Aluminum alloy die casting process data processing method and system based on digital twinning

Abstract

This invention discloses a data processing method and system for aluminum alloy die casting based on digital twins, belonging to the field of detection and control technology. The method includes: acquiring time-series data of the collection dimension; performing time misalignment compensation based on heat and pressure transfer attributes to obtain physical synchronization data and reference values; subtracting the reference value from the measured value to obtain the initial deviation, and converting it into a deviation rate by combining the physical response constant; generating a dynamic threshold based on real-time solid fraction and physical properties; statistically analyzing the proportion of values ​​exceeding the threshold and combining it with complexity parameters for correction to obtain a calculated index; constructing an energy conservation limiting model to perform thermodynamic limiting and obtain the target adjustment index; calculating the correction amount and performing translation compensation on the measured values ​​to generate target fused data input to the twin for solution. This invention has the advantages of synchronous processing, dynamic adaptation, and stability and reliability.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and specifically to a data processing method and system for aluminum alloy die casting based on digital twins. Background Technology

[0002] In the real-time digital twin mapping of the aluminum alloy die casting process, due to the complex working conditions inside the die casting machine and mold, the system needs to integrate multi-dimensional sensor data such as injection pressure, multi-point temperature of the mold and cavity stress. However, it faces the problem of inconsistency in physical properties in the coordination of multi-source data and state self-consistency processing.

[0003] First, the pressure transmission speed during the die-casting filling stage is on the order of the speed of sound, while the heat diffusion in the mold steel is relatively slow. This difference in transmission delay across media can cause data collected by external sensors at the same time stamp to be physically misaligned, making direct numerical fusion prone to deviations in the boundary conditions of the twin model. Second, industrial sensor packaging components typically have inherent thermal inertia and response hysteresis, and there is a lack of unified measurement standards between different physical dimensions. Conventional algebraic subtraction operations are insufficient to separate the sensor's own polarization hysteresis error. Furthermore, when aluminum alloys solidify and cool, crossing the liquidus and solidus lines to enter the mushy region, the release of latent heat can cause a plateau in the local cooling curve. If a static, fixed tolerance threshold is used to remove abnormal data, fluctuations caused by material phase transitions can easily be misjudged as sensor data drift, leading to data calibration errors. Furthermore, conventional data coordination algorithms typically rely on proportional amplitude limiting operators. When aligning multidimensional temperature and pressure deviations, they do not fully consider the energy conservation laws of physical entities. This data translation operation may introduce additional heat and mechanical energy calculation variables into the digital twin model. If the introduced values ​​exceed the physical range of the latent heat of solidification of the die casting, it may lead to iterative divergence of the multiphysics partial differential equations. At the same time, it may also smooth out the high-frequency oscillation characteristics that originally reflected the turbulence of the molten metal and micro-shrinkage defects, thereby affecting the fidelity and reliability of the twin model in reproducing the die casting process state. Summary of the Invention

[0004] To address the technical problems existing in the background art mentioned above, the present invention provides a method and system for processing aluminum alloy die casting process data based on digital twins.

[0005] A data processing method for aluminum alloy die casting process based on digital twins includes: acquiring time-series data of multiple acquisition dimensions of the process entity and counting the total number of acquisition dimensions; performing time misalignment compensation on the time-series data of the acquisition dimensions based on the heat transfer and pressure transfer properties of the medium to acquire physical synchronization data and corresponding independent reference values; subtracting the independent reference values ​​from the current measured values ​​of the physical synchronization data to obtain the initial deviation, and converting the initial deviation into a deviation rate by combining the physical response constant; acquiring the real-time solidity parameter of the process entity, and generating a dynamic threshold based on the real-time solidity parameter and alloy physical properties; counting the number of acquisition dimensions whose absolute value of the deviation rate exceeds the dynamic threshold, dividing them by the total number to obtain a preliminary adjustment index; correcting the preliminary adjustment index by combining the dimensional complexity parameter to obtain a calculated index; constructing a control volume energy conservation limiting model, thermodynamically limiting the calculated index based on the alloy specific heat capacity and latent heat parameters to obtain a target adjustment index; multiplying the initial deviation by the target adjustment index to obtain a correction amount, performing translation compensation on the current measured values ​​based on the correction amount to obtain target fused data, and inputting the target fused data into the digital twin for state calculation.

[0006] Optionally, time misalignment compensation is performed on the time-series data of the acquisition dimensions based on the thermal transfer properties and pressure transfer properties of the medium to obtain physically synchronized data, including: obtaining the spatial coordinate parameters corresponding to each acquisition dimension; calculating the physical transmission delay difference of the time-series data of different acquisition dimensions reaching the same observation point based on the spatial coordinate parameters, combined with the pressure transfer property representing the pressure transfer wave velocity and the thermal transfer property of the medium representing the thermal diffusivity; and performing translation compensation on the time axis of the time-series data of the acquisition dimensions according to the physical transmission delay difference to generate physically synchronized data with physical cross-section alignment.

[0007] Optionally, the initial deviation is converted into a deviation rate by combining the physical response constant, including: obtaining the hysteresis compensation rate for quantifying the response hysteresis characteristics by combining the rate of change of the current measured value of the time series data of each acquisition dimension with the response time constant of the corresponding acquisition dimension; dividing the initial deviation by the independent reference value to obtain the basic relative error; and adding the basic relative error to the hysteresis compensation rate to obtain the deviation rate after compensating for thermal inertia hysteresis.

[0008] Optionally, a dynamic threshold is generated based on real-time solidity parameters and alloy physical properties, including: obtaining the basic tolerance value in the liquid phase state, and obtaining the phase transformation relaxation coefficient based on the latent heat of melting parameter of the aluminum alloy, the average specific heat capacity parameter of the mushy region, and the temperature span difference between the liquidus and solidus lines; obtaining the currently measured alloy melt temperature, and calculating the real-time solidity parameter based on the relative proportional difference between the alloy melt temperature, solidus temperature parameter, and liquidus temperature parameter; dynamically scaling the phase transformation relaxation coefficient using the real-time solidity parameter, and adding the scaling result to the basic tolerance value to generate a dynamic threshold.

[0009] Optionally, the preliminary adjustment index is corrected by combining the dimensionality complexity parameter to obtain the computational index, including: obtaining the number of standard dimensions that match the digital twin solution model; determining whether the total number exceeds the number of standard dimensions; if so, dividing the difference between the total number and the number of standard dimensions by the number of standard dimensions to obtain the dimensionality complexity parameter; if not, setting the dimensionality complexity parameter to zero; adding the dimensionality complexity parameter to a constant to obtain the correction ratio; and multiplying the preliminary adjustment index by the correction ratio to generate the computational index.

[0010] Optionally, an energy conservation limiting model for the control volume is constructed. Based on the alloy specific heat capacity and latent heat parameters, the calculated index is thermodynamically limited to obtain the target adjustment index, including: calculating the maximum latent heat limit energy that can be released during the solidification stage of the casting based on the product of the liquid alloy density, casting volume, and latent heat parameters; multiplying and summing the mold density parameter, mold specific heat capacity parameter, volume parameter representing the local control volume, and the absolute value of the initial deviation to obtain the total virtual energy change when the current deviation is fully corrected; dividing the maximum latent heat limit energy by the total virtual energy change to obtain the energy tolerance ratio; and selecting the minimum value between the calculated index and the energy tolerance ratio as the target adjustment index.

