Multi-factor coupling casting global comprehensive force high-precision digital intelligence real-time prediction method

Abstract

The invention relates to a multi-factor coupling casting global comprehensive force high-precision digital intelligence real-time prediction method, which comprises the following steps: constructing a material process-solidification cooling speed and sequence-mechanical property digital twinning quantitative relation model under a defect-free condition, and determining a model formula coefficient through physical twinning; based on physical and digital twinning, a defect-mechanical property quantitative relation model of a material belt defect area under the defect condition is constructed, and a corresponding material casting global comprehensive force high-precision digital intelligence prediction model is constructed in combination with machine learning; designing and manufacturing characteristic standard castings containing defect-free and multi-level defect areas, comparing comprehensive force properties of the characteristic standard castings obtained by model prediction and physical twinning, and verifying and correcting the model; a digital and physical twin product casting is adopted, the solidification cooling speed and sequence and defect data are collected in real time, and the model is combined to predict the global comprehensive force of the product casting in real time with high precision. Real-time digital-intelligent visualization of casting force is achieved, and high-reliability, light-weight and digital-intelligent design and manufacturing of castings can be supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cast alloy material performance prediction and casting preparation, and particularly relates to a high-precision numerical intelligence real-time prediction method for coupling multi-factor comprehensive mechanical properties of castings. BACKGROUND

[0002] With the urgent demand for extreme lightweight in the fields of aerospace and automobiles, aerospace large complex castings and automobile large thin-walled castings are developing towards high-reliability, rapid lightweight and high-precision numerical intelligence design and manufacturing. High-reliability design of castings requires comprehensive mechanical properties (such as yield strength, tensile strength, elongation and Young's modulus) of the castings to meet the standards; rapid lightweight design needs to be designed and checked according to the strength and stiffness of the castings to ensure that the castings meet the load requirements after weight reduction. However, the traditional simulation can only simulate the filling and solidification process and internal defects of the castings, and cannot predict the comprehensive mechanical properties of the castings, thereby limiting the development of high-reliability, rapid lightweight and high-precision numerical intelligence design and manufacturing of castings. Therefore, it is urgent to develop a cast mechanical property real-time numerical visualization technology based on a cast mechanical property prediction method to realize high-reliability, rapid lightweight and high-precision numerical intelligence design and manufacturing of castings.

[0003] On the other hand, the existing cast microstructure and mechanical property prediction method has significant limitations in the application of cast high-reliability, rapid lightweight and high-precision numerical intelligence design and manufacturing, and cannot meet the requirements of cast high-reliability, rapid lightweight and high-precision numerical intelligence design and manufacturing. Therefore, it is urgent to develop a high-precision numerical intelligence real-time prediction method for coupling multi-factor (grain size, secondary dendrite arm spacing, microcomponent difference and defects) comprehensive mechanical properties of castings including yield strength, tensile strength, elongation and stiffness prediction data to overcome the above problems. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a high-precision numerical intelligence real-time prediction method for coupling multi-factor comprehensive mechanical properties of castings, which combines machine learning to construct and correct a high-precision numerical intelligence prediction model for comprehensive mechanical properties of castings, realizes high-precision online real-time prediction of comprehensive mechanical properties of product castings, and solves the problem that traditional methods cannot numerically and intelligently predict the comprehensive mechanical properties of castings in real time with high precision.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a high-precision numerical intelligence real-time prediction method for coupling multi-factor comprehensive mechanical properties of castings, comprising the following steps: S1, construct a material process-solidification cooling rate and sequence digital twin quantitative relationship model under defect-free conditions and a material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model, determine the coefficients of each formula in the two digital twin quantitative relationship models based on the physical twin, construct a defect-mechanical property quantitative relationship model of the material under the condition of having a defect area based on the physical twin and the digital twin, combine the machine learning method to establish a high-precision numerical and intelligent prediction model of the corresponding material casting overall comprehensive mechanical property, realize the prediction of the mechanical property of the casting, and the mechanical property of the casting includes yield strength, tensile strength, elongation, and Young's modulus; S2, design and manufacture a characteristic standard casting containing both a defect-free area and a multi-level defect area, compare the mechanical property prediction obtained from the casting overall comprehensive mechanical property high-precision numerical and intelligent prediction model with the mechanical property measured data obtained from the physical twin, and correct the casting overall comprehensive mechanical property high-precision numerical and intelligent prediction model; S3, establish a product casting digital twin model, manufacture a product casting, collect the solidification cooling rate and defect data of the product casting in real time through the physical twin, and import the collected data into the casting overall comprehensive mechanical property high-precision numerical and intelligent prediction model in real time to predict the overall comprehensive mechanical property of the product casting in real time.

