average of the principal curvatures is taken as the actual curvature of the mesh cell. The rate of change of curvature is calculated by: on the fitted local surface, calculating the spatial gradient magnitude of the actual curvature along the tangent direction of the surface, and obtaining it by solving the partial derivative of the curvature distribution function.

[0033] The equal-volume sphere curvature calculation submodule is used to establish a curvature benchmark for an ideal shape without distortion. The specific calculation steps are as follows: First, a local volume threshold is set to obtain the total volume of the 3D model of the ferrous metal irregular part and determine the number of partitions; the total volume is divided by the number of partitions to obtain the local volume threshold. The number of partitions ranges from 5000 to 50000, and the specific value can be determined through convergence testing based on the geometric complexity of the irregular part to ensure that the local volume accurately reflects the local geometric features of the corresponding region. In this embodiment, for an irregular part with a total volume of 0.5 cubic meters, the number of partitions is determined to be 10000 after convergence testing, and the calculated local volume threshold is 5 × 10⁻⁶. -8 First, using the center of the current grid cell as the center of the sphere, expand the neighborhood layer by layer outwards, expanding one layer of adjacent grid cells at a time, accumulating the volume of the included grid cells until the accumulated volume first reaches or exceeds the local volume threshold. This accumulated volume is then taken as the local region volume represented by the current grid cell. Next, calculate the radius of a sphere with the same volume as this local region. Finally, calculate the curvature of the sphere with the same volume, which is equal to the reciprocal of the radius.

[0034] The gate distance calculation submodule is used to obtain the gate distance based on the flow path length of each grid cell from the preset gate mark. First, the gate position is marked in the 3D model. For a single-gate system, the geometric center point of the gate region is marked as the starting point; for a multi-gate system, the center points of all gate regions are marked to form the starting point set. Then, using the entire 3D solid domain of the casting cavity as the flow space, the equation describing the wavefront propagation distance field is solved on its discretized grid model. Its physical meaning is that the wavefront originating from the gate position expands outward at a uniform unit velocity. This embodiment uses the fast travel method to numerically solve this equation. This algorithm calculates the flow distance at each point within the entire cavity by simulating the outward expansion of the wavefront from the gate position. Finally, for each grid cell, the flow distance value corresponding to its center point is extracted as the gate distance for that grid cell.

[0035] By comprehensively characterizing the defect sensitivity of each region of the irregular part from three dimensions—thermal points, stress distortion, and flow ends—the subsequent partitioning module can formulate differentiated encryption and monitoring strategies for different physical causes, namely thermal point type, stress type, and cold shut type, thus realizing the physical mapping from geometric features to risk distribution.

[0036] Specifically, the partitioning module includes: The distortion factor calculation submodule is used to calculate the distortion factor of each grid cell based on the actual curvature of each grid cell and the curvature of the equal-volume sphere. The risk calculation submodule is used to perform min-max normalization on the hot spot volume, curvature change rate, gate distance and distortion factor of each grid cell to obtain normalized hot spot volume, normalized curvature change rate, normalized gate distance and normalized distortion factor, and then combine them with the corresponding preset weight coefficients for weighted fusion to obtain the risk coefficient of each grid cell. The type determination submodule is used to determine the encryption type of each grid cell based on the comparison results of the normalized hot spot volume, normalized rate of curvature change, normalized distortion factor, and normalized gate distance of each grid cell with the corresponding preset threshold; wherein the encryption type includes at least hot spot type, stress type, and cold shut type. The region determination submodule is used to determine the continuous grid cell region with a risk coefficient greater than the preset encryption threshold as the encryption zone.

[0037] In this embodiment, the distortion factor calculation submodule is used to calculate the distortion factor of each grid cell based on the actual curvature of each grid cell and the curvature of the equal-volume sphere. First, the curvature difference is calculated, which is obtained by subtracting the curvature of the equal-volume sphere from the actual curvature of the same grid cell. Then, the curvature difference is divided by the curvature of the equal-volume sphere to obtain the dimensionless distortion factor. In this embodiment, the sign and absolute value of the distortion factor calculation result are fully preserved for use by the subsequent partitioning module.

[0038] In this embodiment, normalization uses the min-max method: Normalized value = (original value - minimum value) / (maximum value - minimum value). Wherein, a zero hot spot volume indicates no hot spot risk; a larger rate of curvature change indicates more drastic curvature change; a larger gate distance indicates greater distance from the gate and higher risk at the flow end; the distortion factor needs to be taken as its absolute value before normalization, with a larger value indicating more severe geometric distortion.

[0039] In this embodiment, the preset weighting coefficients are the proportions of four normalized parameters, with the sum of the weights being 1. The weight values ​​are determined based on the physical mechanism of ferrous metal casting: the hot spot volume has the highest weight, initially 0.40; the curvature change rate is second, initially 0.30; the distortion factor is third, initially 0.20; and the gate distance is the lowest, initially 0.10. This weight can be dynamically adjusted according to the dominant defect type of the irregular part: when shrinkage cavity risk is dominant, the hot spot volume weight is increased to 0.50; when cold shut risk is dominant, the gate distance is increased to 0.30; and when hot crack risk is dominant, the curvature change rate is increased to 0.40, so that the risk coefficient adaptively adapts to different irregular parts.

[0040] In this embodiment, for each grid cell, the four normalization parameters are multiplied by their corresponding weight coefficients, and then the four products are summed to obtain the risk coefficient of that grid cell. The risk coefficient ranges from 0 to 1, and serves as the input for the subsequent region determination submodule, used to filter high-risk grid cells and delineate the encrypted area.

[0041] In this embodiment, the preset thresholds for the four normalization parameters are uniformly set to 0.6. This value is based on statistical analysis of 50 typical defect samples: the normalization parameter values ​​for known defect locations are concentrated in the range of 0.52 to 0.85, and the mean values ​​of each parameter are between 0.60 and 0.68. After weighing the options, a threshold of 0.6 results in a defect recognition rate of 87%, a false positive rate of 18%, and an encrypted area ratio of 19% to 23%; a threshold of 0.5 increases the recognition rate to 94% but raises the false positive rate to 35%, significantly increasing costs; a threshold of 0.7 reduces the false positive rate to 9% but decreases the recognition rate to 71%, resulting in an excessively high risk of missed detections. Therefore, 0.6 was selected as the uniform preset threshold, which is applicable to different ferrous metal irregular-shaped parts.

