Top pouring rapid solidification forming method for thin-wall complex casting

Abstract

The invention discloses a top pouring and rapid solidification forming method for a thin-wall complex casting, and relates to the technical field of metal casting and rapid solidification, a melt flow field, a solid-liquid interface and solute distribution are collected at the moment of pouring and matched with a numerical model, and a segregation map is generated; activating the coil according to the map partition to generate Lorentz force to guide the circulating current; switching the heat conduction mode of the heat conduction sheet along the circulation direction and heating to synchronously push a solid-liquid interface; constructing an error field by using a spectrum-thermal image, and applying magnetic-thermal disturbance to the high-risk layer until convergence; and finally, melting and segmentally supplementing materials according to a static pressure curve by virtue of a memory alloy corrugated pipe riser, monitoring pores and displacement through sound transmission and speckles, and smoothly releasing pressure and compacting after reaching the standard. And real-time coupling control is achieved, segregation, cold shut and shrinkage cavities are eliminated on the premise that the period is not prolonged, and the overall structure uniformity and size stability of the casting are improved.

Description

Technical Field

[0001] The invention relates to the technical field of metal casting and rapid solidification, in particular to a top pouring and rapid solidification forming method for thin-walled complex castings. Background Art

[0002] Modern aircraft engine casings, additive-cast hybrid blades, and high-efficiency heat exchangers widely utilize top-cast, thin-walled, complex castings to reduce mass and increase internal flow channel integration. This inverted trapezoidal filling path requires the molten metal to be diverted, folded back, and connected within a very short distance. Any flow instability will result in incomplete fusion or cold shut defects. To suppress rapid freezing in thin-walled areas, the industry's common practice is to increase the pouring superheat, thicken the ceramic shell, or install insulating risers on the top. However, this phenomenon of "first solidification followed by blockage" at cross-sectional abrupt changes is still difficult to avoid. To address macro-shrinkage, foundries generally rely on risers, insulating sleeves, and heating sleeves to provide feed metal, supplemented by numerical methods such as the Niyama criterion to adjust the solidification direction. For microstructure control, low-frequency or rotary electromagnetic stirring has been used to refine grains and reduce central segregation, while online infrared temperature measurement and shortwave imaging are used to capture the cold-heat gradient to optimize the shell preheat curve. Defect detection technology has also shifted from post-process inspections to in-process monitoring. Ultrasonic arrays combined with digital radiography can locate hidden holes and cracks within seconds. Despite significant advances in single-physics fields, public literature indicates that the simultaneous use of rapid solidification and complex mold filling can introduce new coupled instabilities. Thermal, flow, composition, and stress fields often interact at the sub-second level, making real-time coordination difficult with current isolated control methods.

[0003] During the top pouring and rapid solidification process, the volume of the residual liquid zone shrinks on a millisecond-by-millisecond basis. If the feeding pressure and feeding rhythm do not match the real-time pore generation rate, hidden shrinkage cavities or capsule-like pores will form at the intersection of the rib root and the outer edge. Overly rapid pressurization can easily trigger cold cracking due to dendrite interlocking. Existing riser systems are unable to maintain a seal while simultaneously compensating for submillimeter displacement as the solid shell shrinks, making it difficult to achieve a balance between final stress and density. Relying solely on post-process ultrasonic or X-ray spot checks only reveals defects at the finished product stage, resulting in costly repairs and an inability to trace the instantaneous mismatch mechanism.

[0004] Therefore, there is an urgent need for a top pouring rapid solidification forming method for thin-walled complex castings to achieve a full chain closed loop from melt filling to solid-solid transformation, and completely solve the root cause problem of shrinkage and cracking of thin-walled complex castings. Summary of the Invention

[0005] (1) Technical problems solved To address the shortcomings of existing technologies, the present invention provides a top-pouring rapid solidification forming method for thin-walled complex castings. This method involves capturing the melt flow field, solid-liquid interface, and solute distribution at the moment of pouring and matching them with a numerical model to generate a segregation map. Based on the map's partitioning, coils are activated to generate Lorentz forces to guide the circulation. The heat transfer mode of the heat conducting plate is then switched according to the direction of the circulation, and heating is performed to synchronize the solid-liquid interface. Spectral-thermal imaging is then used to construct an error field, and magnetic-thermal perturbations are applied to high-risk layers until convergence. Finally, a memory alloy bellows riser is used to melt and feed material in sections according to the static pressure curve. Porosity and displacement are monitored using acoustic and speckle patterns. Once the desired conditions are met, pressure is released smoothly for densification. This method achieves real-time coupled control, eliminating segregation, cold shuts, and shrinkage without extending the cycle time, thus resolving the technical issues discussed in the background art.

[0006] (2) Technical solution To achieve the above objectives, the present invention is implemented through the following technical solutions: A top-pouring rapid solidification forming method for thin-walled complex castings involves collecting melt flow field, solid-liquid interface, and solute distribution data at the moment the pouring valve is opened, matching this data with a preset filling-solidification model, and generating a real-time, time-updated, three-dimensional segregation risk map for subsequent field control. Based on the three-dimensional segregation risk map, the outer edge and rib root coils are activated, and a programmable current is used to generate a soft Lorentz force against the solute gradient, stabilize the solute field distribution, and guide the melt microcirculation to diffuse from the edge to the core; Read the circulation vector, switch the heat conduction plate plane or volume heat conduction and allocate the heating block power in different zones, so that the heat flow direction of the rib root and transition zone is consistent with the circulation, and maintain the synchronous advancement and stability of the solid-liquid interface; Synchronously collect dual-band spectra and short-wave infrared thermal images, construct an error field and calibrate the high-risk layer, inject a micro-magnetic field and milliwatt pulse heat flow into it, iterate until the error converges and output a mark; The shape memory alloy bellows riser is stretched and sealed, the hydraulic servo valve applies pressure according to the static pressure curve and simultaneously melts the segmented feeding core rod, the pore volume fraction and speckle tracking displacement are monitored acoustically, and after meeting the standards, the pressure is smoothly released to complete the densification.

[0007] Furthermore, a unified time base is used to trigger the three-way synchronous sampling of the ultrasonic array, infrared camera and laser-induced spectrometer, and the three-way voxels are aligned to the mold coordinate system at one time through a three-point laser reference, generating a voxel stack with a weighted quality factor, providing high-reliability raw data for the three-dimensional segregation risk map.

[0008] Furthermore, the temperature-velocity dual calibration algorithm is used to iteratively correct the turbulent viscosity coefficient and latent heat coefficient. After the temperature difference error of the numerical model is lower than the set threshold, the segregation risk index is calculated, and a rolling updated three-dimensional segregation risk map is generated through Kalman incremental splicing.

[0009] Furthermore, according to the partitioning results of the three-dimensional segregation risk map, annular array coils, U-shaped embedded coils and axial through-hole coils are called, and the current density vector is optimized using a genetic algorithm. A time series package with amplitude, phase and harmonic identification is output to drive the corresponding coils.