[0006] Further, in step S1, the specific process includes the following steps: S11, construct a material process-solidification cooling rate and sequence digital twin quantitative relationship model under defect-free conditions, collect the solidification cooling rate and solidification sequence of the material, and the material process-solidification cooling rate and sequence digital twin quantitative relationship model is defined as: ; ; Wherein, is the solidification cooling rate of the material in different areas, is the thermal conductivity of the metal, ρ is the density of the metal, is the second derivative of , is the specific heat capacity of the metal, h is the interface heat transfer coefficient, is the characteristic length, is the material temperature, T mold is the mold temperature, is an array of the solidification cooling rate of the material in the area from large to small; S12, construct a digital twin quantitative relationship model of material solidification cooling rate and sequence-mechanical properties under defect-free conditions to obtain the structure data of the material and the mechanical properties of the material under defect-free conditions, wherein the structure data includes microstructure and microcomponent difference; S13, determine the formula coefficients in the digital twin quantitative relationship model of material process-solidification cooling rate and sequence under defect-free conditions and the digital twin quantitative relationship model of material solidification cooling rate and sequence-mechanical properties by physical twinning the process parameters, solidification cooling rate, structure data and mechanical properties of the material under defect-free conditions; S14, based on the data obtained by physical twinning, fit a quantitative relationship of the defect, tensile strength and elongation data of the defect area of the material under defect conditions, and then construct a quantitative relationship model of defect-mechanical properties of the defect area of the material under defect conditions based on digital twinning, and obtain the defect and mechanical property data of the defect area of the material under defect conditions by combining the model constructed in steps S11-S13 and the yield strength and Young's modulus data, wherein the defect data includes shrinkage, shrinkage, hot crack and inclusion; S15, combine the machine learning method to sort the importance of each structure data of the material and the defect data of the defect area to the comprehensive mechanical properties of the casting, and based on the sorting result, couple the digital twin quantitative relationship model of material process-solidification cooling rate and sequence under defect-free conditions, the digital twin quantitative relationship model of material solidification cooling rate and sequence-mechanical properties, and the quantitative relationship model of defect-mechanical properties of the defect area of the material under defect conditions to intelligently construct a high-precision digital intelligent prediction model of the comprehensive mechanical properties of the casting.