[0042] In this embodiment, the preset encryption threshold is a criterion value used to screen high-risk grid cells, and its value ranges from 0 to 1. In this embodiment, the preset encryption threshold is set to 0.7. Based on the statistical verification of the aforementioned 50 typical samples, the risk coefficient of the grid cell where the known defect location is located is concentrated between 0.65 and 0.92, with a mean of 0.78; while the risk coefficient of the non-defect area is concentrated between 0.25 and 0.55, with a mean of 0.42. When the threshold is set to 0.7, the recognition rate of known defect samples is 82%, and the total area of ​​the encryption zone accounts for 12% to 18% of the total area of ​​the casting, avoiding excessive encryption while ensuring the defect detection rate. This threshold can be adjusted according to the quality requirements and cost constraints of specific irregular parts: for applications with stringent quality requirements such as aerospace, the threshold can be reduced to 0.6 to improve the defect detection rate; for cost-sensitive mass production, the threshold can be increased to 0.8 to reduce the number of sensors deployed.

[0043] In this embodiment, the region determination submodule traverses all grid cells and marks grid cells with a risk coefficient greater than a preset encryption threshold as high-risk cells. Then, a connected component labeling algorithm is used to aggregate high-risk cells. Specifically, a two-pass scanning method is used to merge adjacent high-risk cells (i.e., cells sharing faces or edges) into the same connected region. The first scan assigns a temporary label to each high-risk cell and records the equivalence relationship. The second scan merges the equivalence labels, so that each connected region obtains a unique region identifier. Each connected region is determined as an encryption zone. Isolated connected regions with fewer grid cells than a preset isolation threshold are discarded and not processed as encryption zones because they may come from numerical noise or local geometric singularities. Finally, several discrete encryption zones are output. Each encryption zone has a clear boundary range, the grid cells it belongs to, and the risk coefficient distribution, which serves as the spatial basis for sensor deployment in the subsequent acquisition module.

[0044] In this embodiment, the preset isolation threshold refers to the lower limit of the number of grid cells used to determine whether a connected region should be removed due to its excessively small scale. In this embodiment, the preset isolation threshold is set to 3 grid cells. The basis for this value is that, after min-max normalization and connected component aggregation, high-risk connected regions consisting of one or two grid cells typically originate from boundary singularities in grid division, local numerical noise, or computational errors in individual grid cells, rather than actual geometric hotspots or structural abrupt changes. Statistical analysis of the risk distribution of 50 typical irregularly shaped components shows that the densified area corresponding to a real defect covers at least 3 consecutive grid cells. Therefore, setting the isolation threshold to 3 can retain the real defect area while filtering out noise interference, avoiding overly fragmented densified areas caused by numerical noise, preventing the placement of sensors at isolated high-risk cells without actual physical meaning, thereby saving sensor hardware costs, reducing the complexity of subsequent data acquisition and processing, and making the division of the densified area more engineering-practical and physically interpretable.

[0045] Understandably, when the risk coefficient is less than or equal to the preset encryption threshold, the corresponding grid cell is determined to be a low-risk area, is not included in any encryption area, and no monitoring sensors are deployed.

[0046] Through a three-tiered progressive architecture of risk calculation, type determination, and region identification, a complete mapping from physical attributes to the densified zone space is achieved. The risk calculation submodule uses min-max normalization to eliminate dimensional differences and differentiates the weighting coefficients to adaptively reflect the coupling effects of four types of inducing factors: thermal points, stress, distortion, and flow ends. The type determination submodule accurately classifies grid cells into thermal point type, stress type, or cold shut type by comparing each parameter with an independent threshold and arbitrating physical priorities. This allows the densification type to directly guide the subsequent differentiated sensor deployment, avoiding information redundancy from indiscriminate deployment. The region identification submodule filters high-risk cells using thresholds and aggregates continuous regions using connected component labeling, while eliminating noise islands below the isolation threshold. Through the collaborative work of these three modules, the risk zoning results possess physical interpretability, spatial continuity, and engineering economy, providing the acquisition module with accurate spatial basis and type labels.

[0047] Specifically, the distortion factor calculation submodule includes: A curvature difference calculation unit is used to calculate the difference between the actual curvature of the mesh cell and the curvature of the equal-volume sphere to obtain the curvature difference value. A distortion factor calculation unit is used to divide the curvature difference by the curvature of the equal-volume sphere to obtain the distortion factor.

[0048] By transforming the abstract geometric distortion features of irregularly shaped parts into quantifiable and comparable physical risk indicators through distortion factors, core criteria are provided for subsequent partitioning modules to identify stress-type encrypted areas and configure acoustic emission sensors differently, realizing a physical mapping from geometric shape to casting defect risk.

[0049] Please see Figure 2 As shown, this is a flowchart of the type determination submodule for determining the encryption type in this embodiment.

[0050] Specifically, the excess factor determination unit is used to determine the excess factor based on the normalized hot spot volume, the normalized rate of change of curvature, the normalized distortion factor, the normalized gate distance, and the corresponding preset threshold. A single-condition determination unit is used to determine the encryption type of the corresponding grid cell based on the number of the excess factors when the number of excess factors is 1. A multi-condition arbitration unit is used to determine the encryption type according to a preset priority order when the number of the excess factors is greater than 1.

[0051] In this embodiment, the out-of-standard factor determination unit compares the normalized hot spot volume, normalized rate of curvature change, normalized distortion factor, and normalized gate distance of each grid cell with their respective preset thresholds, and marks the parameters that are greater than the corresponding preset thresholds as out-of-standard factors.

[0052] In this embodiment, the single condition determination unit is used to directly determine the mesh cell's densification type based on the type of the exceeding factor when the number of exceeding factors is exactly 1. The specific correspondence is as follows: exceeding the hot spot volume corresponds to the hot spot type, exceeding the curvature change rate or distortion factor corresponds to the stress type, and exceeding the gate distance corresponds to the cold shut type.