[0010] Furthermore, after excitation, the measured circulating current and the model predicted difference energy are compared in real time, and the coil current is fine-tuned through the adaptive gain matrix and high-order harmonics or short-time inverted pulses are injected to break the root-locked vortex and converge the difference energy to below the preset threshold.

[0011] Furthermore, the heat flux density vector field is inverted by the dual grid finite volume method. After calculating the heat-flux misalignment index, the heat conducting plate is switched between plane heat conduction and volume heat conduction mode, and pulse power is output to the rib root heating block according to the weighted solute residual. Terahertz tomography and digital speckle patterning are used to jointly monitor the solid-liquid interface velocity. The excitation intensity of the thermal conductive plate and the heating power coefficient are adjusted according to the incremental proportional-integral method, so that the three-dimensional interface velocity decreases synchronously and the opening time of the feeding channel is dynamically extended.

[0012] Furthermore, the solute peak intensity is obtained by exciting a ring fiber array with dual-band laser pulses, and short-wave infrared thermal images are collected simultaneously. After calculating the spectral equilibrium error index and temperature equilibrium error index, the voxels are stratified into three layers: high-risk, medium-risk, and stable according to the energy threshold.

[0013] Furthermore, a sinusoidal perturbation and pseudo-random pulse accounting for one percent of the original current are applied to the high-risk layer coil, the corresponding thermal conductive sheet performs high-frequency small-amplitude warping and the heating block outputs negative step power. The convergence factor and cross-coupling degree are calculated every two cycles and adaptive amplitude modulation or delay is performed.

[0014] Furthermore, based on the octree-encrypted temperature-solid fractional field inversion, the pore generation rate is inverted, the soft static pressure curve is segmentedly fitted and the multi-melting point feeding mass flow rate is calculated. Then, after verification through the safety valve, the packaging pressure and feeding schedule are issued.

[0015] Furthermore, the riser of the shape memory alloy bellows is stretched to press the ceramic valve core and form a three-layer seal. The hydraulic servo valve applies pressure according to the pressure schedule and the fuse melts the core rod in stages. The pore volume fraction is monitored acoustically and the surface displacement is tracked with digital speckle patterns. After the densification threshold is met, the pressure is released according to the three-segment curve and the bellows is driven to retract to complete the densification.

[0016] (3) Beneficial effects The present invention provides a top pouring and rapid solidification forming method for thin-walled complex castings, which has the following beneficial effects: Using real-time benchmark mapping as the entry point, three types of dynamic information, namely melt flow field, solid-liquid interface and solute distribution, are uniformly projected onto a three-dimensional segregation risk map. With the map coordinates as the only target, the disconnection of multi-sensor data and the delay and inaccuracy of conversion of control instructions in traditional processes are avoided, significantly improving the information integrity and control consistency of the entire filling-solidification process.

[0017] Gradient magnetic field homogenization locks the soft Lorentz force field inside the high-risk cluster through partitioned coils and programmable current density, and then dynamically refines the shear layer with the help of harmonic injection and inverted pulses, so that the solute microcirculation has both directional propulsion capabilities and will not tear the primary solid shell, fundamentally resolving the problem of negative segregation stripes that are prone to occur in thin-walled parts.

[0018] The coordinated heat flux regulation utilizes the anisotropic switching of the thermal conductive sheet and the partitioned pulse of the external resistance heating block to construct a heat flux density network consistent with the direction of the microcirculation. It cooperates with the solid-liquid interface equipotential synchronization algorithm to continuously flatten the temperature gradient, and automatically extends the effective period of the liquid channel through the feeding window life prediction, allowing subsequent feeding to be completed at a lower static pressure.

[0019] The spectral-thermal image correction couples the dual-band laser spectrum and the short-wave infrared thermal image into a three-dimensional error field, and uses milliamplitude magnetic field disturbances and milliwatt-level heat flux pulses to implement pseudo-random precision needle modulation on the high-risk layer, eliminating residual segregation and temperature distortion without destroying the solid shell. The convergence factor and cross-coupling degree are used as dual indicators to ensure that the magnetic-thermal perturbations do not cancel each other out, thus filling the "last centimeter" quality control blind spot at the end of rapid solidification.

[0020] Directed shrinkage feeding and densification is achieved by integrating soft static pressure and multi-melting point feeding dual channels through shape memory alloy bellows risers. With the help of endoscopic acoustic transmission and surface displacement dual monitoring, the pore volume fraction and stress state are verified in real time, and the pressure slope and feeding rate are dynamically adjusted. Finally, the pressure is smoothly unloaded and the tissue closure is completed, upgrading the "hard pressure-hard top" of the traditional rigid riser to a "soft pressure-intelligent feeding" closed-loop densification system. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The figure is a schematic flow chart of the top pouring rapid solidification forming method of thin-walled complex castings of the present invention. DETAILED DESCRIPTION

[0022] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] See also Figure 1The present invention provides a top pouring rapid solidification forming method for thin-walled complex castings, comprising: Thin-walled complex castings in the aviation, energy, and high-end equipment industries perform multiple functions, including pressure bearing, flow guidance, and heat exchange. Their service reliability directly depends on the uniformity of the microstructure and dimensional stability during the forming stage. Unlike traditional bottom or side pouring, the top pouring process causes the molten metal to dive along the direction of gravity into the slender and multi-turn mold cavity. The overall filling path is an inverted trapezoidal streamline from top to bottom, followed by backfilling of the ribs. When a rapid solidification strategy is introduced to shorten the production cycle, refine the grain size, and reduce macro-shrinkage, the flow field, temperature field, and solute field are rapidly coupled on a sub-second scale. Any slight mismatch at any step may be frozen into an irreversible segregation band or unfused defect before the solid-liquid interface freezes.

[0024] Traditional practices tend to increase the metal superheat or slow down the cooling rate to allow time for diffusion, but this inevitably sacrifices the advantage of refined structure; conversely, relying solely on overall electromagnetic stirring makes it difficult to take into account the local gradients of narrow flow channels inside thin walls, making it difficult to achieve a balance between high-speed solidification and uniform composition.

[0025] Step 1: At the moment of pouring start, the transient multi-source information of the melt is quickly mapped to a unified geometric-physical coordinate system to generate a three-dimensional segregation risk map that can be incrementally corrected over time, providing quantitative input for subsequent gradient magnetic field and heat flow control.

[0026] The initial shock wave of top pouring causes highly nonlinear coupling of the flow field, temperature field and solute field. If the monitoring and model coupling are not timely, any subsequent field control measures may fail due to misalignment.

[0027] Step 101: Multimodal data acquisition and one-time registration The pouring valve opening time is recorded as a unified time base The multimodal monitoring module simultaneously starts the high-frequency ultrasonic array to obtain the instantaneous volume fraction field of the melt, the high-speed infrared camera to obtain the instantaneous temperature field, and the laser induced spectrometer to obtain the instantaneous solute mass fraction field of the melt inlet surface. The three beam signals are written into the buffer array along the same time base to avoid false phase drift caused by time difference.