[0007] Further, in step S12, the specific process includes the following steps: S121, obtain the microstructure of the material, including grain size and secondary dendrite arm spacing : ; ; In the formula, is a proportional constant, is the solidification cooling rate of the material, , is an exponential constant, is a proportional constant representing the solidification characteristics of the material; S122, calculate the material composition concentration to obtain the microcomponent difference of the material: ; wherein, is the concentration of the i-th material component at a certain solid phase position, i is the concentration of the i-th material component at a certain solid phase position, For allocation coefficients, For the first i Initial concentration of the material This represents the total time it takes for the material temperature to rise from the liquidus to the solidus. For material coefficients, The liquidus temperature of the material. The solidus temperature of the material; S123. Obtain the yield strength of the material. : ; in, This represents the total number of different types of components in the material. For the first i The influence coefficient of the content of a certain component on the yield strength of the material matrix. For the first i The contribution of the standard content of each component to the yield strength of the material matrix. For material constants, The damage coefficient of the yield strength is the secondary dendrite arm spacing. The sensitivity coefficient of the secondary dendrite arm spacing to the yield strength; S124. Obtain the tensile strength of the material. : ; in, For the first i The influence coefficient of the content of a certain component on the tensile strength of the material matrix. For the first i The contribution of the standard content of each component to the tensile strength of the material matrix. For material constants, The secondary dendrite arm spacing is the damage coefficient against tensile strength. The secondary dendrite arm spacing is a sensitivity coefficient to tensile strength. S125, Obtaining the elongation of the material : ; in, For the first i The influence coefficient of the content of a certain component on the elongation of the material matrix. For the first i The contribution of the standard content of each component to the elongation of the material matrix. For material constants, The factor representing the damage to elongation caused by the secondary dendrite arm spacing is denoted as . This is the sensitivity coefficient of the secondary dendrite arm spacing to the elongation. S126. Obtain the Young's modulus of the material. : ; wherein, is the Young's modulus of the material component, i is a material constant, is a damage coefficient of the secondary dendrite arm spacing on the elongation, is a sensitivity coefficient of the secondary dendrite arm spacing on the elongation.

[0008] Further, in step S14, the specific process includes the following steps: S141, a defect-mechanical property quantitative relationship model of the material with a defective area under the defective condition, in which the yield strength and the Young's modulus of the area with defects are determined by the yield strength and the Young's modulus of the material under the non-defective condition determined by the solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model of the material under the non-defective condition; S142, a defect-mechanical property quantitative relationship model of the material with a defective area under the defective condition, in which the tensile strength and the elongation of the area with defects are calculated in the following manner: ; ; wherein, is the tensile strength of the material with a defective area under the defective condition, is the tensile strength of the material under the non-defective condition, is an empirical coefficient, is the volume fraction of defects, is the elongation of the material with a defective area under the defective condition, is the elongation of the material under the non-defective condition, is the base of the natural logarithm, is an empirical coefficient.

[0009] Further, in step S15, the specific process includes the following steps: S151, combining the machine learning method, sorting the importance of the structure data of the material under the non-defective condition on the comprehensive mechanical properties of the casting, and based on the sorting result, coupling the process-solidification cooling rate and sequence digital twin quantitative relationship model of the material under the non-defective condition and the solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model of the material, intelligently constructing a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the non-defective area of the casting corresponding to the material; ​S152, combine the structure data of the defect area of the material belt under the defect condition and the importance of the defect data on the comprehensive mechanical properties of the casting based on the machine learning method, and based on the sorting result and the defect-mechanical property quantitative relationship model of the material belt defect area under the defect condition, intelligently build a high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting with the corresponding material belt defect area; S153, based on the machine learning coupling, intelligently build a high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting with the corresponding material belt defect area.

[0010] Further, in step S2, the specific process includes the following steps: S21, a U-shaped groove-shaped feature standard casting with a reinforcing rib structure is constructed, the feature standard casting has a cold speed gradient distribution covering 0.5-30K / s solidification cold speed range, and three levels of defects in the reinforcing rib part, including first-level defects with a defect size of 0-1mm, second-level defects with a defect size of 1-5mm, and third-level defects with a defect size of 5-30mm; S22, the feature standard casting is predicted by using the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting to obtain the mechanical property prediction, and the mechanical property test data of the feature standard casting is obtained by applying physical twinning, the prediction error of each mechanical property is calculated, and if the yield strength prediction error is greater than 2%, the tensile strength prediction error is greater than 4%, the elongation prediction error is greater than 6%, or the Young's modulus prediction error is greater than 2%, the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting is corrected.

[0011] Further, in step S3, the specific process includes: establishing a product casting digital twin model, constructing a product casting mold and manufacturing a product casting, arranging temperature sensors and defect sensors inside the product casting mold, collecting the solidification cold speed and sequence and defect data of the product casting during the casting process in real time, and inputting the collected data to the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting in real time, and predicting the comprehensive mechanical properties of the product casting in real time.