[0053] In this embodiment, the multi-condition arbitration unit is used to determine the encryption type according to a preset priority order when the number of exceeding factors is greater than 1. The preset priority order is thermal break type over stress type, and stress type over cold shut type, that is, the type with the highest priority is selected as the encryption type. The basis for this priority is that once shrinkage cavity defects are formed, they are difficult to remedy, while thermal cracks and cold shuts are relatively controllable. Therefore, thermal break type has the highest priority.

[0054] Understandably, when all normalized parameters are less than or equal to the corresponding preset threshold, no out-of-limit factors are generated, and the grid cell is determined to have no specific defect cause and is not included in the encrypted area type classification.

[0055] Through a three-tiered progressive logic of exceeding-standard factor determination, single-condition judgment, and multi-condition arbitration, precise classification of defect causes is achieved. The exceeding-standard factor determination unit independently triggers judgment signals for three types of physical causes: thermal points, stress, and cold shuts, preventing the risk of other parameters from being masked by excessively high single parameters. The multi-condition arbitration unit arbitrates according to priority when multiple parameters exceed the standard simultaneously, conforming to the defect evolution law. Through the above three-tiered judgment, each grid cell is assigned a unique encrypted type label, enabling the acquisition module to configure corresponding sensors for different causes—thermocouples for thermal points, acoustic emission sensors for stress, and temperature-pressure composite sensors for cold shuts. This achieves precise mapping from defect causes to sensing strategies, avoiding data redundancy and cost waste from indiscriminate deployment.

[0056] Please see Figure 3 As shown, it is the deployment strategy and data acquisition flowchart of the acquisition module in this embodiment.

[0057] Specifically, the acquisition module includes: The deployment strategy submodule is used to differentiate the sensor type and sampling frequency according to the encryption type; The partition acquisition submodule is used to acquire the casting monitoring data in real time at the sampling frequency corresponding to each sensor, and associate and store the casting monitoring data with the corresponding encrypted zone; the casting monitoring data includes at least the first temperature data of the hot-spot type encrypted zone, the elastic wave signal of the stress type encrypted zone, and the second temperature data and pressure data of the cold-spot type encrypted zone.

[0058] In this embodiment, the deployment strategy submodule is used to differentiate the sensor type and sampling frequency in each encryption zone according to the encryption type determined by the partitioning module. The specific configuration strategy is as follows: For hot-spot type confined zones, thermocouple arrays are deployed within the zone to collect temperature field changes during solidification. The sampling frequency is set to 1 Hz to 10 Hz to monitor temperature plateau characteristics before shrinkage cavity formation. The physical basis for selecting thermocouples in hot-spot type confined zones is that the core risk is shrinkage cavity defects, whose precursory characteristics are abnormal solidification rate and prolonged temperature plateau. Thermocouple arrays can continuously monitor the temperature evolution process, capturing two key indicators: abrupt changes in solidification rate and the duration of the temperature plateau.

[0059] For stress-type insulated zones, acoustic emission sensors are deployed within the zone to collect elastic wave signals generated by the release of thermal stress during solidification. The sampling frequency is set to 100 kHz to 1 MHz to capture high-frequency signals during the initiation of hot cracks. The physical basis for selecting acoustic emission sensors in stress-type insulated zones is that the core risk of these zones is hot cracking defects, the precursor of which is the release of elastic waves caused by the obstruction of solidified shell contraction. This signal has a high frequency and strong transient response, making acoustic emission sensors a standard means of capturing such high-frequency elastic waves.

[0060] For cold-shut type encryption zones, high-response thermocouples and pressure sensors are deployed within the zone to collect high dynamic response data during the filling process. The sampling frequency is set to 100 Hz to 1000 Hz to monitor the temperature drop and pressure fluctuation at the molten metal front. The physical basis for selecting a temperature-pressure composite sensor in cold-shut type encryption zones is that the core risk of cold-shut type encryption zones is premature solidification of the molten metal front, the precursor characteristics of which are a sudden temperature drop and abnormal fluctuations in filling pressure. High-response thermocouples can capture millisecond-level temperature changes, and pressure sensors can simultaneously monitor changes in filling resistance. The combined use of the two can accurately identify the critical state before cold shut occurs.

[0061] The principle of differentiated configuration of the above sampling frequency is as follows: cold-stop type and stress type densification zone need to capture transient changes. The high-frequency elastic wave of the stress type and the millisecond-level temperature drop of the cold-stop type are therefore sampled at a significantly higher frequency than the thermal-stop type densification zone. The temperature field change of the thermal-stop type densification zone is relatively gentle.

[0062] It should be noted that the sensors are not deployed one-to-one with individual grid cells, but rather within a specific spatial range of the densified area. Several sensors are configured within each densified area according to a preset density, with the sensor installation location preferentially selected from the spatial coordinates corresponding to the grid cell with the highest risk factor within the densified area. Specific deployment rules are as follows: For thermal break type densified areas, the spacing between thermocouple array points is 10 mm to 50 mm, with at least 3 measurement points deployed in each densified area; for stress type densified areas, the spacing between acoustic emission sensors is 20 mm to 100 mm, with at least 1 sensor deployed in each densified area; for cold shut type densified areas, the spacing between thermo-pressure composite sensors is 10 mm to 30 mm, with at least 2 measurement points deployed in each densified area.

[0063] The zone acquisition submodule is used to acquire monitoring data in real time at the sampling frequency corresponding to each sensor, and associate the acquired monitoring data with the corresponding encrypted zone for storage, establishing a mapping relationship between encrypted zone identifiers, sensor identifiers, and monitoring data time sequences. The monitoring data includes at least: first temperature data for the hot-spot type encrypted zone, reflecting the solidification rate and the duration of the temperature plateau; elastic wave signals for the stress type encrypted zone, reflecting the stress release frequency and amplitude characteristics; and second temperature and pressure data for the cold-stop type encrypted zone, reflecting temperature changes at the molten metal front and fluctuations in filling pressure.