[0011] Optionally, the initial deviation is obtained by subtracting the independent reference value from the current measured value of the physical synchronization data, including: obtaining a calculated difference by subtracting the independent reference value from the current measured value of the physical synchronization data; determining whether the calculated difference representing the temperature is lower than the set background thermal noise boundary; if so, calibrating the calculated difference as an effective offset based on the industrial thermal inertia characteristics and using it as the initial deviation; if not, directly using the calculated difference as the initial deviation; and obtaining the target fusion data by performing translation compensation on the current measured value based on the correction amount, including: extracting the high-frequency oscillation features caused by fluid turbulence in the current measured value; and in the process of performing translation compensation on the current measured value based on the correction amount, only the baseline offset caused by the response temperature drift is offset and translated, retaining and outputting the target fusion data carrying the high-frequency oscillation features.

[0012] A data processing system for aluminum alloy die casting based on digital twins is also provided, including: a data compensation module for acquiring time-series data of multiple acquisition dimensions of the process entity and counting the total number of acquisition dimensions; performing time misalignment compensation on the time-series data of the acquisition dimensions based on the heat transfer properties and pressure transfer properties of the medium to acquire physical synchronization data and corresponding independent reference values; a deviation evaluation module for subtracting the independent reference value from the current measured value of the physical synchronization data to obtain the initial deviation, and converting the initial deviation into a deviation rate based on the physical response constant; and a dynamic threshold module for acquiring the real-time solidity parameter of the process entity and, based on the real-time solidity parameter and alloy physical properties... The system consists of several modules: a dynamic threshold generation module; an initial index calculation module, which counts the number of data collection dimensions whose absolute deviation exceeds the dynamic threshold, divides it by the total number to obtain the initial adjustment index; a dimensionality complexity parameter is used to correct the initial adjustment index to obtain the calculated index; an energy limiting module, which constructs an energy conservation limiting model for the control volume, thermodynamically limits the calculated index based on alloy specific heat capacity and latent heat parameters to obtain the target adjustment index; and a twin fusion module, which multiplies the initial deviation by the target adjustment index to obtain the correction amount, performs translation compensation on the current measured value based on the correction amount to obtain the target fused data, and inputs the target fused data into the digital twin for state calculation.

[0013] Optionally, the dynamic threshold module is also used to: obtain the basic tolerance value in the liquid phase state, and obtain the phase transformation relaxation coefficient based on the latent heat of melting parameter of the aluminum alloy, the average specific heat capacity parameter of the mushy region, and the temperature span difference between the liquidus and solidus lines; obtain the currently measured alloy melt temperature, and calculate the real-time solid fraction parameter based on the relative proportional difference between the alloy melt temperature, solidus temperature parameter, and liquidus temperature parameter; dynamically scale the phase transformation relaxation coefficient using the real-time solid fraction parameter, and add the scaling result to the basic tolerance value to generate a dynamic threshold.

[0014] Optionally, the energy limiting module is also used to: calculate the maximum latent heat limit energy that can be released during the solidification stage of the casting based on the product of the liquid alloy density, casting volume, and latent heat parameters; multiply and sum the mold density parameter, mold specific heat capacity parameter, volume parameter representing the local control volume, and the absolute value of the initial deviation to obtain the total virtual energy change when the current deviation is fully corrected; divide the maximum latent heat limit energy by the total virtual energy change to obtain the energy tolerance ratio; and select the minimum value between the calculated index and the energy tolerance ratio as the target adjustment index.

[0015] The beneficial effects of this invention are reflected in:

[0016] In the entire data processing method for aluminum alloy die casting based on digital twins, firstly, a physical transmission delay misalignment compensation mechanism based on the thermal transfer properties of the medium and the pressure wave velocity is adopted to alleviate the spatiotemporal lag differences in the propagation of multimodal signals inside the mold, providing synchronous data for physical section alignment for subsequent model calculations. Furthermore, by combining the sensor's response time constant to extract the hysteresis compensation rate, the response hysteresis caused by hardware packaging is compensated, and the absolute numerical difference including background thermal noise is converted into a cross-dimensional deviation rate, making the offset assessment of each dimension closely approximate the transient change process of the melt. Furthermore, a dynamic threshold is constructed using the alloy's latent heat of fusion and real-time solidity, enabling the system to adapt to the temperature plateau period caused by the release of latent heat in the mushy region, reducing the misjudgment of physical fluctuations caused by alloy phase transformation as sensor measurement drift, and mitigating the computational divergence risk that may arise during high-dimensional data aggregation by using dimensional complexity parameters. Furthermore, an energy conservation limiting model for the control volume is constructed to restrict the data correction amount to the maximum latent heat range of the casting, reducing the possibility of thermodynamic logic conflicts caused by energy errors introduced into the digital twin during data translation operations. At the same time, when compensating for measured values ​​during translation, the high-frequency oscillation characteristics caused by fluid turbulence are preserved, ensuring that the target fusion data input to the digital twin maintains the integrity of microscopic dynamic details while satisfying the macroscopic energy conservation boundary conditions. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 This is a schematic diagram illustrating the steps of the data processing method for aluminum alloy die casting based on digital twins according to the present invention;

[0019] Figure 2 This is a schematic diagram of a portion of step S1 in the data processing method for aluminum alloy die casting based on digital twins of the present invention;

[0020] Figure 3 This is a schematic diagram of a portion of step S2 in the data processing method for aluminum alloy die casting based on digital twins of the present invention;

[0021] Figure 4 This is a schematic diagram of a portion of step S3 in the data processing method for aluminum alloy die casting based on digital twins of the present invention;

[0022] Figure 5 This is a schematic diagram of a portion of step S4 in the data processing method for aluminum alloy die casting based on digital twins of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms first, second, etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] This invention provides a data processing method for aluminum alloy die-casting process based on digital twins, such as... Figure 1 As shown, in one embodiment, the method includes:

[0027] S1. Acquire time-series data of multiple acquisition dimensions of the process entity, and count the total number of acquisition dimensions. Perform time misalignment compensation on the acquisition dimension time-series data based on the heat transfer and pressure transfer properties of the medium, and acquire physical synchronization data and corresponding independent reference values.

[0028] S2. Subtract the independent benchmark value from the current measured value of the physical synchronization data to obtain the initial deviation, and combine the initial deviation with the physical response constant to convert the initial deviation into the deviation rate;

[0029] S3. Obtain real-time solid fraction parameters to generate dynamic thresholds, and combine them with dimensionality complexity parameters to calculate preliminary adjustment indicators and calculation indicators.

[0030] S4. Construct a control volume energy conservation and limiting model to obtain the target adjustment index, then obtain the correction amount and perform translation compensation on the current measured value, generate target fusion data, input the digital twin for state calculation.

[0031] In this embodiment, it should be noted that in S1, multiple acquisition dimensions of the process entity's time-series data are first acquired, and time misalignment compensation is performed to address the issue of differences in the cross-medium transmission delay of different physical quantities during the die-casting filling stage. Taking the integrated die-casting of an automotive engine block as an example, 12 acquisition dimensions are deployed inside the mold, including 4 pressure sensors and 8 temperature sensors. Since the transmission speed of pressure waves in liquid aluminum alloy is approximately 4500 m / s, while the diffusion of heat in mold steel is limited to a thermal diffusivity of approximately 0.0000075 m² / s, for a temperature measurement point with a burial depth of 5 mm, the heat conduction delay is approximately 3.33 s, while the pressure wave delay is only approximately 0.0000011 s. Based on spatial coordinate parameters, the pressure data stream at time t and the temperature data stream at time t minus 3.33 s are shifted and aligned to generate synchronous data characterizing the state of the physical space at the same instant, and independent reference values ​​such as the peak injection pressure of 32.0 MPa and the steady-state temperature of 823.15 K are extracted. This misalignment compensation technique based on the physical properties of the medium alleviates the spatiotemporal lag differences in the propagation of multimodal signals inside the mold, providing synchronous data from the same physical cross section for subsequent model solving.