[0012] By the above technical solution, the application provides a multi-factor coupling casting comprehensive mechanical property high-precision numerical intelligence real-time prediction method, which at least has the following beneficial effects: (1) The present application constructs the quantitative relationship models of material "process-solidification cooling rate and sequence", "solidification cooling rate and sequence-mechanical properties" and "defects-mechanical properties" by digital and physical twins, combining machine learning to analyze the importance of each influencing factor of mechanical properties, constructs a high-precision numerical and intelligent prediction model of comprehensive mechanical properties of castings with microstructure, microcomponent differences and defects as multi-factor input data, and yield strength, tensile strength, elongation and Young's modulus as multi-factor output data, which supports high-reliability, rapid lightweight and high-precision numerical and intelligent design and manufacturing of castings; (2) The present application constructs the quantitative relationship models of material "process-solidification cooling rate and sequence", "solidification cooling rate and sequence-mechanical properties" and "defects-mechanical properties" based on the multi-factor variable data obtained by physical twin in the whole process of casting manufacturing, realizes high-precision prediction of comprehensive mechanical properties of castings in defect-free and defect areas under different solidification cooling rates and sequences by combining digital and physical experimental twins in the whole process; (3) The present application realizes wide-range accurate verification and correction of high-precision numerical and intelligent prediction model of comprehensive mechanical properties of castings by quantitatively designing "U-shaped groove with rib structure" characteristic standard castings, so that the characteristic standard castings have comprehensive coverage of solidification cooling rate of 0.5-30 K / s and can generate multi-level defect areas in the castings, and then the comprehensive mechanical properties data of castings obtained by model prediction and physical twin are compared, thereby realizing wide-range accurate verification and correction of high-precision numerical and intelligent prediction model of comprehensive mechanical properties of castings, and obtaining high-precision comprehensive mechanical properties prediction model of castings; (4) The present application realizes online real-time in-situ regulation of casting process parameters in the production process by real-time monitoring and collecting solidification cooling rate and sequence and defect data in the casting process, and importing the solidification cooling rate and sequence and defect data into the high-precision numerical and intelligent prediction model of comprehensive mechanical properties of castings to realize online real-time high-precision prediction of comprehensive mechanical properties of castings. BRIEF DESCRIPTION OF DRAWINGS

[0013] The drawings described herein are used to provide further understanding of the present application, and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application, and do not constitute improper limitations on the present application. In the drawings: Figure 1 It is a flow chart of the present application of high-precision numerical and intelligent prediction model and real-time prediction method of comprehensive mechanical properties of castings; Figure 2 It is the design of characteristic standard castings in the present application; Figure 3 It is three multi-level defect areas of characteristic standard castings in the present application; Figure 4 It is the selection of areas for calculating prediction error in the characteristic standard castings in the present application; Figure 5For real-time monitoring of product casting defects in the present application; Figure 6 For the quantitative relationship between yield strength and cooling rate of the model in Example 1 of the present application; Figure 7 For the global yield strength prediction result of the product casting in Example 1 of the present application; Figure 8 For the selection of the mechanical property test area of the product casting in Examples 1-4 of the present application; Figure 9 For the quantitative relationship between tensile strength and cooling rate of the model in Example 2 of the present application; Figure 10 For the global tensile strength prediction result of the product casting in Example 2 of the present application; Figure 11 For the quantitative relationship between elongation and cooling rate of the model in Example 3 of the present application; Figure 12 For the global elongation prediction result of the product casting in Example 3 of the present application; Figure 13 For the quantitative relationship between the material Young's modulus and the microcomponent difference of the model in Example 4 of the present application; Figure 14 For the global Young's modulus prediction result of the product casting in Example 4 of the present application. DETAILED DESCRIPTION

[0014] In order to make the above-mentioned purposes, features and advantages of the present application more obvious, specific embodiments will be further described in detail below with reference to the accompanying drawings and specific embodiments. The realization process of how to apply technical means to solve technical problems and achieve technical effects can be fully understood and implemented by the applicant.