[0064] It should be noted that data is stored at the encrypted zone level, with each encrypted zone serving as an independent data unit. The defect assessment module directly reads monitoring data unit by unit from the encrypted zones and calculates the defect index for each zone. This storage mechanism significantly reduces data storage volume and computational complexity while ensuring the accuracy of risk assessment, making it more suitable for real-time control needs in industrial settings.

[0065] Through a collaborative architecture of global acquisition and local differentiated monitoring, accurate data acquisition of the casting process of ferrous metal irregular parts is achieved. This ensures that the sensor type matches the physical causes, the sampling frequency matches the signal characteristics, and the data granularity matches the judgment requirements. This provides high-quality and highly targeted data input for the judgment module, while avoiding hardware redundancy and data overload caused by indiscriminate sensor deployment. An optimized balance is achieved between monitoring accuracy and engineering economy.

[0066] Please see Figure 4 As shown, it is a flowchart of the deviation calculation and double correction of the determination module in this embodiment.

[0067] Specifically, the determination module includes: The deviation calculation submodule is used to calculate the degree of deviation, total deviation and deviation rate of the pouring monitoring data of each sensor in each encrypted zone relative to the preset expected value. The index calculation submodule is used to take the maximum value of the sensor defect indices of all sensors in each encryption zone as the encryption zone defect index of each encryption zone; wherein, the sensor defect index is the maximum value of the deviation degree, the total deviation and the deviation rate; The correction submodule is used to correct the defect index of the encryption area based on the encryption type and the solidification temperature difference of the ferrous metal irregular part, so as to obtain the correction index. The risk assessment submodule is used to compare the correction index with the corresponding preset assessment threshold. When the correction index is greater than the preset assessment threshold, it is determined that there is a defect risk in the corresponding encryption area.

[0068] In this embodiment, the preset expected value refers to the standard reference value of the sensor monitoring data in each densified zone under normal casting process conditions for ferrous metal, serving as the benchmark for calculating the degree of deviation, total deviation, and deviation rate. In this embodiment, the preset expected value is mainly determined by solidification simulation to conform to the underlying logic of the prior zoning of this application—that is, the preset expected value can be obtained before casting, so that the deviation calculation is logically consistent with the preliminary steps such as risk zoning and sensor deployment. Specifically, for temperature monitoring data, including first temperature data and second temperature data, the finite element method is used to simulate the filling and solidification process of the same grade of ferrous metal under the same casting process parameters, and the theoretical cooling curve of each sensor installation position is extracted as the initial preset expected value; on this basis, the simulation curve can be progressively corrected by 3 to 5 trial castings and the measured data of defect-free samples, so that the preset expected value gradually approaches the actual process state. For the elastic wave signal, the preset expected value is set to the background noise level of the acoustic emission sensor during the defect-free casting process. The root mean square value of the background noise refers to the value obtained by continuously collecting 1000 signal points at a preset sampling frequency before the start of casting or during the defect-free casting process, square the voltage value of each signal point, sum them, divide by 1000, and then take the square root. In this embodiment, the root mean square value is taken as 20 mV to 50 mV. For the pressure data, the preset expected value is set to the pressure curve during the normal filling process. This curve is obtained through fluid dynamics simulation or actual measurement calibration under standard process conditions. Typical values ​​are 0.1 MPa to 0.3 MPa in the initial filling stage, 0.2 MPa to 0.5 MPa in the middle filling stage, and 0.3 MPa to 0.6 MPa in the final filling stage. Optionally, for castings that have entered the mass production stage, the preset expected value can be obtained directly by statistically averaging the measured data of at least 10 sets of defect-free castings. The preset expected value is determined mainly by simulation, which ensures that the deviation from the calculation benchmark can be obtained before casting. This avoids the model error of relying solely on simulation and reduces the trial casting cost of relying solely on actual measurement.

[0069] In this embodiment, the correction index is equal to the encryption zone defect index multiplied by the first correction coefficient and then by the second correction coefficient.

[0070] Specifically, the first correction factor depends on the sensitivity of the encryption type to deviations in the monitoring data: the hot spot type has the highest sensitivity, set at 1.2; the stress type is the baseline, set at 1.0; and the cold shut type is in between, set at 1.1. These factors are determined based on sensitivity analysis of typical defect samples and can be adaptively adjusted according to specific casting materials and process conditions.

[0071] Specifically, the second correction coefficient depends on the solidification temperature difference of the ferrous metal irregularly shaped part. In this embodiment, the solidification temperature difference is defined as the relative deviation between the actual cooling rate and the process-set cooling rate: Solidification temperature difference = (Actual cooling rate - Process-set cooling rate) / Process-set cooling rate. The actual cooling rate is calculated by linear regression of the measured cooling curve; the process-set cooling rate is obtained through simulation using finite element solidification simulation software. The formula for calculating the second correction coefficient is: Second correction coefficient = 1 + Solidification temperature difference × Preset gain factor. Optionally, for thermally concentrated areas, a temperature gradient can be introduced as an auxiliary correction term.

[0072] In this embodiment, the preset gain factor is a scaling factor used to adjust the degree of influence of cooling rate deviation on the second correction coefficient. Sensitivity analysis of 50 typical defect samples verifies that: when the preset gain factor is below 0.3, the correction is insufficient, and the false negative rate in high-risk areas increases; when the preset gain factor is above 0.7, the correction is excessive, and the false positive rate increases significantly; the defect identification rate reaches its optimal balance when the preset gain factor is 0.5. In this embodiment, the preset gain factor is set to 0.5, quantifying the impact of cooling rate deviation on risk as an adjustable scaling parameter, making the correction mechanism both physically interpretable and engineering flexible.

[0073] In this embodiment, the preset judgment threshold is a critical value used to determine whether there is a defect risk in the encrypted area. The value ranges from 0.5 to 0.8. Based on statistical verification of 50 typical defect samples, the correction index distribution in the known defective encrypted area is concentrated between 0.55 and 0.85, with a mean of 0.68; while in the defect-free encrypted area, the distribution is concentrated between 0.20 and 0.50, with a mean of 0.35. In this embodiment, it is uniformly set to 0.6. The recognition rate for known defective samples is 85%, and the false alarm rate for defects in the encrypted area is 12%. The balance between the recognition rate and the false alarm rate meets the requirements of engineering applications.