[0028] Three laser marking points embedded in the mold form a rigid triangular benchmark, and the infrared camera captures the pixel coordinates of the three light spots. , after the projection matrix and measured spatial coordinates Build the homography matrix After the coordinate mapping is completed, the high-frequency ultrasound voxels and spectral pixels are in the same mold coordinate system. The registration is completed once to eliminate the imaging distortion differences between devices.

[0029] In order to solve the infrared noise caused by high temperature radiation and the spectral drift caused by the overlap of solute spectral lines, a weight field is constructed through adaptive threshold filtering. and , combined with the volume fraction field credibility weight Calculate the overall quality factor :

[0030] Among them: comprehensive quality factor : Used to characterize the overall credibility of multi-source data at the same point, the range ; Temperature weight : Output by infrared correction algorithm, range , reflecting the pixel radiation error; Solute weight : Output calculated by spectral line signal-to-noise ratio, range ; Volume fraction weight : Obtained by normalizing the ultrasonic echo intensity, range .

[0031] If a voxel If the value is below the threshold, it will be automatically eliminated or transferred to Kalman interpolation to improve the accuracy of subsequent model alignment.

[0032] Through one-time registration and noise suppression, the three-source data can speak the same language in the same space-time framework, eliminating redundant coordinate conversion for subsequent continuous mapping.

[0033] Through hardware-level "phase locking" and unified time base, the clock deviation and exposure difference of multi-source imaging equipment are eliminated at the data acquisition level, avoiding the huge computational cost of post-time compensation of high-speed flowing molten metal within an extremely short time scale; primary geometric alignment uses three laser beams as a rigid triangulation reference to project all sensors into the same mold coordinate system, achieving "one-time alignment, lifelong use" and reducing the time spent on recalibrating each frame; the dynamic weight field automatically adjusts the credibility according to the real-time imaging quality, allowing data purification and model alignment to proceed in parallel, thereby improving the resilience and stability of the measurement chain.

[0034] Step 102: Model alignment and three-dimensional segregation risk map generation The preset filling-solidification numerical model includes flow equations, energy equations and solute transport equations. According to the temperature field output in step 101, and volume fraction field , quickly optimize the turbulent viscosity coefficient and solid-liquid phase change latent heat coefficient , so that the error between the initial simulation results and the measured results of the model is lower than the convergence threshold .

[0035] Where: Heat flow error : The normalized difference between the model temperature field and the measured temperature field, dimensionless, with a value range of ; Model temperature field :Numerical model predicts temperature, numerical model in spatial location and time Temperature forecast; Measurement area : three-dimensional integration domain, which is the entire thin-walled casting cavity; Turbulent viscosity coefficient , latent heat coefficient : Model adjustable parameters used for iterative correction.

[0036] When the model deviation satisfies When calling the measured solute mass fraction field , model velocity field With temperature gradient Calculating the segregation risk index : Among them: Segregation risk index : The larger the value, the easier it is to produce negative solute segregation, dimensionless; gradient weight coefficient : Empirical coefficient, satisfying ; Reference solute mass fraction : nominal value of alloy composition; reference temperature : Liquidus temperature; reference speed : is the average inlet flow velocity.

[0037] Segregation Risk Index In the mold coordinate system The lower part is spliced ​​into a colored voxel cloud, and the color level is mapped by the risk index value to form a three-dimensional segregation risk map . The map is refreshed every time step That is, calling the Kalman filter to correct the historical trajectory , and add the current moment increment to form a risk evolution sequence that rolls over time.

[0038] Eventually With the optimized model parameter package The segregation map is output to the next step, the gradient magnetic field equalization module, through a 3D interface. This interface includes time scale, spatial resolution, and coordinate system identification, ensuring that downstream modules can directly analyze the data without repeating the analysis process. Through risk index calculation and incremental iteration of the map, high-dimensional instantaneous data is compressed into visual, quantifiable, and traceable field information, providing clear targeting points and priority ranking for gradient magnetic field and heat flow strategies, achieving a seamless connection between the data and control chain.

[0039] The on-site temperature measurement and the numerical model are iteratively coupled in a "small step, fast run" manner. The turbulent viscosity coefficient and latent heat coefficient can be immediately calibrated without waiting for the full range of global data, significantly shortening the "warm-up time" of the model; the segregation risk index simultaneously introduces three-dimensional quantitative standards of composition gradient, temperature gradient and stagnation rate, avoiding the subjectivity of traditional processes that make decisions based on visual observation or a single temperature difference threshold, and embodies the creativity of "multi-field coupling-quantitative grading"; the three-dimensional risk map adopts an incremental iterative method to convert high-frequency noise into a low-pass smooth curve, significantly reducing command bumps without losing spatial resolution.

[0040] By establishing a continuous mapping chain from physical monitoring to numerical modeling and then to risk assessment, the instantaneous multi-source information of the melt is compressed into a dynamic and visual segregation risk map. , and synchronously output the corrected model parameter package, so that the subsequent gradient magnetic field homogenization and heat flow coordinated regulation can obtain a unified and real-time updated decision basis.

[0041] The three-dimensional segregation risk map generated in step 1, along with the calibrated fluid-thermal-composition parameter package, provides quantitative and unified coordinates and thresholds for subsequent field control. The gradient magnetic field homogenization step must respond quickly in the high-temperature, high-electromagnetic-noise, and spatially confined casting environment while avoiding mechanical disturbances to the thin solid shell.

[0042] Step 2: Based on the three-dimensional segregation risk map, detachable coils are synchronously activated at the outer edge of the mold and at key ribs. This creates a soft Lorentz force field that acts in the opposite direction of the solute gradient in real time, guiding the melt microcirculation to diffuse from the edge to the core, reducing local solute enrichment and providing a flow basis for coordinated heat flow regulation. The three-dimensional segregation risk map highlights the presence of high solute gradients in the edge cold zone and the thick-to-thin transition zone at the root of the rib; if these gradients are allowed to evolve, the solid-liquid interface will lock in compositional differences before freezing in seconds. Therefore, a driving force that does not rely on mechanical stirring and can dynamically adjust the intensity and direction within the narrow cavity is required.

[0043] Step 201: Risk Map - Coil Current Primary Mapping First, the three-dimensional segregation risk map voxel cloud is converted into the mold coordinate system The system is divided into three subdomains: the outer edge, the rib transition zone, and the central stagnation zone. Each subdomain is associated with a pre-placed coil group: the outer edge corresponds to the annular array coil, the rib transition zone corresponds to the U-shaped embedded coil, and the central zone corresponds to the axial through-hole coil. This subdomain division not only considers geometric topology but also incorporates voxel risk weighted averaging to align the coil coverage with the risk center, avoiding null excitation and energy waste.

[0044] Master control algorithm call risk index and the corrected model velocity field , solve the minimization objective function in each subdomain : The goal is to adjust the coil current density vector Make the objective function Minimum, thus ensuring that the direction of the Lorentz force is consistent with the direction of risk reduction. The current density instruction is stored in the form of a vector, including amplitude, phase and harmonic identification.