[0015] Those of ordinary skill in the art can understand that all or part of the steps in the above-mentioned embodiment methods can be completed by programs instructing related hardware, therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program codes.

[0016] Please refer to Figure 1 The embodiment of the present application provides a coupled multi-factor casting global comprehensive high-precision numerical intelligence real-time prediction method, which realizes high-precision online real-time prediction of the global comprehensive mechanical properties of the product casting by combining machine learning to construct and correct the high-precision numerical intelligence prediction model of the global comprehensive mechanical properties of the casting. S1. Construct a quantitative digital twin model of the material process-solidification cooling rate and sequence under defect-free conditions and a quantitative digital twin model of the material solidification cooling rate and sequence-mechanical properties. Based on the physical twin, determine the coefficients of each formula in the two constructed quantitative digital twin models. Based on the physical twin and digital twin, construct a quantitative model of the defect-mechanical properties of the material with defective regions under defective conditions. Combine machine learning methods to establish a high-precision digital intelligent prediction model of the comprehensive mechanical properties of the corresponding material casting, so as to predict the mechanical properties of the casting, including yield strength, tensile strength, elongation, and Young's modulus.

[0017] S11. Construct a quantitative digital twin model of the relationship between material processing and solidification cooling rate and sequence under defect-free conditions, and collect the solidification cooling rate and solidification sequence of the material. The quantitative digital twin model of the relationship between material processing and solidification cooling rate and sequence is defined as shown in Formulas 1 and 2: Formula 1 Formula 2 in, To determine the solidification cooling rate of materials in different regions, For the thermal conductivity of metals, ρ For metal density, for The second derivative, Specific heat capacity of the metal h The interfacial heat transfer coefficient, For characteristic length, For material temperature, T mold For mold temperature, An array sorted by the solidification cooling rate of materials within the region from largest to smallest.

[0018] S12. Construct a digital twin quantitative relationship model of material solidification cooling rate and sequence under defect-free conditions to obtain the material's structural data and mechanical properties under defect-free conditions. The structural data includes microstructure and micro-composition differences. As a preferred embodiment of step S12, the specific process includes the following steps: S121. Based on Formulas 3 and 4, obtain the microstructure of the material, including grain size. and secondary dendrite wall spacing : Formula 3 Formula 4 In the formula The proportionality constant is... The solidification cooling rate of the material. , It is an exponential constant. A proportionality constant characterizing the solidification properties of a material; S122. Based on Formula 5, calculate the material composition concentration to obtain the microscopic compositional differences of the material: Formula 5 in, For a certain solid phase position, the first i The concentration of the material components, For allocation coefficients, For the first i The initial concentration of the material This represents the total time it takes for the material temperature to rise from the liquidus to the solidus. For material coefficients, The liquidus temperature of the material. The solidus temperature of the material; S123. Based on Formula 6, obtain the material yield strength. : Formula Six in, This represents the total number of different types of components in the material. For the first i The influence coefficient of the content of a certain component on the yield strength of the material matrix. For the first i The contribution of the standard content of each component to the yield strength of the material matrix. For material constants, The damage coefficient of the yield strength is the secondary dendrite arm spacing. The sensitivity coefficient of the secondary dendrite arm spacing to the yield strength; S124. Based on Formula 7, obtain the tensile strength of the material. : Formula 7 in, For the first i The influence coefficient of the content of a certain component on the tensile strength of the material matrix. For the first i The contribution of the standard content of each component to the tensile strength of the material matrix. For material constants, The secondary dendrite arm spacing is the damage coefficient against tensile strength. The secondary dendrite arm spacing is a sensitivity coefficient to tensile strength. S125. Obtain the elongation of the material based on Formula 8. : Formula 8 in, For the first i The influence coefficient of the content of a certain component on the elongation of the material matrix. For the first i The contribution of the standard content of each component to the elongation of the material matrix. For material constants, The factor representing the damage to elongation caused by the secondary dendrite arm spacing is denoted as . This is the sensitivity coefficient of the secondary dendrite arm spacing to the elongation. S126. Based on Formula 9, obtain the Young's modulus of the material. : Formula Nine in, For the first i Young's modulus of a material composition. For material constants, The factor representing the damage to elongation caused by the secondary dendrite arm spacing is denoted as . This is the sensitivity coefficient of the secondary dendrite arm spacing to the elongation.