[0032] In S2, the initial deviation is obtained by subtracting the independent reference value from the current measured value after physical synchronization processing. This initial deviation is then converted into a deviation rate by combining the physical response constant, thereby eliminating the inherent thermal inertia and hysteresis errors of the industrial sensor hardware package. Continuing with the aforementioned temperature dimension processing, if the currently measured temperature is 827.15K, subtracting the reference value of 823.15K yields an initial deviation of 4K. This value is taken as the effective offset after determining that it exceeds the background thermal noise boundary of 0.1K. Considering that the temperature change rate detected by the sensor may be as high as 20K / s during the transient phase of die casting filling, a physical response constant corresponding to the sensor at this temperature measurement node (0.05s) is specifically introduced. By dividing the initial deviation by the reference value to obtain the basic relative error, and adding the polarization attenuation term derived from the rate of change and the physical response constant, the final calculated deviation rate after compensating for thermal inertia hysteresis is 0.6072%. This processing mechanism effectively extracts the hysteresis of the sensor during the dynamic heating process, transforms the absolute numerical difference into a cross-dimensional relative evaluation benchmark, and makes the offset assessment of each dimension closer to the transient physical change process of the molten metal, reducing the data evaluation error caused by direct algebraic calculation.

[0033] In S3, real-time solidity parameters are acquired to generate a dynamic threshold. Preliminary adjustment and calculated indices are then combined with dimensional complexity parameters to address the physical plateau caused by latent heat release during the alloy's solidification and cooling into the mushy region. When a localized casting temperature of 833.15 K is identified, falling between the liquidus (853.15 K) and solidus (788.15 K), the real-time solidity parameter is calculated to be 0.307 based on the temperature difference. Using a basic tolerance of 0.03 and a latent heat to sensible heat ratio of 6.23 calculated from latent heat and specific heat capacity, the phase transformation relaxation coefficient is dynamically scaled, resulting in a relaxed dynamic threshold of 8.73%. Traversing 12 acquisition dimensions, three dimensions were found to have deviation rates exceeding this threshold, leading to a preliminary adjustment index of 0.25. Furthermore, since the total number of dimensions (12) exceeds the set standard number of dimensions (10), the calculated dimensional complexity parameter is 0.2. Multiplying the initial adjustment index by the correction ratio yields a calculated index of 0.3. This scheme can adapt to fluctuations caused by material phase transitions, reduces the possibility of data rejection errors, and mitigates the risk of computational divergence during high-dimensional data aggregation through complexity compensation.

[0034] In S4, a control volume energy conservation limiting model is constructed to obtain the target adjustment index, ensuring that the data translation operation conforms to the thermodynamic boundary of the physical entity. Subsequently, the correction amount is obtained and the target fusion data is generated. For the aforementioned local casting weighing approximately 1.35 kg and with a volume of 0.0005 cubic meters, the maximum latent heat limit energy that can be released during the solidification stage is calculated to be 525150 J. Simultaneously, the mold control volume represented by the temperature acquisition dimension is statistically analyzed, and it is calculated that a full correction of the current 4K deviation will generate a total virtual energy change of 14352 J within the twin, thus yielding an energy tolerance ratio of 36.5. The minimum value between the calculated index 0.3 and this tolerance ratio is selected as the target adjustment index 0.3. Based on this, the correction amount for the temperature dimension is calculated to be 1.2 K, and the measured value is translated to obtain the corrected fusion value of 825.95 K. When processing pressure data, only the 1.2 MPa slow-varying baseline offset caused by temperature drift is offset, while the positive and negative 0.5 MPa high-frequency oscillation envelope reflecting the turbulent characteristics of the molten metal is retained. This step limits the correction to the latent heat range, reduces the possibility of equation divergence due to fictitious energy, and maintains the feature integrity of microscopic dynamic details when inputting into a digital twin.

[0035] In summary, the entire data processing method for aluminum alloy die casting based on digital twins firstly employs a physical transmission delay misalignment compensation mechanism based on the thermal transfer properties of the medium and the pressure wave velocity. This alleviates the spatiotemporal lag differences in the propagation of multimodal signals within the mold, providing synchronized data for physical section alignment in subsequent model calculations. Furthermore, by combining the sensor's response time constant to extract the hysteresis compensation rate, the response hysteresis caused by hardware packaging is compensated, and the absolute numerical difference including background thermal noise is converted into a cross-dimensional deviation rate, making the offset assessment in each dimension closely approximate the transient changes in the melt. Moreover, by utilizing the alloy's latent heat of fusion and real-time solidity to construct a dynamic threshold, it can adapt to the temperature plateau period caused by the release of latent heat in the mushy region, reducing the misjudgment of physical fluctuations caused by alloy phase transformation as sensor measurement drift. Additionally, the dimensional complexity parameter reduces the risk of computational divergence that may occur during high-dimensional data aggregation. Furthermore, an energy conservation limiting model for the control volume is constructed to restrict the data correction amount to the maximum latent heat range of the casting, reducing the possibility of thermodynamic logic conflicts caused by energy errors introduced into the digital twin during data translation operations. At the same time, when compensating for measured values ​​during translation, the high-frequency oscillation characteristics caused by fluid turbulence are preserved, ensuring that the target fusion data input to the digital twin maintains the integrity of microscopic dynamic details while satisfying the macroscopic energy conservation boundary conditions.

[0036] like Figure 2 As shown, in one specific embodiment, S1 includes: S11, acquiring time-series data of multiple acquisition dimensions of the process entity. In the aluminum alloy die-casting process, the acquisition dimension time-series data refers to the multi-source data stream acquired in real time from the sensor cluster at the production site, such as the pressure fluctuation sequence of the injection mechanism, the temperature response time sequence of the mold inner wall, and the stress signal on the cavity surface. Simultaneously, the total number of acquisition dimensions is counted to define the data scale of the digital twin mapping.

[0037] S12. Execution Time Dislocation Compensation. Based on the spatial coordinate parameters of the acquisition point within the mold, combined with the pressure transmission attribute representing the pressure transmission wave velocity and the medium heat transfer attribute representing the medium's heat conduction rate, the physical transmission delay difference between signals from different acquisition dimensions and the observation point is calculated. Here, time dislocation compensation refers to using an algorithm to retrospectively calibrate the time lag caused by differences in physical propagation speed. Based on the physical transmission delay difference, the time-series data of the acquisition dimensions are shifted along the time axis to generate physically synchronized data representing the state of the physical space at the same instant.

[0038] S13. Establish independent baseline values. Based on the set values ​​of the process card or the statistical characteristics of historical stable periods, bind independent baseline values ​​for each dimension.

[0039] In this embodiment, it should be noted that in S11, time-series data of multiple acquisition dimensions of the process entity are acquired, and the total number of acquisition dimensions is counted. Taking the integrated die-casting production scenario of a large automotive aluminum alloy engine block as an example, a total of 12 acquisition dimensions are deployed in key parts of the mold inner wall and during injection. These include 4 high-sampling-frequency pressure sensors located near the ingate and 8 temperature sensors evenly distributed 5mm on the surface of the mold cavity. Based on this, the total number of acquisition dimensions is counted as 12. The technical means of this step is to sort out the deployment of sensing hardware on the die-casting site, obtain the spatial coordinate parameters corresponding to each acquisition dimension, and clarify the exact mapping position of each sensor in three-dimensional physical space. This solves the problem of unclear spatial attributes of multi-source sensor data, provides a low-level spatial coordinate reference for subsequent transmission delay calculation based on physical distance, and enables the acquired process data to form a close physical binding relationship with the actual structure of the mold.