[0074] Understandably, when the correction index is less than or equal to the preset judgment threshold, the corresponding encrypted area is judged to have no defect risk, and the subsequent defect identification and control process is not triggered.

[0075] A four-level architecture—deviation calculation, index aggregation, dual correction, and risk assessment—achieves a quantitative mapping from monitoring data to defect risk status. Specifically, the deviation calculation submodule extracts abnormal features from monitoring data across three dimensions: amplitude, cumulative value, and trend, simultaneously characterizing three risk modes: sudden anomalies, gradual defects, and trend changes. The index calculation submodule takes the maximum value of each of the three dimensions at the sensor level to avoid masking local risks through averaging; at the encryption zone level, it takes the maximum value of the defect indices of all sensors within the zone to ensure the sensitivity of risk assessment. The correction submodule introduces a dual correction mechanism based on encryption type and solidification temperature difference, mapping the original defect index to a risk metric that more closely reflects physical reality. The risk assessment submodule uses a preset threshold as the decision boundary to balance the false alarm rate and the missed detection rate. These four modules work collaboratively to provide reliable risk trigger signals to the identification module.

[0076] Specifically, the deviation calculation submodule includes: A deviation calculation unit is used to calculate the ratio of the absolute value of the difference between the pouring monitoring data and the preset expected value to the preset expected value, so as to obtain the deviation degree; The deviation calculation unit is used to calculate the cumulative value of the deviation degree from the start time of pouring to the current time, so as to obtain the total deviation. The deviation rate calculation unit is used to calculate the quotient of the difference in deviation between two adjacent samples divided by the sampling time interval to obtain the deviation rate.

[0077] In this embodiment, the total deviation is calculated by summing the deviation at each sampling time from the start of the pouring to the current time to obtain the total deviation.

[0078] In this embodiment, the deviation rate is calculated as follows: the deviation at the current sampling time is subtracted from the deviation at the previous sampling time to obtain the deviation difference, and then divided by the sampling time interval to obtain the deviation rate.

[0079] This system quantifies abnormal states in monitoring data from multiple perspectives using three dimensions: deviation degree, total deviation, and deviation rate. Deviation degree, expressed as a relative deviation, eliminates dimensional differences between different sensors, making temperature, pressure, and elastic wave signals comparable on the same scale. Dimensionlessness also ensures thresholds are universally applicable across sensors. Total deviation accumulates the deviation degree at each sampling time, reflecting the cumulative effect of the gradual defect formation process and compensating for the short-sightedness of instantaneous judgment. Deviation rate captures the trend of deviation degree changes through the difference between adjacent samples; positive values ​​indicate increased risk, while negative values ​​indicate mitigation, providing earlier warnings of sudden defects than amplitude indicators. These three dimensions correspond to the current amplitude, historical accumulation, and development trend of the abnormal state, respectively. This complementary information allows subsequent index calculations to extract the most severe risk signal by taking the maximum value, avoiding missed or false judgments caused by single-dimensional judgments.

[0080] Specifically, the identification module includes: The hot spot identification submodule is used to extract the temporal features of the first temperature data when the encryption type is hot spot type. If it meets the preset hole shrinkage pattern features, the defect type of the corresponding encryption area is identified as a hole shrinkage defect. The stress identification submodule is used to extract the frequency characteristics of the elastic wave signal when the encryption type is stress type. If it meets the preset thermal cracking mode characteristics, the defect type of the corresponding encryption area is identified as a thermal cracking defect. The cold shut identification submodule is used to extract the temporal features of the second temperature data when the encryption type is cold shut type, and if it meets the preset cold shut mode features, then the defect type of the corresponding encryption area is identified as a cold shut type defect; and to extract the waveform features of the pressure data, and if it meets the preset insufficient filling mode features, then the defect type of the corresponding encryption area is identified as an insufficient filling type defect.

[0081] In this embodiment, the specific quantification thresholds for the preset shrinkage mode feature are: the cooling curve shows a temperature plateau or rebound near the solidus line, the plateau duration is more than twice that of the normal cooling curve, the cooling rate is less than 50% of the process-set cooling rate, and the maximum temperature difference in the region exceeds 30 degrees Celsius. The specific quantification thresholds for the preset thermal cracking mode feature are: a high-frequency burst signal of 50 kHz to 200 kHz appears in the elastic wave signal, the signal amplitude exceeds three times the root mean square value of the background noise, and the event count rate increases from less than 10 times to more than 100 times within 1 second and then decays. The specific quantification thresholds for the preset cold shut mode feature are: the temperature curve shows a sudden drop exceeding three times the normal cooling rate, and the temperature drops by more than 100 degrees Celsius within 2 seconds. The specific quantification thresholds for the preset insufficient filling mode feature are: the pressure peak value is less than 60% of the normal filling pressure peak value, the pressure rise rate is less than 50% of the normal pressure rise rate value, and the pressure fluctuation amplitude exceeds twice the normal pressure fluctuation amplitude value.

[0082] In this embodiment, the event count rate refers to the number of acoustic emission events occurring per unit time. An acoustic emission event is defined as a single transient waveform of the elastic wave signal exceeding a preset trigger threshold, which is set to twice the root mean square value of the background noise. The event count rate is calculated as follows: using a 1-second time window, the number of acoustic emission events occurring within that window is counted; updated every 0.1 seconds, using a sliding window calculation. The methods for obtaining the process-set cooling rate and the root mean square value of the background noise are described above and will not be repeated here.

[0083] In this embodiment, the normal cooling curve, normal cooling rate, normal peak filling pressure, normal pressure rise rate, and normal pressure fluctuation amplitude all refer to the standard reference values ​​collected in each confined zone of castings that have been inspected and confirmed to be defect-free under the same casting process parameters. The above normal values ​​are obtained as follows: using the same process conditions as the casting to be monitored, including the same grade of ferrous metal, the same pouring temperature, pouring speed, and initial mold temperature, 3 to 5 castings that have been confirmed to be defect-free by X-ray inspection or ultrasonic testing are continuously poured as standard samples; monitoring data of the entire pouring and solidification process are collected at the same sampling frequency at the sensor deployment locations corresponding to each confined zone; the monitoring data of 3 to 5 sets of defect-free samples are statistically averaged to obtain the standard cooling curve, standard cooling rate, standard pressure curve, standard peak pressure, standard pressure rise rate, and standard pressure fluctuation amplitude at each confined zone; the above standard values ​​are stored in the system as a comparison benchmark for pattern recognition. For new products where defect-free samples cannot be obtained, a theoretical standard curve can be obtained through solidification simulation. Then, the curve can be corrected using the measured data of the first batch of castings that have been confirmed to be defect-free during the trial casting. Subsequent batches can be gradually optimized.