[0019] S13. Using multiple factors such as process parameters, solidification cooling rate in the solid-liquid phase range, microstructure, microstructure differences, yield strength, tensile strength, elongation, and Young's modulus of the material under defect-free physical twin conditions, determine the coefficients of each formula in the digital twin quantitative relationship model of material process-solidification cooling rate and sequence and the digital twin quantitative relationship model of material solidification cooling rate and sequence-mechanical properties under defect-free conditions. S14. Since the quantitative impact of defects on tensile strength and elongation is unclear, firstly, physical twinning is used to obtain data on defects, tensile strength, and elongation in the defective region of the material under defective conditions. Based on the data obtained from the physical twinning, a quantitative relationship is fitted. Then, a quantitative relationship model of defects-mechanical properties in the defective region of the material under defective conditions is constructed based on digital twinning. Combined with the model constructed in steps S11-S13 and the obtained data, the defect and mechanical property data of the defective region of the material under defective conditions are obtained. The defect data includes shrinkage cavities, shrinkage porosity, thermal cracks, and inclusions. As a preferred embodiment of step S14, the specific process includes the following steps: S141, the defect-mechanical property quantitative relationship model of the material with defective region under defective conditions, wherein the yield strength and Young's modulus of the defective region in the model are determined by the material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model under defect-free conditions. S142. The quantitative relationship model between defects and mechanical properties of the material with defective regions under defective conditions, wherein the tensile strength and elongation of the defective regions are calculated in the model as shown in Formulas 10 and 11: Equation Ten Equation Eleven wherein, is the tensile strength of the defective region of the material with defects, is the tensile strength of the material without defects, is an empirical coefficient, is the volume fraction of defects, is the elongation of the defective region of the material with defects, is the elongation of the material without defects, is the base of the natural logarithm, is an empirical coefficient.

[0020] S15, in combination with the machine learning method, the importance of the structure data of the material and the defect data of the defective region to the comprehensive mechanical properties of the casting is sorted, and based on the sorting result, the process-solidification cooling rate and sequence digital twin quantitative relationship model of the material without defects, the material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model, and the defect-mechanical property quantitative relationship model of the defective region of the material with defects are coupled, and the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting is intelligently constructed.

[0021] As a preferred embodiment of step S15, the specific process includes the following steps: S151, in combination with the machine learning method, the importance of the microstructure and microcomponent difference of the material to the comprehensive mechanical properties of the casting is sorted, and based on the sorting result, the process-solidification cooling rate and sequence digital twin quantitative relationship model of the material without defects and the material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model are coupled, and the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting without defects corresponding to the material is intelligently constructed; S152, in combination with the machine learning method, the importance of the structure data of the material and the defect data of the defective region to the comprehensive mechanical properties of the casting is sorted, and based on the sorting result and the defect-mechanical property quantitative relationship model of the defective region of the material with defects, the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting with defective region corresponding to the material is intelligently constructed; S153, based on the machine learning coupling of the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting without defects corresponding to the material and the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting with defective region corresponding to the material, the high-precision intelligent prediction model of the comprehensive mechanical properties of the casting corresponding to the material is intelligently constructed.

