An intelligent decision system for aluminum alloy piston casting process parameters

Abstract

This invention relates to the field of aluminum alloy piston casting technology, specifically disclosing an intelligent decision-making system for aluminum alloy piston casting process parameters. The system acquires real-time composition data of the molten aluminum alloy to be cast, forming a unique compositional characteristic for the current furnace batch. This unique compositional characteristic is input into a physical field mapping framework to generate a dynamic thermophysical property parameter set. Based on the dynamic thermophysical property parameter set and the piston's geometric structure characteristics, a dynamic solidification sensitivity coefficient matrix is ​​constructed. Using the dynamic solidification sensitivity coefficient matrix as a constraint boundary, the optimal combination of process parameters is solved to simultaneously achieve preset thresholds for mold filling integrity and solidification density. The optimal combination of process parameters is output as the casting control benchmark, and the correlation of the physical field mapping framework is updated based on the quality characteristic data fed back after casting. This invention achieves adaptive decision-making of process parameters under compositional fluctuations, improving the stability of the casting process and the yield.

Description

Technical Field

[0001] This invention relates to the field of aluminum alloy piston casting technology, and more specifically to an intelligent decision-making system for aluminum alloy piston casting process parameters. Background Technology

[0002] In the casting production of aluminum alloy pistons, the setting of process parameters (such as pouring temperature, cooling intensity, and pressurization sequence) directly affects the filling integrity, internal density, and final yield of the casting. Currently, the industry generally adopts two main approaches to process parameter decision-making: one is the "trial and error method" based on the experience of process engineers and process manuals, which involves gradually approaching the qualified process window through repeated trials, defect detection, and manual parameter adjustment. This method is highly dependent on personnel experience, has a long cycle, and is difficult to guarantee batch-to-batch consistency. At the industrial control system level, it usually only executes fixed parameters through a programmable logic controller, lacking parameter self-adjustment capabilities. The other approach is the offline optimization method based on numerical simulation software, which involves establishing a physical model of the piston casting process through the finite element method or finite difference method, simulating and analyzing multiple sets of process parameters using orthogonal experimental design, selecting the theoretically optimal parameter combination, and then loading it into the industrial control system as static setpoints for production guidance.

[0003] The lack of an online sensing and dynamic response mechanism for fluctuations in the actual composition of aluminum alloy melt in existing technologies means that the physical property parameters on which process parameters are determined are always based on static standard alloy grades. This makes it impossible to achieve real-time adaptive adjustment of process parameters driven by composition, resulting in poor stability of the casting process and a continuous mismatch between process parameters and the actual state of the melt. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent decision-making system for aluminum alloy piston casting process parameters to solve the problems mentioned above.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] An intelligent decision-making system for aluminum alloy piston casting process parameters includes:

[0007] The composition sensing module acquires real-time composition data of the aluminum alloy melt to be cast. The real-time composition data is obtained directly before the melt is poured through online spectral analysis and serves as the exclusive composition characteristics of the melt for the current furnace batch.

[0008] The property reconstruction module inputs the specific composition features into a pre-built physical field mapping framework. The physical field mapping framework establishes the relationship between composition and thermal properties based on historical casting data, and generates a dynamic thermal property parameter set corresponding to the specific composition features. The dynamic thermal property parameter set includes liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature.

[0009] The sensitivity analysis module, based on the dynamic thermophysical property parameter set, combined with the geometric structural characteristics of the piston to be cast and the preset casting quality target, constructs a dynamic solidification sensitivity coefficient matrix by quantifying the attenuation of filling driving force and the sensitivity shift of feeding channel caused by the deviation of exclusive composition characteristics from the reference composition.

[0010] The parameter optimization module uses the dynamic solidification sensitivity coefficient matrix as the constraint boundary for multi-objective optimization. It solves for the optimal combination of process parameters, which is the casting temperature setting, cooling intensity distribution, and pressurization pressure sequence that simultaneously achieve the preset thresholds for filling integrity and solidification density under the current specific composition characteristics.

[0011] The execution and feedback module outputs the optimal combination of process parameters to the casting execution system as the casting control benchmark for the current furnace melt, and receives the quality characteristic data fed back after casting to update the correlation of the physical field mapping framework.

[0012] As a further aspect of the present invention: the generation of the dynamic thermophysical property parameter set corresponding to the specific component characteristics specifically includes:

[0013] Using the content of each element in the exclusive component feature as an index factor, at least two sets of reference components with the highest similarity to the exclusive component feature are matched in the association mapping of multiple sets of reference components and corresponding thermophysical parameters stored in the physical field mapping framework.

[0014] Extract the thermophysical parameters corresponding to at least two sets of reference components, and perform weighted interpolation calculation on the extracted thermophysical parameters according to the deviation weight of the content of each element in the specific component characteristics relative to the at least two sets of reference components to obtain the curves of liquidus temperature, latent heat of crystallization and viscosity as a function of temperature under the specific component characteristics.

[0015] The calculated curves of liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature are integrated into a dynamic thermophysical parameter set for output.

[0016] As a further aspect of the present invention: the construction of the dynamic solidification sensitivity coefficient matrix specifically includes:

[0017] Obtain the liquidus temperature and viscosity as a function of temperature curves from the dynamic thermophysical parameter set. Combine the wall thickness ratio of the thin-walled region to the thick-walled region in the geometric structure of the piston to be cast, and calculate the filling resistance index of the thin-walled region at the preset pouring temperature and the feeding potential energy gradient of the thick-walled region at the solidification critical point.

[0018] The filling resistance index and the feeding potential energy gradient are respectively compared with the standard deviation value caused by the deviation of the specific component characteristics from the reference component to obtain the filling sensitivity coefficient of the thin-walled region and the feeding sensitivity coefficient of the thick region.

[0019] The filling sensitivity coefficients of the thin-walled region and the feeding sensitivity coefficients of the thick-walled region are arranged in a matrix according to the spatial distribution of the piston's geometric features to generate a dynamic solidification sensitivity coefficient matrix.