[0045] Where: objective function : A volume energy function that comprehensively evaluates solute gradient and stagnation conditions; is the segregation risk index, is the model velocity field; Solute weight coefficient: , Reflects the importance of risk index; stagnation weight coefficient: , satisfy ; Subdomain integral domain : outer edge area, rib root transition area or central stagnation area; coil current density vector : Contains amplitude and phase, used for driving Zone coil; For each coil current density vector Introducing a cap To avoid local heating of the coil caused by excessive current; at the same time, the electromagnetic-thermal composite model is used to predict the mold wall temperature. If the wall temperature approaches the allowable limit, the coil current density vector is automatically reduced. The amplitude is adjusted and the subdomain boundaries are redivided to ensure the excitation-heat dissipation balance.

[0046] All subdomain coil current density commands are encoded into time-series packets and transmitted via optical fiber with millisecond latency to a detachable coil control box. After decoding, the control box drives the power amplifier to output the corresponding phase-amplitude waveform, and the coil group begins to energize. A single mapping transforms the spatial risk distribution into coil current commands, achieving a one-to-one correspondence between risk and excitation and ensuring the correct direction of the driving force.

[0047] The successful conversion of static risk maps into a dynamic coil excitation scheme allows for rapid reconfiguration of the zones when risk distribution changes, compared to the traditional "coil arrangement first, then performance evaluation" model. This allows for a truly one-to-one correspondence between coil space and risk space. This allows the current density vector to achieve an adaptive balance between risk reduction and thermal safety, avoiding the vicious cycle of "melt homogenization—mold overheating—shell collapse."

[0048] Step 202: Lorentz force-microcirculation dynamic iteration After the excitation is started, the infrared camera and the ultrasonic array synchronously detect the core position and speed of the melt elliptical circulation and calculate the actual circulation vector and model predictions The difference energy function , a measure of circulation prediction error: The difference energy is used to calculate the current density vector of the next cycle The amplitude of the measured velocity field is fine-tuned to reduce the gap between prediction and measurement with the help of the adaptive gain matrix. : 3D velocity vector field obtained by infrared-ultrasonic joint inversion; gain matrix : Adaptive coefficient, used to correct Amplitude, mapping the error gradient into a symmetric positive definite matrix of current correction; Current density vector , and subdomain Bound current density vector time series, including amplitude, phase and harmonic identification; In order to avoid the standing wave generated in the melt by long-term single-frequency excitation, the current density vector of the coil Inject higher harmonic components The harmonic order automatically switches according to the peak value of the risk index. The higher the peak value, the lower the harmonic order, to ensure that high-risk areas receive continuous and gentle driving. When the risk decreases, the harmonic order automatically increases to refine the circulating shear layer and diffuse the solute, avoiding excessive disturbance of the solid shell.

[0049] The transition zone of the reinforcement root is prone to circulation locking due to its narrow geometry. A short-time anti-phase pulse is applied to the area approaching zero, reversing the direction of the Lorentz force by two and a half cycles, generating a local counter-vortex core and breaking the locked structure; then the original phase is automatically restored, keeping the overall transport direction unchanged.

[0050] Difference energy function Each refresh is consistent with the threshold Compare; if it is lower than the threshold for ten consecutive cycles, the circulating current state is considered stable, the system freezes the current vector and notifies the next step of the thermal flow coordinated regulation module to switch the thermal conductive plate mode; if it is higher than the threshold, the harmonic adjustment and anti-phase pulse mechanism are continued until convergence.

[0051] Through the multi-scale adjustment of prediction-measurement-correction cycle and harmonics and back-pulses, the Lorentz force field is synchronized with the real circulation to ensure that the solute transport target is achieved without destroying the integrity of the solid shell.

[0052] Step 201 completes the risk-current primary mapping and converts the spectrum data into a current density vector that can directly drive the coil; Step 202 uses the circulating current difference energy as the core iteration indicator and fine-tunes the coil current density vector through harmonics and anti-phase pulses. , so that the Lorentz force, microcirculation and solute migration remain in dynamic phase.

[0053] The gradient magnetic field homogenization step uses a three-dimensional segregation risk map as its sole input. A single mapping of the coil current locks the excitation space-time layout. The predicted results are then synchronized with the field feedback through a Lorentz force-microcirculation iteration, ultimately constructing a gentle, directional transport network within the thin wall. This network provides ample solute diffusion time for the rib roots and thin-wall transition zones while minimizing the phase difference between the flow-heat-composition coupling and thus preventing the risk of shell tearing.

[0054] The harmonic injection strategy provides the Lorentz force field with adaptive pulse width, enabling it to adjust the driving force scale according to the risk level, thereby quickly leveling out peak risk levels while delicately maintaining uniformity during low-risk periods. Anti-phase pulses precisely target root locks, breaking up local vortex cores through brief directional reversals without disrupting the overall flow pattern. Real-time energy convergence monitoring prevents an endless cycle of excitation, measurement, and parameter adjustment, ensuring system convergence within a manageable timeframe and automatic handover to the next step.

[0055] After this step, the melt temperature-flow distribution has become stable, providing a front-end circulation field for the coordinated regulation of heat flow, and the iteratively converged current vector The three-dimensional interface of the coil flow field is encapsulated synchronously with the circulation status identification to ensure that the heat conducting plate controller can directly read the circulation direction and intensity information, and realize the seamless connection of the next step of heat flow coordinated regulation.

[0056] At the end of the gradient magnetic field homogenization step, a stable microcirculation channel has been established in the melt from the outer edge to the rib root, and the solute field fluctuation has dropped below the risk threshold; however, the solid-liquid interface is still advancing at a high speed. If the heat flow field distribution still follows the traditional unidirectional heat dissipation path, the temperature drop gradient between the rib root and the thick-thin transition zone will widen again, and the shrinkage channel will be closed before it is fully densified, forming a foreshadowing of secondary segregation and thermal cracking. The heat flow coordinated regulation step aims to utilize the bidirectional heat flow control capabilities of the embedded heat conductive plate array and the external resistive heating block to integrate the circulation direction and intensity information received from the coil flow field three-dimensional interface with the real-time temperature-solid phase fraction data, dynamically shaping a radial heat dissipation network with the same heat and flow direction, thereby maintaining the synchronous advancement of the solid-liquid interface in the form of an equipotential surface and extending the flow time of the molten metal within the shrinkage window.

[0057] Step 3: According to the direction of the microcirculation generated by the soft Lorentz force field, the heat conduction mode of the embedded heat conductive plate and the power distribution of the external heating block are switched in real time, so that the heat flux density vector of the rib root and the thick-thin transition zone is consistent with the melt flow direction, maintaining the equipotential advancement of the solid-liquid interface and extending the unobstructed time of the shrinkage channel.

[0058] When microcirculation guides solute diffusion from the edge to the center, it also carries heat in the same direction. If the heat dissipation of the mold wall remains axially symmetrical, a heat-flow phase misalignment will occur between the rib root and the transition zone, causing premature closure of the solidified shell. Therefore, a heat conduction regulation mechanism that can be partitioned and quickly switched is required to align the local thermal driving force with the melt flow direction, allowing the solid shell to follow the circulation rather than collide with the circulation head.