[0084] In this embodiment, pattern matching employs a multi-dimensional threshold comparison method. For temperature time-series characteristics, the plateau length, cooling rate, and temperature gradient of the cooling curve are calculated in real time and compared with corresponding preset thresholds. When all three indicators exceed the threshold, it is determined to match a concave-hole pattern. For elastic wave frequency characteristics, the signal is subjected to a Fast Fourier Transform to obtain the spectrum, and the energy proportion in the 50 kHz to 200 kHz frequency band is calculated. When this proportion exceeds 60% and the signal amplitude exceeds three times the root mean square value of the background noise, it is determined to match a thermal cracking pattern. For pressure waveform characteristics, the pressure peak value, rise rate, and fluctuation amplitude are calculated in real time and compared with corresponding preset thresholds. When any one of them is below the threshold, it is determined to match an underfilling pattern. The above determination rules are stored in the system in the form of embedded code and are executed once per sampling cycle.

[0085] Understandably, when the monitoring data characteristics of the encrypted area do not match the preset pattern characteristics of the corresponding defect type, it is determined that the encrypted area does not have this type of defect, and monitoring continues or the process of identifying other defect types is initiated.

[0086] By matching the encryption type with the characteristics of the monitoring data, the accurate identification of defect types is achieved. Each identification submodule only processes the feature dimensions related to its physical cause, avoiding the problems of feature redundancy and large computational overhead in general pattern recognition, and realizing fast and accurate defect type determination.

[0087] Specifically, the control module includes: The grading submodule is used to classify the severity of defects as mild, moderate, or severe based on the correction index. The shrinkage cavity control submodule is used to adjust the pouring speed and / or holding time in the pouring parameters when the defect type is a shrinkage cavity and the severity is mild, performing one of reducing the pouring speed and extending the holding time; when the severity is moderate or above, performing both simultaneously. The hot crack control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is hot crack and the severity is mild, performing either reducing the pouring temperature or reducing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The cold shut control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is a cold shut defect and the severity is mild, by increasing the pouring temperature and increasing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The underfill control submodule is used to adjust the pouring speed and / or pouring pressure in the pouring parameters when the defect type is underfill defect and the severity is mild, by increasing the pouring speed and increasing the pouring pressure; when the severity is moderate or above, both are executed simultaneously.

[0088] In this embodiment, the preset first threshold, preset second threshold, and preset third threshold are grading boundary values ​​used to classify the severity of defects. Based on statistical analysis of 50 typical defect samples, when the correction index is in the range of 0.6 to 0.8, the defect size is less than 5 mm and the performance reduction is less than 10%; when it is in the range of 0.8 to 1.0, the defect size is 5 to 15 mm and the performance reduction is 10% to 25%; and when it is greater than 1.0, the defect size is greater than 15 mm and the performance reduction is greater than 25%. The above thresholds are aligned with the risk judgment threshold of 0.6, making risk judgment and graded control directly correspond. In this embodiment, the preset first threshold is 0.6, the preset second threshold is 0.8, and the preset third threshold is 1.0, so that the control intensity matches the severity of the defect. Mild defects are eliminated with minimal intervention, while severe defects are directly controlled with strong control, achieving a balance between defect improvement effect and process stability.

[0089] Understandably, when the correction index is less than or equal to the preset first threshold, it is determined to be without defect risk and no regulation is triggered.

[0090] In this embodiment, based on sensitivity analysis of 50 typical defect samples, various process parameters were adjusted at different time lengths and the defect improvement effects were recorded. After achieving a balance between the defect improvement rate and process side effects, the values ​​of each control measure were determined. Specifically, adjusting the pouring speed by 5%, 8%, and 10% can improve the shrinkage cavity and hot crack defects by 25%, 40%, and 55%, respectively. Further increases significantly increase the risk of cold shut. Extending the holding time by 10, 20, and 30 minutes can improve the shrinkage cavity by 20%, 38%, and 52%, respectively. Exceeding 30 minutes significantly reduces production efficiency. Adjusting the pouring temperature by 10, 15, and 20 degrees Celsius can improve the hot crack by 22%, 40%, and 48%, respectively. Further decreases increase the risk of cold shut. Adjusting the pouring pressure by 10% and 15% can improve the incomplete filling rate by 58% to 65% under speed control. Further increases increase the risk of splashing.

[0091] In this embodiment, the shrinkage cavity control submodule is used to implement corresponding control measures according to the severity of the defect when the defect type is shrinkage cavity. For mild cases, a single control measure is selected based on the location of the shrinkage cavity: if the shrinkage cavity is located near the riser or in a thick part, the holding time is extended by 10 minutes; if the shrinkage cavity is located in the transition zone between thin and thick walls, the pouring speed is reduced by 5%. For moderate cases, the pouring speed is reduced by 8% and the holding time is extended by 20 minutes. For severe cases, the pouring speed is reduced by 10% and the holding time is extended by 30 minutes.

[0092] In this embodiment, the hot crack control submodule is used to implement corresponding control measures based on the severity of the hot crack defect. For mild defects, the pouring temperature is reduced by 10 degrees Celsius first. If the defect is not eliminated after three consecutive pours, the pouring speed is reduced by 5%. For moderate defects, the pouring temperature is reduced by 15 degrees Celsius and the pouring speed is reduced by 8%. For severe defects, the pouring temperature is reduced by 20 degrees Celsius and the pouring speed is reduced by 10%.