[0020] As a further aspect of the present invention: the calculation process of the filling resistance index is as follows:

[0021] Obtain the viscosity versus temperature curve from the dynamic thermophysical parameter set, and read the corresponding melt viscosity value from the viscosity versus temperature curve according to the preset casting temperature.

[0022] Obtain the minimum cross-sectional thickness and filling path length of the thin-walled region in the geometric features of the piston to be cast. Divide the product of the melt viscosity value and the filling path length by the minimum cross-sectional thickness to obtain the first intermediate value.

[0023] The sum of the absolute values ​​of the deviations between the content of each element in the exclusive component characteristics and the content of the corresponding element in the benchmark component is obtained, and the product of the sum of the absolute values ​​of the deviations and the first intermediate value is used as the filling resistance index of the thin-walled region at the preset casting temperature.

[0024] As a further aspect of the present invention: the calculation process of the compensation potential energy gradient is as follows:

[0025] Obtain the liquidus temperature and latent heat of crystallization from the dynamic thermophysical parameter set. Based on the maximum cross-sectional thickness of the thick region in the current geometric structure of the piston to be cast, calculate the latent heat of crystallization released per unit time in the process of the thick region cooling from the liquidus temperature to the solidus temperature.

[0026] Obtain the cross-sectional area and feeding distance of the feeding channel between the thick region and the adjacent thin-walled region, and divide the product of the latent heat of crystallization and the cross-sectional area of ​​the feeding channel by the feeding distance to obtain the second intermediate value;

[0027] Obtain the ratio of magnesium content to silicon content in the specific composition characteristics, and use the product of the ratio and the second intermediate value as the feeding potential energy gradient of the thick region at the solidification critical point.

[0028] As a further aspect of the present invention: the output process of the optimal combination of process parameters is as follows:

[0029] Obtain the dynamic solidification sensitivity coefficient matrix, and use the ratio of the filling sensitivity coefficient of the thin-walled region to the feeding sensitivity coefficient of the thick region in the matrix as the adjustment weight to generate the pouring temperature adjustment step size, cooling intensity distribution adjustment coefficient and pressurization timing adjustment factor, respectively.

[0030] Starting with the preset combination of benchmark process parameters, the pouring temperature adjustment step size, cooling intensity distribution adjustment coefficient and pressurization pressure timing adjustment factor are adjusted iteratively in sequence. After each round of adjustment, the current filling integrity evaluation value and solidification density evaluation value are calculated respectively.

[0031] The optimal combination of process parameters is output when both the filling integrity evaluation value and the solidification density evaluation value reach the preset threshold, along with the pouring temperature setting, cooling intensity distribution, and pressurization timing.

[0032] As a further aspect of the present invention: the calculation process of the filling integrity evaluation value is as follows:

[0033] Obtain the current iteratively adjusted pouring temperature setting value, and read the melt viscosity value at the corresponding temperature from the viscosity-temperature change curve in the dynamic thermophysical parameter set;

[0034] Obtain the minimum cross-sectional thickness and filling path length of all thin-walled regions in the geometric features of the piston to be cast. Divide the filling path length of each thin-walled region by the minimum cross-sectional thickness of the thin-walled region and multiply it by the melt viscosity value to obtain the filling resistance value of each thin-walled region.

[0035] The reciprocal of the sum of the filling resistance values ​​of each thin-walled region and the deviations of the content of each element in the specific composition characteristics from the baseline composition is taken as the current filling integrity evaluation value.

[0036] As a further aspect of the present invention: the calculation process for the solidification density evaluation value is as follows:

[0037] Obtain the cooling intensity distribution and pressurization timing after the current iteration adjustment, determine the cooling rate of the thick region at the solidification critical point based on the cooling intensity distribution, and calculate the shrinkage per unit volume of the thick region by combining the latent heat of crystallization in the dynamic thermophysical parameter set.

[0038] Obtain the solid fraction threshold corresponding to the start time of pressurization in the pressurization time series, and divide the shrinkage per unit volume of the thick region by the product of the pressurization pressure value and the cross-sectional area of ​​the shrinkage compensation channel to obtain the shrinkage compensation capacity value.

[0039] The shrinkage compensation capacity value and the ratio of magnesium content to silicon content in the specific component characteristics are normalized and used as the current solidification density evaluation value.

[0040] As a further aspect of the present invention: the received quality characteristic data after casting is used to update the correlation of the physical field mapping framework, specifically including:

[0041] The shrinkage cavity size and distribution density of the piston at the preset detection position after casting, as well as the contour integrity of the annular groove region and the skirt region, are obtained and used as the first quality feature data and the second quality feature data, respectively.

[0042] The first quality feature data and the second quality feature data are compared with the preset hole reduction threshold and contour threshold respectively, and the quality deviation vector corresponding to the current exclusive component feature is calculated.

[0043] Based on the sign and magnitude of each component in the mass deviation vector, the mapping values ​​of the thermophysical parameters associated with the current specific component characteristics in the physical field mapping framework are reversed, and the corrected mapping relationship is stored for the generation of dynamic thermophysical parameter sets of the melt in subsequent furnaces.

[0044] The beneficial effects of this invention are:

[0045] (1) Eliminating the negative impact of batch fluctuations in alloy composition on casting quality and improving process stability. This invention obtains the actual composition of each batch of molten aluminum in real time through online spectral analysis, and dynamically reconstructs the curves of liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature. This allows process parameter decisions to no longer rely on static alloy standard grades, but to be specifically optimized for the actual physical properties of the melt in the current batch. When the magnesium content is too low or the iron content is too high, causing a decrease in the fluidity of the molten aluminum, this invention can automatically adjust the pouring temperature, cooling intensity, and pressurization timing to compensate, avoiding the risk of batch scrap caused by manual adjustment after defects occur in traditional methods, and ensuring that the yield remains highly consistent across different batches and under different environmental conditions.