[0059] The specific countermeasures are: first, locate the heat-flow misalignment intensity through heat flux density inversion; then switch the thermal conductivity of the heat conducting plate in different crystal directions and accurately distribute the external heating power; then use the solid-liquid interface equipotential synchronization algorithm to map the thermal conductivity adjustment results to the interface morphology, and finally ensure that the shrinkage feeding window is extended in the rib root direction until the circulation stagnates.

[0060] Step 301: Heat flux inversion and heat conduction mode switching Read the circulation vector field and real-time temperature field in the Coil Flow 3D interface, and invert the current heat flux density vector field using the dual grid finite volume method. After assigning a heat-fluid coupling weight to each voxel, calculate the heat-fluid misalignment index. : Where: Heat-flow misalignment index , where zero represents completely co-directional, and a value close to one represents completely perpendicular. The system divides voxels into co-directional, angled, and reversed regions based on the threshold, providing a regional reference for subsequent heat conduction mode switching.

[0061] Where: Heat-flow misalignment index : dimensionless, range ; Heat flux vector : The direction is the same as heat transfer, unit is W·m⁻²; Measured velocity field : local melt velocity vector, unit: m·s⁻¹.

[0062] The embedded thermal pad utilizes a reversible phase-change material layer laminated with a high-thermal-conductivity graphite layer. Under low magnetic fields, the thermal conductivity along the in-plane plane of the pad is higher than that along the vertical plane. When an alternating magnetic field is applied by an external excitation coil, the phase-change layer's grain orientation deflects, causing the thermal conductivity to rise sharply in the vertical direction, enabling rapid switching between in-plane and volumetric heat conduction. The pad's operating mode is determined by the average value of the heat-flow misalignment index for each subdomain: In the in-plane zone, heat conduction is maintained, allowing the heat flow to dissipate rapidly along the melt flow direction; in the angled zone, heat conduction is switched to volumetric heat conduction, buffering heat-flow angle differences; and in the reversed zone, an alternating mode is activated, causing the pad to cycle back and forth between in-plane and volumetric heat within a single magnetic field cycle, generating a pulsed heat flow that suppresses localized heat backflow.

[0063] Mirroring the heat flow regulation of the thermal pad, the external resistance heating blocks are arranged radially around the rib root. High-frequency pulse power is output to the rib root transition zone, constant low power is output to the outer edge zone, and the central stagnant flow zone is kept warm.

[0064] The power allocation curve is composed of risk residuals By weighted superposition with the heat-flow misalignment index, the principle of more active heat compensation in solute-enriched zones is established to reduce supercooling.

[0065] The final heat flux density vector field is the superposition of the heat flux derived from the thermal conductive sheet and the heat flux released by the heating block. Finite element analysis is used to quickly verify that the superimposed field keeps the heat-flux misalignment index below the upper limit. If it exceeds the upper limit, the thermal conductive sheet excitation or the heating block pulse duty cycle is adjusted until the new threshold is met before proceeding to the next step. The heat-flux misalignment index guides the dual-field regulation of the thermal conductive sheet and the heating block, ensuring that the thermal driving force and the direction of the microcirculation converge in space, laying the foundation for interface synchronization.

[0066] Through high-resolution heat flux density reconstruction and tensor smoothing, the heat-flow vector angle is accurately captured, and the heat-flow misalignment index is introduced as a quantitative indicator in thermal management decisions. For the first time, the anisotropic switching of the thermal conductive sheet and the power distribution of the external heating block are driven by the vector field rather than the absolute temperature threshold. The shape memory alloy-phase change microcapsule composite thermal conductive sheet provides sub-second thermal conductivity reversal capability, which is much faster than traditional mechanical inserts or flow channel switching, and can dynamically accompany the solid-liquid interface under conditions of constantly changing propulsion speed. The power distribution model couples the solute residual with the heat-flow misalignment index to achieve the most active heat compensation in places where solutes are easily enriched.

[0067] Step 302: Solid-liquid interface equipotential synchronization and feeding window extension Using terahertz tomography combined with digital speckle, the propulsion rate of the solid-liquid interface in the three directions of the rib root, outer edge, and center is tracked in real time, and the interface velocity vector field is calculated. ,If the interface velocity deviation in a certain direction exceeds the allowed bandwidth, the equipotential synchronization algorithm is triggered.

[0068] The equipotential synchronization algorithm uses the target interface velocity As a benchmark, by correcting the excitation intensity of the thermal conductive sheet Power coefficient of heating block Iterative approximation, iteration basis: The speed deviation Decompose into subdomains and then map back 、 The iteration continues until all subdomain errors fall within the bandwidth. Interface velocity vector : local interface movement speed m·s⁻¹; target interface speed : Determined by the overall cooling plan; thermal conductivity excitation intensity : Parameter corresponding to magnetic field amplitude, range ; Heating power coefficient : Heating block power normalization coefficient, range .

[0069] As the interface velocity becomes synchronous, the remaining opening time of the rib root and the center feeding channel It is predicted by the diffusion-solidification coupling model. When the value is lower than the safety value, the excitation of the heat conducting plate in the center area is automatically weakened to extend the life of the liquid. At the same time, a high-frequency ramp pulse is injected into the heating block in the outer edge area to delay the closure of the outer edge and form a time difference compensation.

[0070] When the angle between the interface velocity and the microcirculation velocity direction in each subdomain falls below a threshold and the heat-flow misalignment index meets the specified criteria, the system determines the three fields are closed, writes to the synchronized 3D interface for correction, and encapsulates the thermal conductivity pattern table, heating power sequence, and feeding schedule, providing a complete initial state for the spectral-thermal image correction step. Using real-time interface velocity monitoring and feeding window prediction, closed-loop regulation of the heat transfer and heating parameters is achieved, ensuring that the solid-liquid interface advances centripetally in the form of an equipotential surface and that the feeding channel is maintained until the directional feeding densification step is initiated.

[0071] Step 301 addresses heat-flux misalignment by aligning the thermal driving force toward the microcirculation through anisotropic switching of the heat conducting plate and heating power distribution. Step 302 further synchronizes the solid-liquid interface velocity with the heat-flux coupling results and dynamically extends the feeding window. This collaborative approach achieves the three fields of heat, fluid, and phase change, providing an initial equilibrium state for subsequent spectral and thermal image correction and freeing up time and space margins for directional feeding and densification.

[0072] The heat-flux coordination step uses the heat-flux misalignment index as the core metric. By leveraging the anisotropic switching of the thermal conductive plate and the power modulation of the external heating block, a heat dissipation channel in phase with the microcirculation is established. Subsequently, through solid-liquid interface velocity monitoring and feeding channel life prediction, dual-parameter iteration of heat conduction and heating is achieved, maintaining equipotential advance of the solid-liquid interface across the entire region and ensuring the existence of the feeding window until the next step is initiated. The results transform the advantages of magnetic field-induced solute homogenization into structural densification during solidification phase transformation, completely eliminating the risks of secondary segregation and thermal cracking while maintaining a rapid solidification cycle. This lays a solid foundation for the integration of heat-fluid-phase transformation and directional feeding densification.