[0022] S2, design and manufacture a feature standard casting containing both a defect-free zone and a multi-level defect zone, and according to the comparison results of the mechanical property prediction obtained by the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting and the mechanical property test data obtained by physical twinning, the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting is corrected; As a preferred embodiment of step S2, the specific process includes the following steps: S21, a U-shaped groove-shaped feature standard casting with a reinforcing rib structure is constructed, the design drawing of the feature standard casting is as shown in Figure 2 The overall axial total length is 600 mm, the transverse width is 62.5 mm, and the vertical height is 61 mm; the two ends are symmetrically configured with 25 mm straight segments and are connected by R10 transition fillets to form a torsional reinforcement zone; a U-shaped weight-reducing groove is arranged in the middle of the casting, the groove wall thickness is 2.5 mm, and the groove is symmetrically distributed on both sides; the groove bottom adopts an R10 fillet and an R7.5 fillet to form a double-curvature coupling structure, constituting a main stress dissipation zone; a symmetrically distributed reinforcing rib is arranged every 150 mm axially to enhance the bending resistance of the casting; the feature standard casting has a cold speed gradient distribution covering the 0.5-30K / s solidification cold speed range, and a defect area with three levels of defects is generated at the reinforcing rib position, as shown in Figure 3 The defect area includes first-level defects with a defect size of 0-1 mm, second-level defects with a defect size of 1-5 mm, and third-level defects with a defect size of 5-30 mm, wherein the cold speed gradient is used to cover the typical solidification working conditions of the material solidification cold speed of 0.5-30K / s, the defect-free area is used to verify the prediction accuracy of the cold speed main line model, and the defect area is used to verify the reliability of the correction effect of the "defect-performance" model, thereby realizing the double verification of the main line and the correction path of the numerical and intelligent prediction model. S22, the feature standard casting is predicted by the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting, and the mechanical property data of the feature standard casting is obtained by physical twinning, the prediction error of each mechanical property is calculated, and if the yield strength prediction error is >2%, the tensile strength prediction error is >4%, the elongation prediction error is >6%, or the Young's modulus prediction error is >2%, the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting is corrected. The area for calculating the prediction error in the feature standard casting is selected as shown in Figure 4 .

[0023] S3, a product casting digital twin model is established, a product casting is manufactured, the solidification cold speed and defect data of the product casting are monitored and collected in real time by physical twinning, and the collected data is imported into the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting in real time, and the comprehensive mechanical properties of the product casting are predicted in real time and with high precision.

[0024] As a preferred embodiment of step S3, it specifically comprises: establishing a product casting digital twin model, constructing a product casting mold and manufacturing a product casting, the product casting mold is internally laid with temperature sensors and defect sensors, and the solidification cold speed and sequence and defect data in the product casting casting process are collected online in real time, wherein the real-time monitoring of the defect data is as shown in Figure 5 The collected data is input in real time to a casting global comprehensive mechanical property high-precision digital intelligent prediction model, and the global comprehensive mechanical property of the product casting is predicted in real time with high precision.

[0025] The above method will be described in detail in combination with four specific embodiments: Example 1

[0026] S1, coupling of a casting global comprehensive mechanical property high-precision digital intelligent prediction model: In this embodiment, the selected alloy is Al6.9Si0.25Mg0.1Fe0.04Mn0.15Ti, which is melted in a 12 kg capacity graphite crucible using a resistance furnace, and the melting temperature is controlled at 750°C, and the single furnace liquid mass is 8 kg. After the melt is homogenized for 1 hour, Al-10wt.% Sr intermediate alloy is added, and pure argon gas is used for rotary jet degassing (rotating speed 350 rpm, time 4 minutes), and after degassing, commercial granular flux is covered, and the casting process is carried out. During the casting process, multiple groups of sensors are laid to monitor the temperature field evolution, solidification behavior and defect formation of the casting under different process conditions in real time. After the casting is completed, the mechanical properties of the casting are tested, and the “process—solidification cold speed and sequence” and “solidification cold speed and sequence—mechanical property” digital twin quantitative relationship models of the material under the condition of no defect are constructed based on digital twin, wherein the quantitative relationship between the yield strength of the casting and the cold speed is as shown in Figure 6 The formula coefficients in the “process—solidification cold speed and sequence” and “solidification cold speed and sequence—mechanical property” digital twin quantitative relationship models of the material under the condition of no defect are determined by the data obtained by physical twin, and the “defect—mechanical property” quantitative relationship model of the material in the defect area under the condition of defect is constructed based on the data obtained by physical twin and digital twin, and the global mechanical property digital intelligent prediction model of the casting in the defect-free and defect area is intelligently constructed by coupling the “process—solidification cold speed and sequence”, “solidification cold speed and sequence—mechanical property” and “defect—mechanical property” quantitative relationship models of the material.