[0046] (2) Breaking through the timeliness bottleneck of offline numerical simulation, enabling rapid response in multi-variety, small-batch production. This invention transforms the orthogonal experiment and numerical simulation process, which originally required hours or even days, into second-level inference through a pre-constructed physical field mapping framework, eliminating the need to remodel for each new piston or each batch with compositional fluctuations. When developing a new piston model or encountering batches with abnormal composition, this invention can quickly generate an initial optimal combination of process parameters based on historical data, and rapidly converge to a parameter solution that satisfies both the filling integrity and solidification density thresholds through iterative adjustments, shortening the process parameter development cycle and reducing material and energy waste during the trial production process. Attached Figure Description

[0047] The invention will now be further described with reference to the accompanying drawings.

[0048] Figure 1 This is a system block diagram of the present invention;

[0049] Figure 2 This is a flowchart of the calculation process of the filling resistance index in this invention;

[0050] Figure 3 This is a flowchart of the calculation process of the compensating potential energy gradient in this invention. Detailed Implementation

[0051] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Please see Figure 1 As shown, this invention is an intelligent decision-making system for aluminum alloy piston casting process parameters, comprising:

[0053] The composition sensing module acquires real-time composition data of the aluminum alloy melt to be cast. The real-time composition data is obtained directly before the melt is poured through online spectral analysis and serves as the exclusive composition characteristics of the melt for the current furnace batch.

[0054] The property reconstruction module inputs the specific composition features into a pre-built physical field mapping framework. The physical field mapping framework establishes the relationship between composition and thermal properties based on historical casting data, and generates a dynamic thermal property parameter set corresponding to the specific composition features. The dynamic thermal property parameter set includes liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature.

[0055] The sensitivity analysis module, based on the dynamic thermophysical property parameter set, combined with the geometric structural characteristics of the piston to be cast and the preset casting quality target, constructs a dynamic solidification sensitivity coefficient matrix by quantifying the attenuation of filling driving force and the sensitivity shift of feeding channel caused by the deviation of exclusive composition characteristics from the reference composition.

[0056] The parameter optimization module uses the dynamic solidification sensitivity coefficient matrix as the constraint boundary for multi-objective optimization. It solves for the optimal combination of process parameters, which is the casting temperature setting, cooling intensity distribution, and pressurization pressure sequence that simultaneously achieve the preset thresholds for filling integrity and solidification density under the current specific composition characteristics.

[0057] The execution and feedback module outputs the optimal combination of process parameters to the casting execution system as the casting control benchmark for the current furnace melt, and receives the quality characteristic data fed back after casting to update the correlation of the physical field mapping framework.

[0058] In the composition sensing module, real-time composition data of the aluminum alloy melt to be cast is acquired. This real-time composition data is obtained directly before the melt is poured using online spectral analysis and serves as the unique compositional characteristics of the melt for the current furnace batch. Specifically, it includes:

[0059] Sampling points are set at the flow channel between the aluminum outlet of the smelting furnace and the ladle. Before the start of each heat of molten aluminum pouring, an online spectroscopic analysis device is activated manually or automatically. The online spectroscopic analysis device uses a laser-induced breakdown spectrometer, which focuses a pulsed laser onto the surface of the molten aluminum flowing through the sampling point, generating plasma on the surface of the molten aluminum. By collecting the plasma emission spectrum and analyzing its characteristic spectral lines, the mass percentage content of magnesium, silicon, iron, and copper in the molten aluminum, as well as the mass percentage content of other trace alloying elements that affect the thermophysical parameters of the aluminum alloy, are quantitatively obtained.

[0060] The contents of magnesium, silicon, iron, copper, and other trace alloying elements are integrated according to a preset data format to form a unique compositional feature corresponding to the current furnace of molten aluminum. This unique compositional feature contains numerical information on the content of each element, used to characterize the actual alloy composition of the melt in the current furnace, and serves as the input basis for generating subsequent dynamic thermophysical property parameters.

[0061] In the property reconstruction module, the specific composition features are input into a pre-constructed physical field mapping framework. Based on the correlation between composition and thermal properties established by historical casting data, the physical field mapping framework generates a dynamic set of thermal property parameters corresponding to the specific composition features. The dynamic set of thermal property parameters includes liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature, specifically including:

[0062] First, a physical field mapping framework is constructed. This framework refers to a pre-established set of correlation mappings between multiple sets of reference components and their corresponding thermophysical parameters, stored in a data storage unit. The reference components are standardized alloy composition data determined for different aluminum alloy grades and combinations of element content. Each set of reference components includes a fixed mass percentage of magnesium, silicon, iron, copper, and other trace alloying elements. The corresponding thermophysical parameters are the liquidus temperature, latent heat of crystallization, and viscosity versus temperature curves obtained by measuring the aluminum alloy samples corresponding to each set of reference components using differential scanning calorimetry and a thermophysical property testing instrument. In this correlation mapping set, each set of reference components forms a one-to-one mapping relationship with its corresponding thermophysical parameter, storing no fewer than 50 sets of reference components and their corresponding thermophysical parameters.

[0063] Secondly, similarity matching is performed between the specific component features and the reference components. Using the element content of each element in the specific component features of the current melt as index factors, the similarity between the specific component features and each set of reference components in the physical field mapping framework is calculated one by one. The similarity calculation method is as follows: calculate the absolute value of the difference between the magnesium, silicon, iron, and copper content in the specific component features and the corresponding element content in the reference components. Add the absolute values ​​of these element content differences and take the reciprocal; the resulting value is the similarity value. The larger the similarity value, the closer the specific component features are to the set of reference components. The reference components are sorted in descending order of similarity value, and at least two sets of reference components with the highest similarity values ​​are selected as matching references. The set with the highest similarity value is designated as the first reference component, and the set with the second highest similarity value is designated as the second reference component.