[0073] When the gradient magnetic field homogenization and the coordinated regulation of heat flow are completed, the melt circulation is consistent with the direction of heat flow, the solid-liquid interface advances synchronously, and the feeding window duration meets the requirements of the top feeding riser. However, in the last few tens of milliseconds of rapidly solidified thin-walled castings, there are still microscopic fluctuations: on the one hand, the solute field appears uniform at the macro level, but trace component drift is formed due to the release of liquid residual heat and local element segregation; on the other hand, the temperature field is smooth overall under the condition of heat and flow in the same direction, but fine-grained temperature distortion may occur due to the warping inertia of the heat conducting plate and the lag of the external pulse heating. If the deviation is not suppressed in time at this stage, the segregation will be locked into a strip-shaped hard and brittle area during the solid-solid transformation, and the temperature distortion will be accompanied by a surge in internal stress and trigger microcracks. The terahertz-speckle joint monitoring integrates the deep-buried interface velocity field and the surface displacement field at one time, avoiding the projection error caused by single-system measurement; the equipotential synchronous power iteration uses the heat conducting plate excitation and heating power as dual-degree-of-freedom parameters, and finely matches the three-dimensional radial interface velocity under the incremental proportional-integral framework, so that the solid-liquid interface changes from non-equidistant to equipotential, significantly reducing the risk of thermal cracking of the rib root.

[0074] The mission of the spectral-thermal image correction step is to perform high-resolution monitoring and local fine-tuning of the solute and temperature when the feeding window enters the end, construct a real-time error field through two channels: spectral absorption peak intensity and short-wave infrared thermal image grayscale, and map the error field into a micro-magnetic field waveform and heat flow power correction instructions. Ultimately, fine needle tissue suturing is completed without destroying the solid shell, retaining a uniform ground state for directional feeding and densification.

[0075] Step 4: Construct a three-dimensional error field using the spectral absorption peak intensity and the short-wave infrared thermal image grayscale. At the end of the feeding window, perform micro-corrections on the gradient magnetic field waveform and heat flux power at the milliwatt-millitesla level to simultaneously return the solute and temperature to the target uniformity.

[0076] Step 401: Error field construction and layered mapping The spectral channel utilizes a ring-shaped fiber matrix, with the fiber endface embedded in a transparent window at the top of the cavity. This fiber shares a synchronous shutter with the laser exciter. The exciter utilizes a dual-band pulse output, with the first band locked to the absorption peak of the main alloy matrix and the second band to the absorption peak of the segregation-prone elements. The detector records peak intensities at the same time base and employs multiple integration to suppress scattering noise. The peak intensity matrix is ​​then subjected to a Hamming window short-time Fourier transform, converting the time-domain pulse tails into frequency-domain weights, from which the instantaneous solute distribution can be inferred.

[0077] The thermal imaging channel uses an InGaAs area array sensor, achieving a frame rate twice as fast as spectral sampling. Each thermal image frame undergoes non-uniform response correction, followed by wavelet threshold filtering to remove RF speckle. The processed grayscale matrix is ​​mapped to a temperature matrix and aligned with the spectral matrix through sub-pixel motion compensation, ensuring a one-to-one correspondence between the two channels in spatial coordinates.

[0078] Calculate the spectral equalization error index for each voxel separately: And the temperature equilibrium error index: Where: Spectral Balance Error Index : voxel solute deviation, range ; Target solute mass fraction : Step 3 output mean; temperature balance error index : voxel temperature deviation, range Target temperature : heat flow co-regulation prediction value; right and Perform weighted superposition to form comprehensive correction energy : Where: Weight coefficient Determined by the process's sensitivity to composition and temperature and meeting .

[0079] according to The voxels are divided into three layers: high-risk, medium-risk, and stable. The high-risk layer is primarily located at the end of the slender channel at the rib root and in the remelting zone at the outer edge; the medium-risk layer is annularly surrounding the high-risk layer; and the stable layer is distributed in the remaining locations. The stratification results are written into the error field label set, providing coordinate indexes for the next micro-control step. Through dual-channel synchronous sampling and error index quantization, step 401 outputs a three-dimensional stratified error field in milliseconds, providing the controller with a precise target.

[0080] Through dual-band excitation and synchronous optical gates, the spectral sampling matrix can simultaneously monitor matrix elements and segregation-sensitive elements, avoiding segregation misjudgment caused by traditional single-peak inference; multi-band integration and Hamming window transformation project the pulse tail into the frequency domain and then restore the peak, effectively suppressing reflection noise and stabilizing the solute mass fraction estimation; InGaAs thermal imaging adds inter-frame dark field correction and wavelet threshold filtering to ensure that the temperature matrix maintains a linear response in a severe radiation background; error field stratification introduces connectivity constraints to avoid unnecessary adjustments triggered by isolated voxels, demonstrating high robustness at the thin-wall microscale.

[0081] Step 402: Adaptive magnetothermal dual modulation and convergence verification The controller first reads the error field tag set, locks the coordinates of the high-risk layer, and then calls the running gradient magnetic field module to add a small sinusoidal perturbation to the current density vector of the corresponding coil. The amplitude is capped at 1% of the original current amplitude, and the frequency is twice the original harmonic frequency to ensure that the perturbation does not disrupt the stable circulation topology. The perturbation direction is aligned with the opposite direction of the solute gradient, so that the Lorentz force produces a localized pin-pulling effect within the high-risk layer, promoting the outward diffusion of residual segregated elements.

[0082] If a high-risk layer is accompanied by temperature deviations, the thermal pad excitation and heater block power interfaces are activated, outputting fine-tuning commands using a proportional-integral-limiting strategy. The thermal pad switches between high-frequency, low-amplitude warping within the high-risk layer to delay rapid heat dissipation. The heater block outputs a single pulse of negative step power to reduce local temperature peaks and prevent overcompensation. In medium-risk layers, the system adjusts only the heater block duty cycle without altering the magnetic field to prevent interference with circulating currents.

[0083] After the controller injects the perturbation, it calculates the real-time convergence factor every two sampling periods: in, is the comprehensive correction energy; : convergence factor : error energy change rate of high-risk layer; High-risk layer integral domain : Error field high-risk layer voxel set; time step : Interval between two sampling cycles like Continue for five cycles, indicating that the overall error decreases; if , the controller automatically weakens the magnetic field disturbance amplitude and extends the heat flux fine-tuning pulse width to avoid error rebound.

[0084] To prevent the magnetic field and heat flux modulation from canceling each other out, the cross-coupling degree is calculated in each subdomain. : Where: is the coil current density vector; and the subdomain The corresponding coil group at time The current density vector; is the heat flux density vector, which is the local heat flux direction and intensity after inversion in step 301 and superposition of heat conduction and heating; If the cross coupling If the threshold is exceeded, it indicates that the Lorentz force and heat flux are in the same direction, resulting in local overdrive. The amplitude is immediately reduced and the heat flux modulation phase is delayed by half a cycle to achieve decoupling. Through extremely small amplitude magnetic-thermal dual modulation and real-time convergence factor monitoring, high-risk layer error attenuation is achieved without interrupting the main circulation and heat dissipation state.