[0093] In this embodiment, the cold shut control submodule is used to implement corresponding control measures based on the severity of the cold shut defect. For mild defects, the pouring temperature is increased by 10 degrees Celsius first; if the defect is not eliminated after three consecutive pours, the pouring speed is increased by 5%. For moderate defects, the pouring temperature is increased by 15 degrees Celsius and the pouring speed is increased by 10%. For severe defects, the pouring temperature is increased by 20 degrees Celsius and the pouring speed is increased by 15%.

[0094] In this embodiment, the incomplete filling control submodule is used to implement corresponding control measures according to the severity of the defect when the defect type is incomplete filling. For mild cases, the pouring speed is increased by 10%; for moderate cases, the pouring speed is increased by 15% and the pouring pressure is increased by 10%; for severe cases, the pouring speed is increased by 20% and the pouring pressure is increased by 15%.

[0095] In this embodiment, the preset defect threshold is set to 5%, which is determined based on the statistics of 50 typical defect samples. When the defect rate drops below 5%, the probability that the comprehensive mechanical properties of the casting meet the design requirements reaches more than 95%, and the marginal contribution of further adjustments to performance improvement slows down. It can be understood that for irregularly shaped components in high-end equipment fields such as aerospace and energy power, the preset defect threshold can be set to 3%.

[0096] After each adjustment, the system continuously monitors the defect identification results for at least the next three castings. If the defect rate of all three castings has dropped below the preset defect threshold, the current process parameters are locked. If some castings meet the standard while others do not, monitoring continues until the fifth casting. If there are still non-compliance issues in the five castings, the next round of adjustment is executed. If a new defect type appears, the system switches to the corresponding control submodule. Based on statistics from 50 sets of samples, the confidence level reaches over 90% when monitoring 3 to 5 castings, achieving a balance between accuracy and production efficiency. Those skilled in the art can adjust the number of castings monitored according to process stability; when fluctuations are large, the number can be increased to 8 to 10 castings. The aforementioned control commands are sent to the actuators of the casting equipment. Specifically: the casting speed is achieved by controlling the rotation speed of the ladle tilting servo motor; the casting temperature is achieved by adjusting the gate heating power or the tapping temperature of the smelting furnace; the holding time is achieved by modifying the holding timer parameters of the control system; and the casting pressure is achieved by adjusting the opening of the pressure control valve, or indirectly by increasing the ladle tilting speed or increasing the height of the pouring cup in gravity casting. The adjusted process parameters serve as a new baseline for the next batch of production. The system continuously monitors the defect identification results and performs hierarchical control, forming a closed-loop adaptive iteration between batches until the process parameters converge to the optimal window.

[0097] A graded control mechanism that adjusts casting parameters differently based on defect type and severity achieves adaptive suppression of defect risk. Control of shrinkage cavities focuses on promoting feeding; for mild cases, reducing casting speed decreases turbulent air entrapment or extending holding time enhances solidification feeding; for moderate and severe cases, both methods work synergistically to accelerate shrinkage cavity healing. Control of hot cracks focuses on reducing thermal stress; for mild cases, reducing casting temperature decreases solidification shrinkage or reducing casting speed slows stress accumulation; for moderate and severe cases, both methods work synergistically to inhibit crack initiation and propagation. Control of cold shuts focuses on improving molten metal fluidity; for mild cases, increasing casting temperature extends flow time or increases casting speed enhances filling kinetic energy; for moderate and severe cases, both methods work synergistically to eliminate premature solidification at the mold front. Control of incomplete filling defects focuses on enhancing filling capacity; for mild cases, increasing casting speed increases filling pressure or directly increases casting pressure; for moderate and severe cases, both methods work synergistically to overcome flow resistance and ensure the cavity is fully filled. The aforementioned hierarchical control strategy is based on the physical mechanism of defect formation. It achieves a dynamic balance between minimal intervention and strong control, avoiding both the decline in production efficiency or the risk of secondary defects caused by over-control and the precise suppression of defects of different degrees. Ultimately, it forms an adaptive closed loop of identification, judgment, control and iteration.

[0098] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. An intelligent control system for casting irregularly shaped metal parts, characterized in that, include: The acquisition module is used to acquire the curvature of the spherical shape with equal volume of ferrous metal, the hot spot volume of each grid cell in the grid cells divided on it, the rate of curvature change, the actual curvature and the gate distance. The partitioning module is used to determine several encryption zones based on the threshold comparison results of the risk coefficient of each grid cell, and to determine the encryption type of each encryption zone. The risk coefficient is determined based on the hot spot volume, the rate of curvature change, the distortion factor and the gate distance. The distortion factor is determined based on the actual curvature and the curvature of the equal-volume sphere. The acquisition module is used to acquire in real time the pouring monitoring data of each encryption zone during the pouring process of ferrous metal irregular parts based on preset pouring temperature and preset pouring speed, wherein the pouring monitoring data is determined based on each encryption type. The judgment module is used to determine whether there is a defect risk in each encryption zone based on the comparison result between the correction index of each encryption zone and the corresponding preset judgment threshold. The correction index is determined based on the defect index, the encryption type and the solidification temperature difference of the ferrous metal irregular part. The defect index is determined based on the degree of deviation, total deviation and deviation rate of the casting monitoring data of each encryption zone. An identification module is used to determine the defect type of each encryption zone based on the defect risk, the encryption type, and the casting monitoring data. The control module is used to adjust one or more of the following during the pouring process: pouring temperature, pouring speed, holding time, and pouring pressure, based on the defect type and the correction index.

2. The intelligent control system for casting irregularly shaped metal parts according to claim 1, characterized in that, The acquisition module includes: The meshing submodule is used to discretize the three-dimensional model of the ferrous metal irregular part to obtain several mesh elements. The hot spot analysis submodule is used to count the volume of the grid cell region where the solidification time exceeds the preset solidification threshold, so as to obtain the hot spot volume; wherein, the solidification time is determined based on the preset pouring temperature of the ferrous metal part, the initial temperature of the mold, and the thermophysical parameters of the ferrous metal material. The curvature analysis submodule is used to calculate the actual curvature and the rate of change of curvature of each grid cell; The equal-volume sphere curvature calculation submodule is used to calculate the curvature of a sphere with the same local volume as each grid cell, so as to obtain the curvature of the equal-volume sphere. The gate distance calculation submodule is used to obtain the gate distance based on the flow path length of each grid cell from the preset gate mark.