[0040] In S12, time misalignment compensation is performed on the acquired dimensional time-series data based on the thermal and pressure transfer properties of the medium to obtain physically synchronized data and corresponding independent reference values. The pressure wave propagation velocity in molten aluminum alloy is approximately 4500 m / s, while the thermal diffusivity of mold steel is approximately 0.0000075 square meters per second. For a temperature sensor buried at a depth of 5 mm, the thermal conduction delay is approximately 3.33 s, while the pressure wave delay at the same location is only approximately 0.0000011 s. Using these spatial coordinate parameters combined with the medium properties, the physical transmission delay difference is calculated, and the pressure data stream at time t and the temperature data stream at time t minus 3.33 s are shifted and compensated on the time axis. This misalignment compensation logic solves the problem of physical spatial misalignment of data with the same timestamp caused by the transmission of different physical quantities across media, effectively alleviating the differences in multimodal signal propagation lag and providing physically aligned synchronized data for subsequent twin model calculations.

[0041] In S13, after completing the physical alignment of space and time, corresponding independent reference values ​​are established for each acquisition dimension. Continuing with the aforementioned die-casting process of the engine cylinder block, based on the historical process card of the ADC12 aluminum alloy product or the statistical characteristics of the current stable production cycle, reference data for each sensor dimension are extracted. For example, the independent reference value for the peak injection pressure during the filling stage is set to 32.0 MPa, and the independent reference value for the local steady-state cooling temperature of the mold is set to 823.15 K. This technique changes the approach of using a single fixed static reference for the entire process, solving the problem of inconsistent state evaluation references caused by different heat dissipation conditions in different mold areas of complex die-cast parts due to the different heat dissipation conditions of thick and thin-walled parts. This operation of binding an independent reference value to each physically synchronized data stream allows subsequent deviation calculations to be based on the initial thermodynamic or kinetic state of each local area.

[0042] like Figure 3 As shown, in one specific embodiment, S2 includes: S21, obtaining the initial deviation. The calculated difference is obtained by subtracting the independent reference value from the current measured value of the physical synchronization data. For the acquisition dimension characterizing temperature, it is determined whether the calculated difference is lower than the set background thermal noise boundary. If it is lower than this boundary, the calculated difference is calibrated to an effective offset based on the industrial thermal inertia characteristics and used as the initial deviation; if it is not lower than this boundary, the calculated difference is directly used as the initial deviation. The initial deviation here reflects the absolute order of magnitude of the deviation of each acquisition dimension from the ideal process state.

[0043] S22. Conversion Deviation Rate. To compensate for the signal response lag caused by sensor packaging, a physical response constant is introduced for dynamic adjustment. The specific calculation logic is as follows:

[0044]

[0045] in, Let be the deviation rate of the i-th acquisition dimension at time t; Let be the initial deviation of the i-th acquisition dimension at time t; This represents the independent baseline value corresponding to the i-th data collection dimension; Let be the physical response constant of the sensor corresponding to the i-th acquisition dimension; Let be the rate of change of the current measured value of the i-th acquisition dimension with respect to time t.

[0046] Under this logic, the initial deviation is divided by an independent reference value to obtain the basic relative error, and then a hysteresis compensation rate generated from the physical response constant (reflecting the sensor's dynamic hysteresis) is added to obtain the deviation rate after compensating for thermal inertia hysteresis. The hysteresis compensation rate is used to quantify the sensor's response inertia during transient changes.

[0047] In this embodiment, it should be noted that in S21, the initial deviation is obtained by subtracting the independent reference value from the current measured value of the physical synchronization data. If the current measured value of a certain temperature acquisition dimension is 827.15K, the calculated difference between it and the independent reference value of 823.15K is 4K, calculated by algebraic subtraction. It is then determined whether this calculated difference is lower than a set background thermal noise boundary, such as 0.1K. Since 4K is higher than this thermal noise boundary, the calculated difference is directly used as the initial deviation.

[0048] If the calculated difference is extremely small and falls within the inherent background thermal noise range of the sensor, it will be calibrated to an effective offset based on the industrial thermal inertia characteristics before being used as the initial deviation. The method for determining the set background thermal noise boundary is as follows: with the die-casting machine unloaded and the mold in a constant-temperature closed state, a background temperature sample sequence is continuously collected within a set time period (e.g., 10 minutes) through the corresponding temperature acquisition channel. The standard deviation of the temperature fluctuation of this sample sequence is calculated, and this standard deviation is multiplied by a preset confidence threshold coefficient (e.g., 3 times) to obtain the upper limit of the boundary covering most random hardware noise. For example, continuously acquiring 1000 unloaded mold temperature data points, the calculated standard deviation is 0.033K. Multiplying this by 3 times the confidence threshold coefficient, the set background thermal noise boundary for this acquisition channel is 0.1K. This logic, by filtering out minute thermal noise without physical guidance significance, solves the problem that conventional algebraic subtraction operations easily mistake sensor noise floor for process fluctuations, ensuring that the initial deviation used as the basis for subsequent evaluation reflects the physical and thermodynamic offset of the mold.

[0049] In S22, the initial deviation is converted into a deviation rate by incorporating the physical response constant; specifically, this is achieved through the calculation of the expression... The core of this calculation process, which converts the initial deviation into a deviation rate, lies in removing the time-domain hysteresis effect caused by the hardware packaging of sensors in industrial settings.

[0050] The first term of the expression is the initial deviation at the current time. Divide by the corresponding independent benchmark value This division operation aims to eliminate the dimensional differences caused by different physical properties and different measurement point magnitudes, transforming them into a standardized basic relative error. For example, for a certain mold temperature acquisition dimension, the initial deviation obtained by subtracting the reference value from the current measured value is 4K, and the independent reference value is 823.15K. The basic relative error obtained by dividing the two is approximately 0.004858.

[0051] However, under the high-speed filling and rapid heating / cooling conditions of aluminum alloy die casting, the protective sheath surrounding the industrial K-type thermocouple induces thermal inertia, causing the sensor's output electrical signal to lag behind the actual physical temperature changes inside the mold. If only static numerical differences are relied upon for state assessment, the digital twin will receive boundary conditions containing time lags. Therefore, a second term is introduced into the expression as feedforward compensation, where... Defined as the dynamic hysteresis compensation rate, based on the underlying physical logic, its calculation results are strictly limited to a technical unit of [missing information]. It is used to reflect the transient rate of change of the measured signal on the time axis.

[0052] In the specific calculation, the rate of change of the current measured value of the temperature measurement node with respect to time, 20 K / s, is extracted, divided by the absolute value of the reference value to maintain the correct direction sign and dimensions, and then multiplied by the inherent physical response time constant of the sensor. That is, 0.05s. The physical response time constant of this sensor is obtained based on step response calibration tests performed on the hardware sensing package: Sensors from the same batch were subjected to a step switch between two physical fluid media with a constant temperature difference (e.g., instantaneous transfer from a 0°C ice-water mixture to 100°C boiling water). The absolute time taken for the sensor output signal to reach 63.2% of the total temperature difference change was recorded, and the average time from multiple tests was taken as the physical response time constant. For example, for this type K thermocouple, 10 ice-boiling water step switch tests were performed, and the average time to reach a steady-state output of 63.2% was measured to be 0.05s. This 0.05s is then used as the inherent physical response constant in the feedforward compensation calculation. The physical meaning of this product operation is to use the current jump slope of the signal to predict the amount of hidden deviation that has not yet been fully perceived by the sensor within the response delay time window, calculating the value of the hysteresis attenuation compensation term to be approximately 0.001214. Finally, the basic relative error is summed with the hysteresis attenuation compensation term to obtain a deviation rate of approximately 0.006072 after compensating for thermal inertia hysteresis, or 0.6072%.