[0064] Next, the thermophysical parameters corresponding to the matching benchmarks are extracted and deviation weights are calculated. The curves of the first liquidus temperature, first latent heat of crystallization, and first viscosity as a function of temperature for the first benchmark component, and the curves of the second liquidus temperature, second latent heat of crystallization, and second viscosity as a function of temperature for the second benchmark component, are extracted from the physical field mapping framework. Based on the degree of deviation of the content of each element in the specific component characteristics relative to the first and second benchmark components, a first deviation weight and a second deviation weight are calculated. The first deviation weight is calculated as follows: the absolute values ​​of the differences between the content of each element in the specific component characteristics and the content of each element in the first benchmark component are added to obtain the first sum of differences; the absolute values ​​of the differences between the content of each element in the specific component characteristics and the content of each element in the second benchmark component are added to obtain the second sum of differences; the second sum of differences is divided by the sum of the first and second sums of differences, and the resulting value is used as the first deviation weight; the first sum of differences is divided by the sum of the first and second sums of differences, and the resulting value is used as the second deviation weight. The sum of the first and second deviation weights equals 1.

[0065] Finally, weighted interpolation calculations are performed, and a dynamic thermophysical property parameter set is output. The first deviation weight is multiplied by the first liquidus temperature to obtain the first weighted liquidus temperature; the second deviation weight is multiplied by the second liquidus temperature to obtain the second weighted liquidus temperature; the first weighted liquidus temperature and the second weighted liquidus temperature are added together to obtain the liquidus temperature under the specific composition characteristics. Following the same method, the first deviation weight is multiplied by the first latent heat of crystallization, and the second deviation weight is multiplied by the second latent heat of crystallization; the sum of these two multiplications yields the latent heat of crystallization under the specific composition characteristics. For the viscosity versus temperature curve, multiple temperature sampling points are selected at preset temperature intervals on the temperature axis. At each sampling point, the first deviation weight is multiplied by the viscosity value of the first viscosity versus temperature curve at that sampling point, and the second deviation weight is multiplied by the viscosity value of the second viscosity versus temperature curve at that sampling point; the sum of these two multiplications yields the viscosity value at that sampling point. Connecting the viscosity values ​​of all sampling points in temperature order yields the viscosity versus temperature curve under the specific composition characteristics. The calculated curves of liquidus temperature, latent heat of crystallization, and viscosity versus temperature are integrated into a dynamic thermophysical parameter set, which is then used as output for subsequent steps.

[0066] Please see Figure 2 or Figure 3 As shown, in the sensitivity analysis module, based on the dynamic thermophysical property parameter set, combined with the geometric characteristics of the piston to be cast and the preset casting quality target, a dynamic solidification sensitivity coefficient matrix is ​​constructed by quantifying the attenuation of the filling driving force and the sensitivity shift of the feeding channel caused by the deviation of the specific composition characteristics from the reference composition. Specifically, this includes:

[0067] First, the liquidus temperature and viscosity as a function of temperature are obtained from the dynamic thermophysical parameter set, and the geometric features of the piston to be cast are acquired. The geometric features of the piston to be cast refer to the parameters of the piston solid model extracted by 3D modeling software, including the division of thin-walled and thick-walled regions, the minimum and maximum cross-sectional thickness of each region, the filling path length, the cross-sectional area of ​​the feeding channel, and the feeding distance. The thin-walled region refers to the area with a piston wall thickness of 6 mm or less, specifically including the piston skirt and the bottom of the annular groove; the thick-walled region refers to the area with a piston wall thickness of 12 mm or more, specifically including the piston top and the area around the pin hole. The preset pouring temperature refers to the pouring temperature value pre-set according to the aluminum alloy grade. For hypereutectic aluminum-silicon alloys, this value is usually set between 740°C and 760°C; for eutectic aluminum-silicon alloys, this value is usually set between 710°C and 730°C.

[0068] Next, the filling resistance index of the thin-walled region at the preset pouring temperature is calculated. The melt viscosity value corresponding to the preset pouring temperature is read from the viscosity-temperature curve in the dynamic thermophysical parameter set and recorded as the first parameter. From the geometric characteristics of the piston to be cast, the minimum cross-sectional thickness and filling path length of the thin-walled region are obtained, where the filling path length refers to the longest flow distance traveled by the molten aluminum from the gate into the mold cavity to the end of the thin-walled region. The melt viscosity value is multiplied by the filling path length to obtain the first product, and the first product is divided by the minimum cross-sectional thickness to obtain the first intermediate value. The sum of the absolute values ​​of the deviations between the content of each element in the specific composition characteristics and the corresponding element content of the reference composition is obtained, where the reference composition refers to the first reference composition with the highest similarity in the physical field mapping framework. The sum of the absolute values ​​of the deviations is multiplied by the first intermediate value to obtain the filling resistance index of the thin-walled region at the preset pouring temperature, calculated according to the following formula:

[0069] ;

[0070] in, Indicates the filling resistance index. This indicates the melt viscosity value at the preset casting temperature. Indicates the filling path length of the thin-walled region. This represents the minimum cross-sectional thickness of the thin-walled region. Indicating the first characteristic of specific components The content of each element Indicates the first reference component The content of each element This indicates the number of elements involved in the calculation, specifically including four elements: magnesium, silicon, iron, and copper. Indicates the sequence number of the thin-walled region. The index indicating the type of element.

[0071] Next, the feeding potential energy gradient of the thick region at the solidification critical point is calculated. The liquidus temperature and latent heat of crystallization from the dynamic thermophysical parameter set are obtained, along with the maximum cross-sectional thickness of the thick region in the geometric features of the piston to be cast. The solidification critical point refers to the moment when the solid fraction reaches 70% during the cooling process from the liquidus temperature to the solidus temperature of the thick region. Based on the maximum cross-sectional thickness of the thick region, the latent heat of crystallization released from this region per unit time is calculated. Specifically, the maximum cross-sectional thickness of the thick region is divided by a preset solidification time coefficient to obtain the solidification rate, and then the latent heat of crystallization is multiplied by the solidification rate to obtain the latent heat of crystallization released per unit time. The cross-sectional area and feeding distance of the feeding channel between the thick region and the adjacent thin-walled region are obtained, where the feeding channel refers to the flow path for molten metal feeding from the thick region to the adjacent thin-walled region. The latent heat of crystallization is multiplied by the cross-sectional area of ​​the feeding channel to obtain a second product, and the second product is divided by the feeding distance to obtain a second intermediate value. Obtain the ratio of magnesium to silicon content in the specific composition characteristics, multiply this ratio by the second intermediate value, and obtain the feeding potential energy gradient of the thick region at the solidification critical point, calculated according to the following formula:

[0072] ;

[0073] in, This represents the gradient of the compensating potential energy. This represents the latent heat of crystallization released per unit time in a thick region at the solidification critical point. This indicates the cross-sectional area of ​​the channel being compensated. Indicates the compensation distance. This indicates the percentage by mass of magnesium in the specific ingredient characteristics. This indicates the percentage by mass of silicon in the specific ingredient characteristics.