[0085] Error field construction transforms spectral and thermal imaging data into actionable three-dimensional energy maps. Adaptive magnetothermal dual modulation perturbs the magnetic field and heat flow according to the energy gradient, achieving rapid and stable error cancellation through convergence factors and cross-coupling. Continuing to use the previously described circulation and heat flow parameters, the system outputs a residual convergence indicator and a temperature-composition equilibrium state, providing a final uniform solute field and stress-controlled temperature field for targeted feeding and densification.

[0086] The spectral-thermal image correction step integrates dual-channel high-speed monitoring with micro-amplitude magnetic-thermal coupling control. Through error index quantization, layered label positioning, perturbation command injection, and convergence factor closure, it achieves last-centimeter correction of residual segregation and localized temperature distortion at the rib root and outer edge. Its output format uses the same protocol as the preceding synchronous 3D correction interface, including a residual convergence flag, a magnetic field perturbation spectrum, and a heat flux fine-tuning sequence. Upon receiving the zero residual flag, the top feeding and densification module enters the gentle static pressure phase, ensuring a seamless transition across the entire process.

[0087] In a relatively stable circulation-heat flow system, only magnetic field perturbations of one percent amplitude and milliwatt-level heat flow fine-tuning are allowed, reflecting the respect for the integrity of the solid shell; the pseudo-random pulse and mobile "thermal curtain" strategy allows the magnetic-thermal perturbations to present a non-repeating pattern in both space and time, which can significantly reduce the risk of local residual segregation being "frozen" by the solid-solid interface; the localization of the convergence factor and the monitoring of the cross-coupling degree ensure that the adjustment will not cause new disturbances globally.

[0088] After four-level coupling of real-time benchmark mapping, gradient magnetic field averaging, heat flow coordinated regulation and spectrum-thermal image correction, the solute field and temperature field inside the thin-walled complex casting have entered a zero residual state. The solid-liquid interface converges synchronously in the direction of the top riser in the form of an equipotential surface, and the rib root-outer edge channel still maintains the shrinkage opening time. However, the final stage of rapid solidification still hides two risks: First, the release of latent heat from crystal growth causes the liquid zone to shrink continuously. If the matching between the feeding pressure and the feeding amount lags, hidden shrinkage cavities will form. Second, during the solid-solid transition, dendrite interlocking creates microcracks. If the external static pressure is too high, this can induce cold cracking or deformation rebound.

[0089] By precisely applying pressure and releasing soluble feeding material through the retractable shrinkage feeding riser and simultaneously monitoring the internal pore volume fraction and surface micro-displacement in real time, the three goals of shrinkage compensation, tissue closure and stress relief are completed simultaneously within a second window, ultimately obtaining a thin-walled complex casting with no shrinkage, no cracks and uniform tissue.

[0090] Step 5: Within the end of the feeding window, use the top retractable feeding riser to simultaneously apply soft static pressure and release the soluble feeding material in a timed manner, and control the pore volume fraction in a real-time closed-loop manner until it reaches the target threshold and relieves residual stress.

[0091] Step 501: Feeding path planning and static pressure curve preset After zero residual, the melt is in a uniform steady state, and the remaining feeding window Because the flow rate is limited, a dual-parameter static pressure-feeding path must be established to ensure that the pressure rise slope matches the soluble feed release rate. Therefore, the predicted pore formation rate is first analyzed, and the static pressure curve is then planned according to the minimum energy principle. The dissolution-diffusion sequence of the soluble feed is then calculated using the solid-solid transition rate. Finally, the two paths are packaged into a dense path 3D interface for reading by the execution unit.

[0092] The controller calls the last temperature field in the spectrum-thermal image correction stage Solid phase fraction , and combined with the converged turbulent viscosity coefficient and latent heat coefficient , using the volume fraction conservation equation to invert the expansion-contraction rate of the residual liquid zone Where: Volume expansion coefficient : alloy thermal expansion coefficient, dimension K⁻¹; residual liquid integration domain : Define area; pore generation rate : Unit: m³·s⁻¹, indicating volume shrinkage rate; To avoid cold cracking, static pressure is increased over time. Adopting a three-stage slow-rise model: Parameter constraints: , ensure slow at the beginning and fast at the end; Soft static pressure curve : Pressure exerted by the top riser; Segment slope : rate of pressure rise, unit: Pa·s⁻¹; time segment :Depend on Determination of extreme points; peak pressure : Less than 50% of the solid shell yield strength; The core rod of the feeding material is made of low melting point alloy rod of the same series. The rod body is divided into three sections along the axial direction, high, medium and low melting point areas. and pore generation rate Coupled, calculate the feed mass flow rate: Among them: Feed efficiency : Consider the ratio of dissolution-diffusion loss, ranging from 0 to 1; alloy density : Unit: kg·m⁻³; feeding mass flow rate : unit kg·s⁻¹; For soft static pressure curve and Perform extreme value check, if at any moment or The system automatically and smoothly refits the pressure schedule and the material feeding schedule. Finally, the pressure schedule and material feeding schedule are packaged into the dense path 3D interface and sent to the execution layer simultaneously with the safety valve pressure relief settings. A valid path flag is also written. Through pore generation prediction, static pressure curve fitting, and material feeding schedule calculation, a safe and efficient pressure-feeding coordinated path is pre-planned.

[0093] Through fine-grained inversion of the pore formation rate, combined with high-low dual-slope static pressure curves and multi-melting point segmented feeding mode, the system enables the pressure-feeding dual parameters to truly form a one-to-one mapping relationship on the time axis for the first time, avoiding the timing misalignment caused by the traditional methods of "first pressure, then feeding" or "first feeding, then pressure".

[0094] Step 502: Soft static pressure execution and real-time verification of density With the path table, the execution layer must precisely execute within a time window of seconds, verifying in real time whether density and stress conditions meet the standards, and fine-tuning material feeding and pressure as necessary. The process begins with a retractable riser with position correction, then continues with pressure application and real-time tracking of pore volume fraction. Finally, once density reaches the threshold, smooth pressure relief completes the tissue closure.

[0095] The riser is composed of a shape memory alloy bellows and a high-modulus ceramic valve core. The preset length of the bellows is equal to the original design riser height. The controller activates the bellows through the thermostrictive socket. The bellows stretches to press the valve core against the top of the solid shell and form a metal-ceramic-metal three-layer sealing gasket, ensuring no leakage during static pressure and the sealing fatigue margin meets ten times the cycle life.

[0096] Hydraulic servo valve follows a soft static pressure curve Output pressure oil to drive the bellows and valve core to pressurize the melt; at the same time, the core rod of the feeding material is pressed by the resistance fuse. The curve is heated and melted in sections. To ensure dual-channel alignment, the oil pressure sensor and the resistor fuse current values ​​are collected every millisecond. If the oil pressure-current phase difference exceeds the safety belt, the oil pressure command is immediately interpolated and corrected.

[0097] The ultrasonic phased array probe adopts a dual-frequency composite mode of array surface and array center. One frequency band focuses on detecting the rib root, while the other frequency band wide beam covers the center area and outputs the pore volume fraction in real time. At the same time, digital speckle is projected onto the casting surface to track the displacement vector , calculate the stress criterion: like This indicates that the surface displacement is abnormal and the static pressure rise rate should be slowed down.