[0053] This computational logic, which combines static proportional error with dynamic first derivative, solves the technical problem that conventional algebraic evaluation is difficult to handle transient large temperature gradient conditions. It enables the calculated deviation rate to reflect the dynamic change trend of the molten metal in advance, preventing subsequent algorithms from making data lag misjudgments when facing high-speed filling. It provides a more realistic underlying evaluation index for overall data collaboration.

[0054] like Figure 4 As shown, in one specific embodiment, S3 includes: S31, generating a dynamic threshold. Aluminum alloys release latent heat in the phase transformation slurry region, causing nonlinear fluctuations in data characteristics. The tolerance boundary is dynamically adjusted using a real-time solids fraction parameter. The real-time solids fraction parameter refers to the volume proportion of the alloy that has transformed into a solid phase during solidification. The expression for generating the dynamic threshold can be as follows:

[0055]

[0056] in, The dynamic threshold at time t; This represents the basic tolerance value in the liquid phase state; These are the latent heat of fusion parameters for aluminum alloys; This represents the average specific heat capacity parameter of the aluminum alloy in the mushy region. These are the liquidus temperature parameters for aluminum alloys. These are solidus temperature parameters for aluminum alloys. This is the temperature of the alloy melt measured at the current moment.

[0057] In the formula, the phase transition relaxation coefficient is first obtained based on the latent heat of melting parameter, the average specific heat capacity parameter, and the liquid-solid phase temperature span; then, the phase transition relaxation coefficient is dynamically scaled based on the real-time solid fraction parameter and the basic tolerance value is superimposed to finally generate the dynamic threshold.

[0058] S32. Calculation and Index Correction. The number of data collection dimensions whose absolute value of the statistical deviation rate exceeds the dynamic threshold is divided by the total number to obtain the initial adjustment index. Subsequently, the standard number of dimensions is obtained. If the total number exceeds the standard number of dimensions, the difference between the two is divided by the standard number of dimensions to obtain the dimension complexity parameter; if it does not exceed the standard number, the dimension complexity parameter is set to zero. The dimension complexity parameter is used to assess the computational convergence risk caused by excessively high data dimensionality. The dimension complexity parameter is added to a constant to obtain the correction ratio. The initial adjustment index is multiplied by the correction ratio to generate the calculated index.

[0059] In this embodiment, it should be noted that in S31, the real-time solid fraction parameter is obtained to generate a dynamic threshold. Specifically, this is achieved by calculating an expression. A dynamic threshold is generated, and this calculation process aims to establish an adaptive tolerance evaluation system that matches the phase transformation thermodynamic process of aluminum alloy metallurgy.

[0060] During the solidification process of aluminum alloy ADC12, the temperature rises from the liquidus. Descending to the solidus The region is called the mushy region, within which the liquid metal gradually grows a dendritic network and releases a large amount of latent heat of fusion. The large-scale release of this latent heat of phase change will offset the cooling heat transfer of the external mold, resulting in a physical plateau in the temperature measurement curve on the time axis where the temperature decrease tends to be gradual. If the fixed static tolerance threshold of the liquid phase region is continued to be used during this stage, the anomaly detection algorithm may easily misjudge this thermal stagnation caused by the material phase change as temperature drift of the temperature sensor or failure of the acquisition channel.

[0061] To solve this problem, the second half of the expression within the parentheses... A linear approximate real-time solidity parameter based on temperature gradient was constructed. Given a set liquidus temperature of 853.15 K and a solidus temperature of 788.15 K, the measured temperature of the alloy melt at the current moment... At 833.15 K, the calculated real-time solids fraction is approximately 0.307, indicating that about 30.7% of the alloy volume has solidified. The first part of the expression (within parentheses) extracts the ratio of the latent heat of fusion of the aluminum alloy (389000 J / kg), the average specific heat capacity of the viscous region (960 J / (kg·K), and the liquid-solid phase temperature span (65 K)). This ratio is approximately 6.23, which physically represents the dimensionless multiple of latent heat to sensible heat; here, it is defined as the phase transformation relaxation coefficient. Multiplying the real-time solids fraction parameter by the phase transformation relaxation coefficient and adding it to a constant 1 yields a tolerance relaxation factor of approximately 2.91. The preset liquid phase state baseline tolerance value is then applied. That is, 0.03, multiplied by this relaxation factor, finally generates a dynamic threshold of 0.0873, or 8.73%, corresponding to the current microsecond-level state. The preset method for determining the basic tolerance value for the liquid phase state is as follows: Monitoring data from historical normal production batches when the alloy was completely in the liquid phase range is extracted. The actual relative deviation of each dimension's time-series data compared to its corresponding benchmark value is calculated. After removing outliers, the maximum envelope value of the historical relative deviation within the 95% confidence interval is extracted as the basic tolerance value. For example, statistical analysis of the temperature relative deviation during the pure liquid filling stage of the past 30 cycles reveals that 95% of the effective deviation data are distributed within 0.03 (i.e., 3%). Therefore, 0.03 is set as the allowable reasonable fluctuation benchmark in the liquid phase state, i.e., the basic tolerance value, and then used as the starting point for subsequent phase transformation tolerance relaxation calculations.

[0062] This computational logic establishes a correlation between static threshold data and the thermodynamic constitutive state of the material. By monitoring the phase transformation progress of the melt in the mushy region in real time and dynamically amplifying the tolerance boundary proportionally, the data filtering algorithm can autonomously adapt to the nonlinear cooling and heat dissipation law of the material, preventing the situation where feature data containing real solidification phase transformation information is mistakenly rejected as abnormal noise.

[0063] In S32, preliminary adjustment and calculation indicators are calculated by combining the dimensionality complexity parameter. Following the aforementioned threshold calculation results, traversing these 12 acquisition dimensions, three acquisition dimensions with an absolute deviation rate exceeding the dynamic threshold of 8.73% are identified. Dividing these by the total number of dimensions (12) yields a preliminary adjustment indicator of 0.25. To assess the risk of high-dimensional data solution, a standard dimension number of 10 is first determined. The method for setting the standard dimension number is as follows: it is determined based on the Jacobian matrix condition number evaluation test of the partial differential equation solver used in the digital twin; during the offline debugging phase of the twin model, the number of sensor dimensions for synchronous input boundary conditions is gradually increased while monitoring the coupled iteration process. The upper limit of the previous stable dimension when the iteration step count increases exponentially or the residual begins to diverge is taken as the standard dimension number. For example, offline testing revealed that when the input dimension reached 11, the condition number of the solution matrix abruptly caused computational instability, while 50 consecutive tests with 10 dimensions showed stable convergence. Therefore, 10 is set as the matching standard dimension number. Subsequently, since the total number of dimensions (12) exceeds the standard number of dimensions, the difference between the two (2) is calculated and divided by 10 to obtain the dimension complexity parameter 0.2. This dimension complexity parameter 0.2 is added to a constant to obtain a correction ratio of 1.2. The initial adjustment index 0.25 is then multiplied by this correction ratio to generate a computational index of 0.3. This adjustment logic, by quantifying the breadth of the deviation data and the scale of the feature dimensions, alleviates the potential iterative computational instability that may occur in multiphysics nonlinear coupling equations when facing non-cooperative inputs.