[0074] Then, the filling sensitivity coefficient for thin-walled regions and the feeding sensitivity coefficient for thick regions are calculated. The standard deviation value caused by deviations of the specific component characteristics from the reference component is obtained. The standard deviation value refers to the root mean square value of the difference between the content of each element in the specific component characteristics and the content of each element in the first reference component. Specifically, the calculation method is as follows: calculate the difference between the content of magnesium, silicon, iron, and copper elements and the corresponding element content in the first reference component, sum the squares of each difference to obtain the sum of squares, divide the sum of squares by the number of element types involved in the calculation (4) to obtain the mean square of the differences, and then take the square root of the mean square of the differences to obtain the standard deviation value. The filling resistance index is divided by the standard deviation value to obtain the filling sensitivity coefficient for thin-walled regions. The feeding potential energy gradient is divided by the standard deviation value to obtain the feeding sensitivity coefficient for thick regions.

[0075] Finally, a dynamic solidification sensitivity coefficient matrix is ​​generated. The filling sensitivity coefficients of the thin-walled region and the shrinkage sensitivity coefficients of the thick-walled region are arranged in a matrix according to the spatial distribution of the piston's geometric features. Specifically, the piston is divided into three sections along the axial direction (top region, annular groove region, and skirt region) and three sections along the radial direction (center region, middle region, and edge region), forming a 3x3 matrix structure. The filling sensitivity coefficient of the thin-walled region or the shrinkage sensitivity coefficient of the thick-walled region corresponding to each region is filled into the corresponding position in the matrix. For regions that contain both thin-walled and thick-walled features, the larger value of the two is used, resulting in the final dynamic solidification sensitivity coefficient matrix.

[0076] In the parameter optimization module, the dynamic solidification sensitivity coefficient matrix is ​​used as the constraint boundary for multi-objective optimization. The optimal combination of pouring temperature setpoint, cooling intensity distribution, and pressurization pressure sequence is determined to achieve both filling integrity and solidification density at preset thresholds under the current specific composition characteristics. This optimal combination of process parameters includes:

[0077] First, a dynamic solidification sensitivity coefficient matrix is ​​obtained. From this matrix, the filling sensitivity coefficients corresponding to all thin-walled regions and the feeding sensitivity coefficients corresponding to all thick regions are extracted. The average filling sensitivity coefficient of each thin-walled region is calculated to obtain the average filling sensitivity coefficient; the average feeding sensitivity coefficient of each thick region is calculated to obtain the average feeding sensitivity coefficient. The average filling sensitivity coefficient is divided by the average feeding sensitivity coefficient to obtain the adjustment weight. Based on the adjustment weight, the pouring temperature adjustment step, the cooling intensity distribution adjustment coefficient, and the pressurization pressure timing adjustment factor are generated. Specifically, the generation method is as follows: the preset reference pouring temperature adjustment step is multiplied by the adjustment weight to obtain the pouring temperature adjustment step; the preset reference cooling intensity distribution adjustment coefficient is multiplied by the adjustment weight to obtain the cooling intensity distribution adjustment coefficient; and the preset reference pressurization pressure timing adjustment factor is multiplied by the adjustment weight to obtain the pressurization pressure timing adjustment factor. The reference pouring temperature adjustment step is set to 5 degrees Celsius, the reference cooling intensity distribution adjustment coefficient is set to 0.1, and the reference pressurization pressure timing adjustment factor is set to 0.05 seconds.

[0078] Secondly, the iterative process begins with a preset combination of benchmark process parameters, which includes a benchmark pouring temperature, a benchmark cooling intensity distribution, and a benchmark pressurization timing. The benchmark pouring temperature is preset according to the aluminum alloy grade, and is set to 720 degrees Celsius for eutectic aluminum-silicon alloys. The benchmark cooling intensity distribution refers to dividing the cooling water flow rate along the piston axial direction into three segments: top cooling flow rate, annular groove cooling flow rate, and skirt cooling flow rate. The benchmark flow rates for each segment are set to 3 liters per minute, 2.5 liters per minute, and 2 liters per minute, respectively. The benchmark pressurization timing refers to the curve of pressurization pressure changing over time, where the pressurization start time is set to 3 seconds after pouring is completed, and the target pressurization pressure value is set to 120 MPa. The baseline process parameter combination is iteratively adjusted according to the pouring temperature adjustment step size, cooling intensity distribution adjustment coefficient, and pressurization timing adjustment factor. Each round of adjustment involves: increasing or decreasing the current pouring temperature by one pouring temperature adjustment step size to obtain a new pouring temperature setpoint; multiplying the current cooling flow rate of each section by the cooling intensity distribution adjustment coefficient and adding this to the current flow rate to obtain a new cooling intensity distribution; and proportionally adjusting the pressure values ​​at each time point in the current pressurization timing according to the pressurization timing adjustment factor to obtain a new pressurization timing. After each round of adjustment, the filling integrity evaluation value and solidification density evaluation value under the current process parameter combination are calculated.