[0098] when After ten sampling cycles, smooth pressure relief is triggered: the hydraulic valve decompresses according to the three-stage curve, and the core rod enters the cooling and locking stage to ensure that the feed metal is completely fused and cannot be sucked out. After pressure relief is completed, the system writes the densification completion flag to the master control database.

[0099] Step 501 outputs the three-dimensional interface of the densified path of the pressure-feeding path table optimized by multiple constraints; Step 502 completes the riser positioning, static pressure-feeding synchronization, pore-stress dual monitoring and smooth pressure relief under the instructions of the path table, and finally writes back the densification completion mark to establish a defect-free benchmark for the demolding-post-processing process.

[0100] The directional feeding densification step utilizes a two-layer system of planning first, then execution, organically coupling the residual porosity generation rate, solid-solid transformation dynamics, and riser-feeding hardware capabilities. This not only achieves precise synchronization of gentle static pressure and quantitative feeding, but also relies on dual acoustic and speckle monitoring to achieve a real-time closed-loop between internal and external densification and stress release. Thus, the top pouring rapid solidification forming method completes five-level serial control, from melt filling, magnetic and thermal synergy, tissue suturing, to final densification. A digital interface runs throughout the entire process, ensuring systematic assurance of the structural integrity and microstructure uniformity of thin-walled, complex castings.

[0101] The shape memory alloy bellows-ceramic valve core composite riser provides a solution that combines micro-displacement tracking capability with high-temperature airtightness, breaking through the limitation of ordinary rigid risers that cannot adapt to the dynamic expansion and contraction of dendrites; the hydraulic-feeding synchronous closed loop realizes the active coordination of feeding and contraction dynamics through real-time correlation coefficient regulation; the internal sound-transmitting-external displacement dual monitoring incorporates internal pore reduction and external stress release into the judgment criteria at the same time, preventing only focusing on internal densification while ignoring external cracks; the smooth pressure relief curve combined with the shape memory alloy reverse heating compensation ensures that there is no secondary air absorption and seam riding at the interface in the final stage.

[0102] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0103] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is only for some logical functions. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0105] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0106] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A top pouring rapid solidification forming method for thin-walled complex castings, characterized by: include, At the moment the pouring valve is opened, the melt flow field, solid-liquid interface and solute distribution data are collected and matched with the preset filling-solidification model, generating a three-dimensional segregation risk map updated over time in real time for subsequent field control calls; Based on the three-dimensional segregation risk map, the outer edge and rib root coils are activated, and a programmable current is used to generate a soft Lorentz force against the solute gradient, stabilize the solute field distribution, and guide the melt microcirculation to diffuse from the edge to the core; Read the circulation vector, switch the heat conduction plate plane or volume heat conduction and allocate the heating block power in different zones, so that the heat flow direction of the rib root and transition zone is consistent with the circulation, and maintain the synchronous advancement and stability of the solid-liquid interface; Synchronously collect dual-band spectra and short-wave infrared thermal images, construct an error field and calibrate the high-risk layer, inject a micro-magnetic field and milliwatt pulse heat flow into it, iterate until the error converges and output a mark; The shape memory alloy bellows riser is stretched and sealed, the hydraulic servo valve applies pressure according to the static pressure curve and simultaneously melts the segmented feeding core rod, the pore volume fraction and speckle tracking displacement are monitored acoustically, and after meeting the standards, the pressure is smoothly released to complete the densification.

2. The top pouring rapid solidification forming method according to claim 1, characterized in that: The ultrasonic array, infrared camera and laser-induced spectrometer are sampled synchronously using a unified time base, and the three-way voxels are registered to the mold coordinate system at one time through a three-point laser reference, generating a voxel stack with a weighted quality factor, providing high-reliability raw data for the three-dimensional segregation risk map.

3. The top pouring rapid solidification forming method according to claim 2, characterized in that: The temperature-velocity dual calibration algorithm is used to iteratively correct the turbulent viscosity coefficient and latent heat coefficient, and the segregation risk index is calculated after the temperature difference error of the numerical model is lower than the set threshold. A rolling updated three-dimensional segregation risk map is generated through Kalman incremental stitching.

4. The top pouring rapid solidification forming method according to claim 3, characterized in that: According to the partition results of the three-dimensional segregation risk map, annular array coils, U-shaped embedded coils and axial through-hole coils are called, and the current density vector is optimized using a genetic algorithm. A time series package with amplitude, phase and harmonic identification is output to drive the corresponding coils.

5. The top pouring rapid solidification forming method according to claim 4, characterized in that: After excitation, the measured circulating current is compared with the difference energy predicted by the model in real time. The coil current is fine-tuned through the adaptive gain matrix and high-order harmonics or short-time reverse pulses are injected to break the root-locked vortex and converge the difference energy to below the preset threshold.

6. The top pouring rapid solidification forming method according to claim 5, characterized in that: The heat flux density vector field is inverted by the dual grid finite volume method. After calculating the heat-flux misalignment index, the heat conducting plate is switched between plane heat conduction and volume heat conduction mode, and pulse power is output to the rib root heating block according to the weighted solute residual. Terahertz tomography and digital speckle patterning are used to jointly monitor the solid-liquid interface velocity. The excitation intensity of the thermal conductive plate and the heating power coefficient are adjusted according to the incremental proportional-integral method, so that the three-dimensional interface velocity decreases synchronously and the opening time of the feeding channel is dynamically extended.

7. The top pouring rapid solidification forming method according to claim 6, characterized in that: The solute peak intensity is obtained by exciting a ring fiber array with dual-band laser pulses, and short-wave infrared thermal images are collected simultaneously. After calculating the spectral equilibrium error index and temperature equilibrium error index, the voxels are stratified into three levels: high-risk, medium-risk, and stable according to the energy threshold.

8. The top pouring rapid solidification forming method according to claim 7, characterized in that: A sinusoidal perturbation of 1% of the original current and a pseudo-random pulse are applied to the high-risk layer coil. The corresponding thermal conductive sheet performs a high-frequency small-amplitude warping and makes the heating block output a negative step power. The convergence factor and cross-coupling degree are calculated every two cycles and the amplitude modulation or delay is adaptively performed.

9. The top pouring rapid solidification forming method according to claim 8, characterized in that: The pore formation rate is inverted based on the octree-encrypted temperature-solid fractional field, the soft static pressure curve is segmentedly fitted, and the multi-melting point feeding mass flow rate is calculated. The packaging pressure and feeding schedule are then issued after verification through the safety valve.

10. The top pouring rapid solidification forming method according to claim 9, characterized in that: The riser of the shape memory alloy bellows is stretched to press the ceramic valve core and form a three-layer seal. The hydraulic servo valve applies pressure according to the pressure schedule and the fuse melts the core rod in sections. The pore volume fraction is monitored acoustically and the surface displacement is tracked with digital speckle patterns. When the densification threshold is met, the pressure is released according to the three-segment curve and the bellows is driven to retract to complete the densification.

Images