[0064] like Figure 5 As shown, in one specific implementation, S4 includes: S41, performing thermodynamic limiting. To ensure that the data correction amount does not violate the energy state of the physical entity, a control volume energy conservation limiting model is constructed. The control volume energy conservation limiting model refers to a set of physical constraint operators based on the first law of thermodynamics, used to limit fictitious energy changes. The expression for obtaining the target adjustment index can be as follows:

[0065]

[0066] in, Adjust the indicators to the target; For calculation indicators; Density of liquid alloy; For the volume of the casting; Here, represents the latent heat of fusion of the aluminum alloy; M represents the total number of dimensions involved in the thermodynamic calculation. The mold density parameter corresponding to the j-th acquisition dimension; The specific heat capacity parameter of the mold corresponding to the j-th acquisition dimension; The absolute value of the initial deviation of the j-th acquisition dimension; Let represent the volume parameter of the local control volume in the j-th acquisition dimension.

[0067] This logic calculates the ratio between the maximum latent heat limit energy released by the casting and the total virtual energy change required to correct all deviations, obtains the energy tolerance ratio, and selects the minimum value between the calculated index and this ratio as the target adjustment index.

[0068] S42. Generate target fusion data. Multiply the initial deviation by the target adjustment index to obtain the correction. During the translation compensation process, extract the high-frequency oscillation characteristics caused by fluid turbulence in the current measured values, and only perform offset translation on the slowly varying baseline caused by response temperature drift, retaining and outputting the target fusion data carrying the high-frequency oscillation characteristics. Finally, input the target fusion data into the digital twin for state calculation.

[0069] In this embodiment, it should be noted that in S41, the expression is used. Construct a control volume energy conservation and limit model.

[0070] In multi-source data collaborative processing, preliminary calculation indicators based on statistics and dimensions are used. The given correction ratio might be 0.3, but a digital twin is not only a container for data, but also a solver for multiphysics partial differential equations. Mathematically translating boundary conditions such as temperature within the model is equivalent to introducing or losing heat arbitrarily in physical space. If the translation magnitude is too large, the sum of the resulting virtual energy changes will exceed the actual energy reserves of the physical entity, leading to iterative divergence of the finite volume solution matrix.

[0071] To avoid this risk, the numerator of the expression The maximum latent heat limit energy that can be released during the solidification stage of the casting was calculated. Substituting the liquid alloy density of 2700 kg / m³, the casting volume of 0.0005 m³, and the latent heat of fusion of 389000 J / kg, the product was 525150 J. The denominator of the expression is an integral summation over all M temperature monitoring dimensions. Specifically, the mold density of the j-th dimension (7800 kg / m³), the mold specific heat capacity (460 J / (kg·K), the absolute value of the initial deviation (4K), and the volume parameter representing the local control volume are multiplied together. The method for determining the volume parameter representing the local control volume is as follows: import the 3D CAD geometric model of the die-casting mold and the casting; based on the precise spatial coordinate system of each temperature sensor inside the mold, use finite element mesh generation technology or Thiessen polygon spatial partitioning algorithm to discretize the mold entity into multiple non-overlapping control subdomains; and directly extract the actual physical volume of the subdomain under the jurisdiction of each sensor node through geometric analysis, and use it as the volume parameter of the local control volume for that dimension. For example, after meshing using a mesh generation tool, the geometric volume of the local mold area controlled by temperature measurement point 1 was measured and extracted to be 0.0002 cubic meters. Based on this, the total virtual energy change generated within the twin if the current temperature deviations at all locations are fully corrected is calculated to be 14352 J. Dividing the numerator by the denominator yields an energy tolerance ratio of approximately 36.5, indicating that the current deviation is within a safe thermodynamic buffering range.

[0072] The outer min operator is responsible for arbitrating and trunculating between the numerical correction requirement and the physical conservation law. It compares the calculated index of 0.3 with the energy tolerance ratio of 36.5 and selects the minimum value of 0.3 as the final target adjustment index.

[0073] This computational mechanism, which converts temperature differences into Joule energy using geometric and physical property parameters such as specific heat capacity and volume, imposes physical conservation boundary constraints on the coordination of purely mathematical data. This prevents abnormal data shifts triggered by the failure of individual sensors, ensuring that the boundary data input to the digital twin is always constrained within the macroscopic energy conservation framework, and maintaining the numerical stability of subsequent thermodynamic coupling calculations.

[0074] In step S42, the correction amount is obtained and the current measured value is shifted and compensated to generate the target fused data input digital twin for state calculation. Based on the aforementioned target adjustment index of 0.3, it is multiplied by the initial deviation of 4K in the temperature dimension to obtain a temperature correction amount of 1.2K. After subtraction and shift compensation of the current measured value of 827.15K, the corrected fused value of 825.95K is obtained. For the pressure data processing, a sliding window high-pass filter with a cutoff frequency of the set fluid turbulence cutoff frequency is used to filter the pressure time series data. The method for determining the fluid turbulence cutoff frequency is as follows: Using Fast Fourier Transform (FFT), frequency domain analysis is performed on the high-frequency pressure time-series raw data of historical standard die castings collected during the filling stage. This separates the low-frequency build-up response band determined by the large-scale mold cavity structure and the microscopic high-frequency oscillation band excited by turbulence and splashing. The frequencies corresponding to the spectral troughs of the energy distribution in these two adjacent bands are selected as the fluid turbulence cutoff frequencies. For example, by performing an FFT on the pressure fluctuations at the ingate of 100 cycles, it was found that there are high-frequency energy islands caused by turbulence above 150Hz. The boundary between these islands and the trough of the system's low-frequency pressure build-up energy region is stable at 120Hz. Therefore, 120Hz is set as the fluid turbulence cutoff frequency. During filtering and subsequent translation compensation, the high-frequency oscillation characteristics of ±0.5MPa caused by fluid turbulence in the current measured value are specifically extracted, and only the 1.2MPa slow-varying baseline offset caused by response temperature drift is offset. This processing method alleviates the problem that conventional filtering can easily smooth out the details of micro-process fluctuations, ensuring that the target fusion data input to the digital twin meets the energy conservation requirements, while maintaining the integrity of reflecting the dynamic details of the metal melt and improving the reliability of the model in reproducing the process.

[0075] A data processing system for aluminum alloy die-casting process based on digital twins is also provided. The system includes:

[0076] The data compensation module is used to acquire time-series data of multiple acquisition dimensions of the process entity and count the total number of acquisition dimensions; it performs time misalignment compensation on the time-series data of the acquisition dimensions based on the heat transfer and pressure transfer properties of the medium, and acquires physical synchronization data and corresponding independent reference values.

[0077] The deviation assessment module is used to subtract the independent benchmark value from the current measured value of the physical synchronization data to obtain the initial deviation, and to convert the initial deviation into a deviation rate by combining the physical response constant.

[0078] The dynamic threshold module is used to obtain the real-time solid fraction parameter of the process entity and generate a dynamic threshold based on the real-time solid fraction parameter and the physical properties of the alloy.

[0079] The initial indicator calculation module is used to count the number of data collection dimensions whose absolute value of the deviation rate exceeds the dynamic threshold, divide it by the total number, and obtain the initial adjustment indicator; the initial adjustment indicator is corrected by the dimension complexity parameter to obtain the calculated indicator.

[0080] The energy limiting module is used to construct an energy conservation limiting model for the control volume, and to thermodynamically limit the calculated indicators based on the alloy specific heat capacity and latent heat parameters to obtain the target adjustment indicators.

[0081] The twin fusion module is used to multiply the initial deviation by the target adjustment index to obtain the correction amount, perform translation compensation on the current measured value based on the correction amount to obtain the target fusion data, and input the target fusion data into the digital twin for state calculation.

[0082] In one specific embodiment, the dynamic threshold module is further used to: obtain the basic tolerance value in the liquid phase state, and obtain the phase transformation relaxation coefficient based on the latent heat of melting parameter of the aluminum alloy, the average specific heat capacity parameter of the mushy region, and the temperature span difference between the liquidus and solidus lines; obtain the currently measured alloy melt temperature, and calculate the real-time solid fraction parameter based on the relative proportional difference between the alloy melt temperature, solidus temperature parameter, and liquidus temperature parameter; dynamically scale the phase transformation relaxation coefficient using the real-time solid fraction parameter, and add the scaling result to the basic tolerance value to generate a dynamic threshold.