[0079] Next, the calculation process for the filling integrity evaluation value is as follows: Obtain the current iteratively adjusted pouring temperature setting. On the viscosity-temperature change curve in the dynamic thermophysical parameter set, read the melt viscosity value corresponding to the pouring temperature setting value using a linear interpolation method. Obtain the minimum cross-sectional thickness and filling path length of all thin-walled regions in the geometric structural features of the piston to be cast. Divide the filling path length of each thin-walled region by the minimum cross-sectional thickness of that region to obtain the length-to-thickness ratio of each thin-walled region. Then multiply the length-to-thickness ratio of each thin-walled region by the melt viscosity value to obtain the filling resistance value of each thin-walled region. Obtain the sum of the absolute values ​​of the deviations between the content of each element in the specific composition feature and the content of each element in the first reference composition. Use this sum of absolute values ​​of deviation as the deviation weighting value. Add the filling resistance value of each thin-walled region to the deviation weighting value to obtain the comprehensive filling resistance of each thin-walled region. Sum the comprehensive filling resistances of all thin-walled regions to obtain the total filling resistance. Divide 1 by the total filling resistance, and the resulting value is used as the current filling integrity evaluation value.

[0080] The calculation process for the solidification density evaluation value is as follows: The cooling intensity distribution after the current iteration is obtained, and the cooling rate of the thick region at the solidification critical point is determined based on this distribution. Specifically, the cooling rate of the thick region at the solidification critical point is obtained through a pre-established cooling rate mapping table. This table records the measured cooling rate values ​​at various locations within the thick region under different cooling intensity distributions, with the cooling intensity distribution characterized by the cooling flow rate of each segment. The latent heat of crystallization is obtained from the dynamic thermophysical parameter set. The cooling rate of the thick region at the solidification critical point is multiplied by the latent heat of crystallization, and then multiplied by a preset unit volume conversion factor to obtain the shrinkage per unit volume of the thick region. The pressurization pressure sequence after the current iteration is obtained, and the solidity threshold corresponding to the pressurization start time is read from the pressurization pressure sequence. This solidity threshold is pre-determined using differential scanning calorimetry and is set to 65% for hypereutectic aluminum-silicon alloys. Obtain the target pressure value and the cross-sectional area of ​​the compensation channel in the pressurization pressure sequence. Divide the shrinkage per unit volume of the thick area by the product of the target pressure value and the cross-sectional area of ​​the compensation channel to obtain the shrinkage compensation capacity value. Obtain the ratio of magnesium content to silicon content in the specific composition characteristics. Use this ratio as a composition correction factor. Divide the shrinkage compensation capacity value by the composition correction factor to obtain the corrected shrinkage compensation capacity value. Then, normalize the corrected shrinkage compensation capacity value by dividing it by a preset baseline shrinkage compensation capacity value. The resulting value is used as the current solidification density evaluation value, where the baseline shrinkage compensation capacity value is set to 0.8.

[0081] Finally, the iterative adjustment and evaluation value calculation process described above is repeated until both the filling integrity evaluation value and the solidification density evaluation value reach the preset thresholds. The preset threshold for the filling integrity evaluation value is set to 0.85, and the preset threshold for the solidification density evaluation value is set to 0.9. When both the filling integrity evaluation value and the solidification density evaluation value simultaneously reach or exceed their respective preset thresholds, the iteration stops, and the pouring temperature setting, cooling intensity distribution, and pressurization timing corresponding to this round of adjustment are output as the optimal combination of process parameters.

[0082] In the execution and feedback module, the optimal combination of process parameters is output to the casting execution system as the casting control benchmark for the current furnace melt, and the quality characteristic data fed back after casting is received to update the correlation of the physical field mapping framework, specifically including:

[0083] First, after the piston is cast, non-destructive testing is performed to obtain quality characteristic data. An industrial X-ray three-dimensional tomography scanner is used to perform a full-size scan of the piston, and the size and distribution density of shrinkage cavities at preset detection locations are extracted as the first quality characteristic data. These preset detection locations include the central area of ​​the piston top, the area around the pin hole, and the transition area between the thick and thin-walled regions, totaling five detection locations. The shrinkage cavity size refers to the maximum diameter of a single shrinkage cavity at the detection location, measured in millimeters; the distribution density refers to the number of shrinkage cavities per unit area at the detection location, counted as the number per square centimeter. Simultaneously, an optical three-dimensional scanner is used to scan the piston's outer contour, obtaining the contour integrity of the annular groove region and the skirt region as the second quality characteristic data. The annular groove region includes three annular grooves: the first annular groove, the second annular groove, and the oil annular groove. Contour integrity refers to the degree of conformity between the actual contour and the standard contour, characterized by calculating the average deviation distance between the corresponding points of the actual contour point cloud and the standard three-dimensional model, measured in millimeters. The skirt area includes the piston thrust surface and the thrust bearing surface, and the integrity of the profile is also characterized by the average deviation distance between the actual profile and the standard profile.

[0084] Next, the first quality feature data is compared with a preset cavity threshold, and the second quality feature data is compared with a preset contour threshold to calculate the quality deviation vector corresponding to the current specific component feature. The cavity threshold includes a cavity size threshold and a distribution density threshold, wherein the cavity size threshold is set to 0.3 mm, and the distribution density threshold is set to 2 per square centimeter. The contour threshold is set to 0.05 mm. The quality deviation vector contains 5 components, corresponding to the cavity quality deviation, the annular groove region contour deviation, and the skirt region contour deviation at the 5 detection locations, respectively. For the cavity quality deviation, the difference between the actual cavity size and the cavity size threshold, and the difference between the actual distribution density and the distribution density threshold are calculated for each detection location. These two differences are divided by the corresponding thresholds to obtain the cavity size deviation coefficient and the distribution density deviation coefficient. The two are added together and the average value is taken as the cavity quality deviation component at that detection location. For the profile quality deviation, the difference between the actual average deviation distance of the annular groove region and the profile threshold is divided by the profile threshold to obtain the profile deviation component of the annular groove region; the difference between the actual average deviation distance of the skirt region and the profile threshold is divided by the profile threshold to obtain the profile deviation component of the skirt region. The above five deviation components together constitute the quality deviation vector.