3. The intelligent control system for casting irregularly shaped metal parts according to claim 2, characterized in that, The partitioning module includes: The distortion factor calculation submodule is used to calculate the distortion factor of each grid cell based on the actual curvature of each grid cell and the curvature of the equal-volume sphere. The risk calculation submodule is used to perform min-max normalization on the hot spot volume, curvature change rate, gate distance and distortion factor of each grid cell to obtain normalized hot spot volume, normalized curvature change rate, normalized gate distance and normalized distortion factor, and then combine them with the corresponding preset weight coefficients for weighted fusion to obtain the risk coefficient of each grid cell. The type determination submodule is used to determine the encryption type of each grid cell based on the comparison results of the normalized hot spot volume, normalized rate of curvature change, normalized distortion factor, and normalized gate distance of each grid cell with the corresponding preset threshold; wherein the encryption type includes at least hot spot type, stress type, and cold shut type. The region determination submodule is used to determine the continuous grid cell region with a risk coefficient greater than the preset encryption threshold as the encryption zone.

4. The intelligent control system for casting irregularly shaped metal parts according to claim 3, characterized in that, The distortion factor calculation submodule includes: A curvature difference calculation unit is used to calculate the difference between the actual curvature of the mesh cell and the curvature of the equal-volume sphere to obtain the curvature difference value. A distortion factor calculation unit is used to divide the curvature difference by the curvature of the equal-volume sphere to obtain the distortion factor.

5. The intelligent control system for casting irregularly shaped metal parts according to claim 4, characterized in that, The type determination submodule includes: The out-of-standard factor determination unit is used to determine the out-of-standard factor based on the normalized hot spot volume, the normalized rate of change of curvature, the normalized distortion factor, the normalized gate distance, and the corresponding preset threshold. A single-condition determination unit is used to determine the encryption type of the corresponding grid cell based on the number of the excess factors when the number of excess factors is 1. A multi-condition arbitration unit is used to determine the encryption type according to a preset priority order when the number of the excess factors is greater than 1.

6. The intelligent control system for casting irregularly shaped metal parts according to claim 5, characterized in that, The acquisition module includes: The deployment strategy submodule is used to differentiate the sensor type and sampling frequency according to the encryption type; The partition acquisition submodule is used to acquire the casting monitoring data in real time at the sampling frequency corresponding to each sensor, and associate and store the casting monitoring data with the corresponding encrypted zone; the casting monitoring data includes at least the first temperature data of the hot-spot type encrypted zone, the elastic wave signal of the stress type encrypted zone, and the second temperature data and pressure data of the cold-spot type encrypted zone.

7. The intelligent control system for casting irregularly shaped metal parts according to claim 6, characterized in that, The determination module includes: The deviation calculation submodule is used to calculate the degree of deviation, total deviation and deviation rate of the pouring monitoring data of each sensor in each encrypted zone relative to the preset expected value. The index calculation submodule is used to take the maximum value of the sensor defect indices of all sensors in each encryption zone as the encryption zone defect index of each encryption zone; wherein, the sensor defect index is the maximum value of the deviation degree, the total deviation and the deviation rate; The correction submodule is used to correct the defect index of the encryption area based on the encryption type and the solidification temperature difference of the ferrous metal irregular part, so as to obtain the correction index. The risk assessment submodule is used to compare the correction index with the corresponding preset assessment threshold. When the correction index is greater than the preset assessment threshold, it is determined that there is a defect risk in the corresponding encryption area.

8. The intelligent control system for casting irregularly shaped metal parts according to claim 7, characterized in that, The deviation calculation submodule includes: A deviation calculation unit is used to calculate the ratio of the absolute value of the difference between the pouring monitoring data and the preset expected value to the preset expected value, so as to obtain the deviation degree; The deviation calculation unit is used to calculate the cumulative value of the deviation degree from the start time of pouring to the current time, so as to obtain the total deviation. The deviation rate calculation unit is used to calculate the quotient of the difference in deviation between two adjacent samples divided by the sampling time interval to obtain the deviation rate.

9. The intelligent control system for casting irregularly shaped metal parts according to claim 8, characterized in that, The identification module includes: The hot spot identification submodule is used to extract the temporal features of the first temperature data when the encryption type is hot spot type. If it meets the preset hole shrinkage pattern features, the defect type of the corresponding encryption area is identified as a hole shrinkage defect. The stress identification submodule is used to extract the frequency characteristics of the elastic wave signal when the encryption type is stress type. If it meets the preset thermal cracking mode characteristics, the defect type of the corresponding encryption area is identified as a thermal cracking defect. The cold shut identification submodule is used to extract the temporal features of the second temperature data when the encryption type is cold shut type, and if it meets the preset cold shut mode features, then the defect type of the corresponding encryption area is identified as a cold shut type defect; and to extract the waveform features of the pressure data, and if it meets the preset insufficient filling mode features, then the defect type of the corresponding encryption area is identified as an insufficient filling type defect.

10. The intelligent control system for casting irregularly shaped metal parts according to claim 9, characterized in that, The control module includes: The grading submodule is used to classify the severity of defects as mild, moderate, or severe based on the correction index. The shrinkage cavity control submodule is used to adjust the pouring speed and / or holding time in the pouring parameters when the defect type is a shrinkage cavity and the severity is mild, performing one of reducing the pouring speed and extending the holding time; when the severity is moderate or above, performing both simultaneously. The hot crack control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is hot crack and the severity is mild, performing either reducing the pouring temperature or reducing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The cold shut control submodule is used to adjust the pouring temperature and / or pouring speed in the pouring parameters when the defect type is a cold shut defect and the severity is mild, by increasing the pouring temperature and increasing the pouring speed; when the severity is moderate or above, both are performed simultaneously. The underfill control submodule is used to adjust the pouring speed and / or pouring pressure in the pouring parameters when the defect type is underfill defect and the severity is mild, by increasing the pouring speed and increasing the pouring pressure; when the severity is moderate or above, both are executed simultaneously.

Images