[0083] In one specific implementation, the energy limiting module is also used to: calculate the maximum latent heat limit energy that can be released during the solidification stage of the casting based on the product of the liquid alloy density, the casting volume, and the latent heat parameter; multiply and sum the mold density parameter, the mold specific heat capacity parameter, the volume parameter representing the local control volume, and the absolute value of the initial deviation to obtain the total virtual energy change when the current deviation is fully corrected; divide the maximum latent heat limit energy by the total virtual energy change to obtain the energy tolerance ratio; and select the minimum value between the calculated index and the energy tolerance ratio as the target adjustment index.

[0084] To enable those skilled in the art to fully understand and implement the technical solutions described in this specification, the following section, using a control scenario containing specific data, provides a detailed deduction and data analysis of the entire implementation principle of a data processing method and system for aluminum alloy die casting based on digital twins.

[0085] In an integrated die-casting production scenario for a large automotive aluminum alloy engine block, to precisely monitor the filling and solidification processes, we deployed 12 data acquisition dimensions at key locations on the mold inner wall and in the injection system. According to method S11, this includes four high-sampling-frequency pressure sensors located near the ingate (for real-time acquisition of cavity pressure). And eight temperature sensors (collecting mold temperature) evenly distributed 5mm from the surface of the mold cavity. For this specific process entity, we first established the spatial topology matrix of the mold sensors, with the geometric coordinates of the pressure and temperature sensors strictly mapped onto a three-dimensional coordinate system. Before production began, the density of the liquid metal was set based on the material properties of the ADC12 aluminum alloy. for The latent heat of fusion of materials for Liquidus temperature for (Right now solidus temperature for (Right now Meanwhile, the specific heat capacity of H13 mold steel was measured. for Mold material density for It should be noted that the methods for determining the values ​​of various physical and thermodynamic properties of alloy materials and mold steel in this scheme are as follows: For specific grades of materials used in actual production batches (such as ADC12 aluminum alloy and H13 mold steel), the corresponding authoritative recommended test values ​​are directly extracted from the standard materials engineering database before production begins; or, to improve the accuracy of twin mapping, physical samples of actual production materials are taken in a laboratory environment, and programmed temperature rise tests are performed using thermal analysis instruments such as differential scanning calorimeter (DSC) to measure their latent heat of phase change and specific heat capacity at each stage. After multiple independent tests, the average value is obtained and used as the preset physical property parameters for system state calculation.

[0086] In step S12, the system performs timestamp misalignment compensation based on physical transmission delay. This is because the pressure wave propagation speed in the molten aluminum alloy is as high as... Heat conduction inside the mold is limited by the thermal diffusivity (approximately...) For a temperature sensor buried at a depth of 5mm, the physical delay time for heat conduction to the sensing point is approximately 3.33s, while the pressure wave delay at the same location is only [missing information]. At the algorithm layer, we physically synchronize and align the pressure data stream at time t with the temperature data stream at time t-3.33s to generate physically synchronized data reflecting the state of the physical space at the same instant. Then, in S13, we extract independent reference values ​​for this moment, such as the reference value of the peak injection pressure. The benchmark value for the local steady-state temperature of the mold is 32.0 MPa. for (Right now ).

[0087] In steps S21 to S22, the system calculates the numerical difference and compensates for the sensor's thermal hysteresis. If the measured temperature at the current moment is... Then the initial deviation Simultaneously, during the transient filling phase, the sensor detected the rate of temperature change. for Given the physical response constant of a type K thermocouple. The value is 0.05s. Substitute this value into the deviation rate formula. The deviation rate after compensating for thermal inertia hysteresis was calculated. ,Right now This calculation logic effectively extracts the hysteresis of the sensor during the dynamic heating process, making the deviation rate closer to the actual physical state of the melt.

[0088] Then steps S31 to S32 are executed. The system identifies the current local temperature of the casting as... (Right now This refers to the pasty region located between the liquidus (853.15 K) and solidus (788.15 K). Based on the dynamic threshold model, the system obtains the basic tolerance value. The real-time solid fraction parameter at this time is calculated as follows: The ratio of latent heat to sensible heat in the phase transition relaxation coefficient term is known to be... Substituting into the formula, the dynamic threshold is calculated. ,Right now At this point, the system iterates through 12 data collection dimensions and finds that the absolute value of the deviation rate in 3 dimensions exceeds the dynamic threshold. Therefore, the second quantity is 3, the first quantity is 12, and the initial adjustment index is... Since the access dimension 12 exceeds the standard dimension count of 10 for digital twin systems, the dimensionality complexity parameter is... The final calculated index is .

[0089] In S41, the system constructs a control volume energy conservation limiting model for thermodynamic verification. The total volume of the engine block casting... for (Weighing approximately 1.35 kg), the maximum latent heat limit energy that can be released during the solidification stage of the casting is calculated to be approximately 1.35 kg ⋅ 389,000 J / kg = 525,150 J. Simultaneously, by statistically analyzing the total volume of the mold control body represented by the eight temperature acquisition dimensions, it is calculated that if the current 4K deviation is fully corrected, the total energy change virtually generated within the digital twin is approximately 14,352 J. The energy tolerance ratio is 525,150 / 14,352 ≈ 36.5. Due to the calculation parameters... Much less than this energy limit ratio, therefore the target adjustment index The value is confirmed to be 0.3. Based on this ratio, the correction for this temperature dimension is calculated to be 4K·0.3=1.2K.

[0090] Finally, in S42, the system generates the target fused data. For the temperature dimension, the corrected fused value is 827.15K - 1.2K = 825.95K. When processing the pressure dimension data, the system identifies that the measured pressure value contains high-frequency oscillation waves (transient spikes with an amplitude of ±0.5MPa) generated by the high-speed entry of molten metal into the cavity. During translation compensation, the algorithm only deducts the 1.2MPa offset of the baseline caused by sensor temperature drift, while fully preserving the high-frequency oscillations that reflect the physical characteristics of fluid turbulence. Finally, the target fused data stream, composed of 12 dimensions, is concurrently input into the digital twin for flow field calculation. Since the input data has been physically synchronized, phase-change compensated, and energy-conserving limited, the digital twin system can accurately map the possible locations of air entrapment and shrinkage porosity inside the casting under self-consistent boundary conditions, providing reliable data support for real-time optimization of the die-casting process.

[0091] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solution of the present invention, and these simple modifications all fall within the protection scope of the present invention.

[0092] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present invention will not describe the various possible combinations separately.

[0093] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed by the present invention.

[0094] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A data processing method for aluminum alloy die-casting process based on digital twins, characterized in that, The methods include: Acquire time-series data from multiple acquisition dimensions of the process entity and count the total number of acquisition dimensions; Based on the heat transfer and pressure transfer properties of the medium, time misalignment compensation is performed on the time-series data of the acquisition dimension to obtain physical synchronization data and corresponding independent reference values. The initial deviation is obtained by subtracting the independent benchmark value from the current measured value of the physical synchronization data, and then converting the initial deviation into a deviation rate by combining the physical response constant. Obtain the real-time solid fraction parameter of the process entity, and generate a dynamic threshold based on the real-time solid fraction parameter and the physical properties of the alloy; The number of data collection dimensions whose absolute value of the statistical deviation rate exceeds the dynamic threshold is divided by the total number to obtain the preliminary adjustment index. The initial adjustment index is corrected by combining the dimensionality complexity parameter to obtain the computational index; A control volume energy conservation limiting model is constructed, and the calculated index is thermodynamically limited based on the alloy specific heat capacity and latent heat parameters to obtain the target adjustment index. The initial deviation is multiplied by the target adjustment index to obtain the correction amount. Based on the correction amount, the current measured value is shifted and compensated to obtain the target fusion data. The target fusion data is then input into the digital twin for state calculation.