[0085] Next, based on the sign and magnitude of each component in the mass deviation vector, the mapping values ​​of the thermophysical parameters associated with the current specific component characteristics in the physical field mapping framework are reverse-corrected. Specifically, the liquidus temperature, latent heat of crystallization, and viscosity versus temperature curves corresponding to the current specific component characteristics are extracted from the physical field mapping framework as the mapping values ​​of the thermophysical parameters to be corrected. The arithmetic mean of all components in the mass deviation vector is calculated to obtain the average deviation coefficient. If the average deviation coefficient is positive, it indicates that the casting quality is too porous or the profile is too large under the current process parameter combination, indicating that there is a deviation between the actual thermophysical parameters of the aluminum melt and the current mapping values, and the mapping values ​​of the thermophysical parameters need to be corrected in the direction of decreasing; if the average deviation coefficient is negative, it indicates that the casting quality is too dense or the profile is too small, and the mapping values ​​of the thermophysical parameters need to be corrected in the direction of increasing. The correction amplitude is determined based on the absolute value of the average deviation coefficient. The absolute value of the average deviation coefficient is multiplied by a preset correction step size to obtain the correction amount. The correction step size for the liquidus temperature is set to 2 degrees Celsius, the correction step size for the latent heat of crystallization is set to 5 joules per gram, and the correction step size for the viscosity versus temperature curve is set to 3% of the current viscosity value. The corrected liquidus temperature is obtained by adding or subtracting the correction amount from the mapped liquidus temperature value; the corrected latent heat of crystallization is obtained by adding or subtracting the correction amount from the mapped latent heat of crystallization value; and the corrected viscosity versus temperature curve is obtained by adding or subtracting the viscosity value at each temperature point in the viscosity versus temperature curve.

[0086] Finally, the corrected curves of liquidus temperature, latent heat of crystallization, and viscosity versus temperature are used as updated thermophysical parameter mapping values. These values ​​are then correlated with the current specific compositional characteristics and stored in the physical field mapping framework for use in generating dynamic thermophysical parameter sets for subsequent furnace melts. Simultaneously, the original mapping relationships before correction are retained as historical records for future analysis.

[0087] The working principle of this invention is as follows: First, the content of elements such as magnesium, silicon, iron, and copper in the molten aluminum alloy to be cast is obtained in real time through online spectral analysis, forming a unique compositional characteristic for the current furnace batch. Then, this unique compositional characteristic is input into a pre-constructed physical field mapping framework. Through similarity matching and weighted interpolation calculation, curves of liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature are generated that dynamically correspond to this compositional characteristic. Next, based on this dynamic thermophysical parameter set and the piston geometry, a dynamic solidification sensitivity coefficient matrix is ​​constructed by calculating the filling resistance index of thin-walled regions and the feeding potential energy gradient of thick regions. Then, using this matrix as a constraint boundary, the filling temperature, cooling intensity distribution, and pressurization timing are iteratively adjusted to calculate the filling integrity evaluation value and solidification density evaluation value, respectively, until both reach a preset threshold, and the optimal process parameter combination is output. Finally, this optimal process parameter combination is executed as the casting control benchmark, and the mapping values ​​of thermophysical parameters associated with this unique compositional characteristic in the physical field mapping framework are corrected in reverse based on the quality characteristic data such as shrinkage cavity size, distribution density, and contour integrity detected after casting, thus realizing the dynamic updating of the correlation.

[0088] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An aluminum alloy piston casting process parameter intelligent decision system, characterized in that, include: The composition sensing module acquires real-time composition data of the aluminum alloy melt to be cast. The real-time composition data is obtained directly before the melt is poured through online spectral analysis and serves as the exclusive composition characteristics of the melt for the current furnace batch. The property reconstruction module inputs the specific composition features into a pre-built physical field mapping framework. The physical field mapping framework establishes the relationship between composition and thermal properties based on historical casting data, and generates a dynamic thermal property parameter set corresponding to the specific composition features. The dynamic thermal property parameter set includes liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature. The sensitivity analysis module, based on the dynamic thermophysical property parameter set, combined with the geometric structural characteristics of the piston to be cast and the preset casting quality target, constructs a dynamic solidification sensitivity coefficient matrix by quantifying the attenuation of filling driving force and the sensitivity shift of feeding channel caused by the deviation of exclusive composition characteristics from the reference composition. The parameter optimization module uses the dynamic solidification sensitivity coefficient matrix as the constraint boundary for multi-objective optimization. It solves for the optimal combination of process parameters, which is the casting temperature setting, cooling intensity distribution, and pressurization pressure sequence that simultaneously achieve the preset thresholds for filling integrity and solidification density under the current specific composition characteristics. The execution and feedback module outputs the optimal combination of process parameters to the casting execution system as the casting control benchmark for the current furnace melt, and receives the quality characteristic data fed back after casting to update the correlation of the physical field mapping framework.

2. An aluminum alloy piston casting process parameter intelligent decision system according to claim 1, characterized in that, The generation of the dynamic thermophysical property parameter set corresponding to the specific component characteristics specifically includes: Using the content of each element in the exclusive component feature as an index factor, at least two sets of reference components with the highest similarity to the exclusive component feature are matched in the association mapping of multiple sets of reference components and corresponding thermophysical parameters stored in the physical field mapping framework. Extract the thermophysical parameters corresponding to at least two sets of reference components, and perform weighted interpolation calculation on the extracted thermophysical parameters according to the deviation weight of the content of each element in the specific component characteristics relative to the at least two sets of reference components to obtain the curves of liquidus temperature, latent heat of crystallization and viscosity as a function of temperature under the specific component characteristics. The calculated curves of liquidus temperature, latent heat of crystallization, and viscosity as a function of temperature are integrated into a dynamic thermophysical parameter set for output.