2. The data processing method for aluminum alloy die-casting process based on digital twins according to claim 1, characterized in that, The process of performing time misalignment compensation on the acquired dimensional time-series data based on the thermal and pressure transfer properties of the medium to obtain physically synchronized data includes: Obtain the spatial coordinate parameters corresponding to each collection dimension; Based on spatial coordinate parameters, combined with the pressure transmission property representing the pressure transmission wave velocity and the medium heat transfer property representing the thermal diffusivity, the physical transmission delay difference of time series data from different acquisition dimensions reaching the same observation point is calculated respectively. Based on the difference in physical transmission delay, the time-series data of the acquired dimensions are shifted and compensated on the time axis to generate physically synchronized data with physical cross-section alignment.

3. The data processing method for aluminum alloy die-casting process based on digital twins according to claim 1, characterized in that, The conversion of initial deviation into deviation rate by incorporating the physical response constant includes: Based on the rate of change of the current measured value of the time series data of each acquisition dimension with respect to time, and combined with the response time constant of the corresponding acquisition dimension, the hysteresis compensation rate used to quantify the response lag characteristics is obtained. The basic relative error is obtained by dividing the initial deviation by the independent reference value. Adding the basic relative error to the hysteresis compensation rate yields the deviation rate after compensating for thermal inertia hysteresis.

4. The data processing method for aluminum alloy die-casting process based on digital twins according to claim 1, characterized in that, The generation of dynamic thresholds based on real-time solidity parameters and alloy physical properties includes: Obtain the basic tolerance value in the liquid phase state, and obtain the phase transformation relaxation coefficient based on the latent heat of melting parameters of aluminum alloy, the average specific heat capacity parameters of the paste region, and the temperature span difference between the liquidus and solidus lines. Obtain the currently measured alloy melt temperature, and calculate the real-time solid fraction parameter based on the relative proportional difference between the alloy melt temperature, solidus temperature parameter, and liquidus temperature parameter. The phase transition relaxation coefficient is dynamically scaled using the real-time solid fraction parameter, and the scaling result is added to the basic tolerance value to generate a dynamic threshold.

5. The data processing method for aluminum alloy die-casting process based on digital twin according to claim 1, characterized in that, The preliminary adjustment index is corrected by combining the dimensionality complexity parameter to obtain the computational index, including: Obtain the number of standard dimensions that match the digital twin solution model; Determine if the total number exceeds the standard number of dimensions. If so, divide the difference between the total number and the standard number of dimensions by the standard number of dimensions to obtain the dimension complexity parameter; otherwise, set the dimension complexity parameter to zero. Add the dimensionality complexity parameter to a constant to obtain the correction ratio, and multiply the initial adjustment index by the correction ratio to generate the calculated index.

6. The data processing method for aluminum alloy die-casting process based on digital twin according to claim 1, characterized in that, The constructed control volume energy conservation limiting model, based on alloy specific heat capacity and latent heat parameters, thermodynamically limits the calculated indicators to obtain the target adjustment indicators, including: The maximum latent heat limit energy that the casting can release during the solidification stage is calculated based on the product of the liquid alloy density, the casting volume, and the latent heat parameter. Multiply and sum the mold density parameter, mold specific heat capacity parameter, volume parameter representing the local control volume, and the absolute value of the initial deviation to obtain the total virtual energy change when the current deviation is fully corrected. Divide the maximum latent heat limit energy by the sum of virtual energy changes to obtain the energy tolerance ratio; The minimum value between the calculated index and the energy tolerance ratio is selected as the target adjustment index.

7. The data processing method for aluminum alloy die-casting process based on digital twin according to claim 1, characterized in that, The step of subtracting the independent reference value from the current measured value of the physical synchronization data to obtain the initial deviation includes: The difference is calculated by subtracting the independent baseline value from the current measured value of the physical synchronization data. Determine whether the calculated difference in temperature characterization is lower than the set background thermal noise boundary. If so, calibrate the calculated difference as an effective offset based on the industrial thermal inertia characteristics and use it as the initial deviation; otherwise, directly use the calculated difference as the initial deviation. The target fused data is obtained by shifting and compensating the current measured values ​​based on the correction amount, including: Extract the high-frequency oscillation characteristics caused by fluid turbulence from the current measured values; In the process of shifting and compensating the current measured value based on the correction amount, only the baseline offset caused by the response temperature drift is offset and shifted, and the target fusion data carrying high-frequency oscillation characteristics is retained and output.

8. A data processing system for aluminum alloy die-casting process based on digital twin, characterized in that, include: The data compensation module is used to acquire time-series data of multiple acquisition dimensions of the process entity and count the total number of acquisition dimensions. Based on the heat transfer and pressure transfer properties of the medium, time misalignment compensation is performed on the time-series data of the acquisition dimension to obtain physical synchronization data and corresponding independent reference values. The deviation assessment module is used to subtract the independent benchmark value from the current measured value of the physical synchronization data to obtain the initial deviation, and to convert the initial deviation into a deviation rate by combining the physical response constant. The dynamic threshold module is used to obtain the real-time solid fraction parameter of the process entity and generate a dynamic threshold based on the real-time solid fraction parameter and the physical properties of the alloy. The initial indicator calculation module is used to count the number of data collection dimensions whose absolute value of the deviation rate exceeds the dynamic threshold, divide it by the total number, and obtain the initial adjustment indicator; the initial adjustment indicator is corrected by the dimension complexity parameter to obtain the calculated indicator. The energy limiting module is used to construct an energy conservation limiting model for the control volume, and to thermodynamically limit the calculated indicators based on the alloy specific heat capacity and latent heat parameters to obtain the target adjustment indicators. The twin fusion module is used to multiply the initial deviation by the target adjustment index to obtain the correction amount, perform translation compensation on the current measured value based on the correction amount to obtain the target fusion data, and input the target fusion data into the digital twin for state calculation.

9. The data processing system for aluminum alloy die-casting process based on digital twin as described in claim 8, characterized in that, The dynamic threshold module is also used for: Obtain the basic tolerance value in the liquid phase state, and obtain the phase transformation relaxation coefficient based on the latent heat of melting parameters of aluminum alloy, the average specific heat capacity parameters of the paste region, and the temperature span difference between the liquidus and solidus lines. Obtain the currently measured alloy melt temperature, and calculate the real-time solid fraction parameter based on the relative proportional difference between the alloy melt temperature, solidus temperature parameter, and liquidus temperature parameter. The phase transition relaxation coefficient is dynamically scaled using the real-time solid fraction parameter, and the scaling result is added to the basic tolerance value to generate a dynamic threshold.

10. The data processing system for aluminum alloy die-casting process based on digital twins according to claim 8, characterized in that, The energy limiting module is also used for: The maximum latent heat limit energy that the casting can release during the solidification stage is calculated based on the product of the liquid alloy density, the casting volume, and the latent heat parameter. Multiply and sum the mold density parameter, mold specific heat capacity parameter, volume parameter representing the local control volume, and the absolute value of the initial deviation to obtain the total virtual energy change when the current deviation is fully corrected. Divide the maximum latent heat limit energy by the sum of virtual energy changes to obtain the energy tolerance ratio; The minimum value between the calculated index and the energy tolerance ratio is selected as the target adjustment index.

Images