3. The intelligent decision system for process parameters of aluminum alloy piston casting according to claim 1, characterized in that, The construction of the dynamic solidification sensitivity coefficient matrix specifically includes: Obtain the liquidus temperature and viscosity as a function of temperature curves from the dynamic thermophysical parameter set. Combine the wall thickness ratio of the thin-walled region to the thick-walled region in the geometric structure of the piston to be cast, and calculate the filling resistance index of the thin-walled region at the preset pouring temperature and the feeding potential energy gradient of the thick-walled region at the solidification critical point. The filling resistance index and the feeding potential energy gradient are respectively compared with the standard deviation value caused by the deviation of the specific component characteristics from the reference component to obtain the filling sensitivity coefficient of the thin-walled region and the feeding sensitivity coefficient of the thick region. The filling sensitivity coefficients of the thin-walled region and the feeding sensitivity coefficients of the thick-walled region are arranged in a matrix according to the spatial distribution of the piston's geometric features to generate a dynamic solidification sensitivity coefficient matrix.

4. The intelligent decision system for process parameters of aluminum alloy piston casting according to claim 3, characterized in that, The calculation process for the filling resistance index is as follows: Obtain the viscosity versus temperature curve from the dynamic thermophysical parameter set, and read the corresponding melt viscosity value from the viscosity versus temperature curve according to the preset casting temperature. Obtain the minimum cross-sectional thickness and filling path length of the thin-walled region in the geometric features of the piston to be cast. Divide the product of the melt viscosity value and the filling path length by the minimum cross-sectional thickness to obtain the first intermediate value. The sum of the absolute values ​​of the deviations between the content of each element in the exclusive component characteristics and the content of the corresponding element in the benchmark component is obtained, and the product of the sum of the absolute values ​​of the deviations and the first intermediate value is used as the filling resistance index of the thin-walled region at the preset casting temperature.

5. The intelligent decision system for process parameters of aluminum alloy piston casting process as claimed in claim 3 wherein, The calculation process for the compensation potential energy gradient is as follows: Obtain the liquidus temperature and latent heat of crystallization from the dynamic thermophysical parameter set. Based on the maximum cross-sectional thickness of the thick region in the current geometric structure of the piston to be cast, calculate the latent heat of crystallization released per unit time in the process of the thick region cooling from the liquidus temperature to the solidus temperature. Obtain the cross-sectional area and feeding distance of the feeding channel between the thick region and the adjacent thin-walled region, and divide the product of the latent heat of crystallization and the cross-sectional area of ​​the feeding channel by the feeding distance to obtain the second intermediate value; Obtain the ratio of magnesium content to silicon content in the specific composition characteristics, and use the product of the ratio and the second intermediate value as the feeding potential energy gradient of the thick region at the solidification critical point.

6. An aluminum alloy piston casting process parameter intelligent decision system according to claim 1, characterized in that, The output process of the optimal combination of process parameters is as follows: Obtain the dynamic solidification sensitivity coefficient matrix, and use the ratio of the filling sensitivity coefficient of the thin-walled region to the feeding sensitivity coefficient of the thick region in the matrix as the adjustment weight to generate the pouring temperature adjustment step size, cooling intensity distribution adjustment coefficient and pressurization timing adjustment factor, respectively. Starting with the preset combination of benchmark process parameters, the pouring temperature adjustment step size, cooling intensity distribution adjustment coefficient and pressurization pressure timing adjustment factor are adjusted iteratively in sequence. After each round of adjustment, the current filling integrity evaluation value and solidification density evaluation value are calculated respectively. The optimal combination of process parameters is output when both the filling integrity evaluation value and the solidification density evaluation value reach the preset threshold, along with the pouring temperature setting, cooling intensity distribution, and pressurization timing.

7. An aluminum alloy piston casting process parameter intelligent decision system according to claim 6, characterized in that, The calculation process for the filling integrity evaluation value is as follows: Obtain the current iteratively adjusted pouring temperature setting value, and read the melt viscosity value at the corresponding temperature from the viscosity-temperature change curve in the dynamic thermophysical parameter set; Obtain the minimum cross-sectional thickness and filling path length of all thin-walled regions in the geometric features of the piston to be cast. Divide the filling path length of each thin-walled region by the minimum cross-sectional thickness of the thin-walled region and multiply it by the melt viscosity value to obtain the filling resistance value of each thin-walled region. The reciprocal of the sum of the filling resistance values ​​of each thin-walled region and the deviations of the content of each element in the specific composition characteristics from the baseline composition is taken as the current filling integrity evaluation value.

8. The intelligent decision-making system for aluminum alloy piston casting process parameters according to claim 6, characterized in that, The calculation process for the solidification density evaluation value is as follows: Obtain the cooling intensity distribution and pressurization timing after the current iteration adjustment, determine the cooling rate of the thick region at the solidification critical point based on the cooling intensity distribution, and calculate the shrinkage per unit volume of the thick region by combining the latent heat of crystallization in the dynamic thermophysical parameter set. Obtain the solid fraction threshold corresponding to the start time of pressurization in the pressurization time series, and divide the shrinkage per unit volume of the thick region by the product of the pressurization pressure value and the cross-sectional area of ​​the shrinkage compensation channel to obtain the shrinkage compensation capacity value. The shrinkage compensation capacity value and the ratio of magnesium content to silicon content in the specific component characteristics are normalized and used as the current solidification density evaluation value.

9. The intelligent decision-making system for aluminum alloy piston casting process parameters according to claim 1, characterized in that, The received quality characteristic data after casting is used to update the correlation of the physical field mapping framework, specifically including: The shrinkage cavity size and distribution density of the piston at the preset detection position after casting, as well as the contour integrity of the annular groove region and the skirt region, are obtained and used as the first quality feature data and the second quality feature data, respectively. The first quality feature data and the second quality feature data are compared with the preset hole reduction threshold and contour threshold respectively, and the quality deviation vector corresponding to the current exclusive component feature is calculated. Based on the sign and magnitude of each component in the mass deviation vector, the mapping values ​​of the thermophysical parameters associated with the current specific component characteristics in the physical field mapping framework are reversed, and the corrected mapping relationship is stored for the generation of dynamic thermophysical parameter sets of the melt in subsequent furnaces.

Images