[0027] S2, yield strength verification of the casting global comprehensive mechanical property high-precision digital intelligent prediction model based on the characteristic standard casting: In this embodiment, a characteristic standard casting of Al6.9Si0.30Mg0.1Fe0.04Mn0.15Ti is prepared, which has a structure design with a cold speed gradient distribution of 0.5-30 K / s, and contains both defect-free zones and different levels of defect zones. During the preparation of the alloy, argon degassing (gas flow 15 L / min, time 25 min) and 750 rpm vortex stirring are combined, and Al3Ti3B grain refiner is added to improve the uniformity of the structure. Based on the established high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting, the solidification cooling rate and solidification sequence information of the three defect-free regions and the three different levels of defect regions of the casting digital twin model are extracted, and the yield strength prediction cloud map of the casting is generated. Select six measurement areas on the casting that are the same as the prediction areas of the casting digital twin model, three of which are located in the defect-free area to verify the prediction accuracy of the yield strength in the numerical and intelligent prediction model of the comprehensive mechanical properties of the casting defect-free area, and three of which are located in the multi-level defect area to verify the prediction accuracy of the yield strength in the numerical and intelligent prediction model of the comprehensive mechanical properties of the casting with defects. By comparing the predicted and measured values of the yield strength of each measurement area, and combining machine learning for error analysis and dynamic correction of model coefficients, the model is continuously iterated and optimized.

[0028] S3, Real-time prediction of the yield strength of the product casting based on the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting: In this embodiment, an Al6.9Si0.30Mg0.1Fe0.04Mn0.15Ti product casting is prepared, and the selected product casting has a complex structure and significant wall thickness variation. First, the digital twin model of the product casting is established, and then the physical twin of the product casting is set up with the following initial process parameters: melting temperature 725°C, mold temperature lower mold 205°C, upper mold 225°C, working pressure 545 mbar, holding time 28 s pressure increase + 42 s solidification, degassing time 18 min, vortex stirring speed 690 rpm. During the casting process, the solidification cooling rate and sequence and defect data of the product casting are collected in real time by K-type thermocouples and X-ray real-time imaging instruments. Based on the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting, the above digital data is input into the model to predict the yield strength of the product casting in real time, and the prediction results are shown in Figure 7 As shown in Figure 8 As shown in Example 2

[0029] S1, High-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting coupled with multiple factors: In this embodiment, the selected alloy is Al6.95Si0.25Mg0.1Fe0.04Mn0.15Ti, which is melted in a 12 kg capacity graphite crucible using an electric resistance furnace, with the melting temperature controlled at 750°C, and the single furnace melt mass being 8 kg. After the melt is homogenized for 1 hour, Al-10wt.% Sr intermediate alloy is added, and rotary argon gas jet degassing is used (rotating speed 350 rpm, time 4 minutes), after which commercial granular flux is applied, and the casting process is carried out after 10 minutes of standing. During the casting process, multiple groups of sensors are arranged to monitor the temperature field evolution, solidification behavior, and defect formation of the casting under different process conditions in real time. After the casting is completed, the mechanical properties of the casting are tested, and the “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” digital twin quantitative relationship models of the material under the condition of no defects are constructed based on digital twinning, in which the quantitative relationship between the tensile strength of the casting and the cooling rate is as shown in Figure 9 The formula coefficients in the “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” digital twin quantitative relationship models of the material under the condition of no defects are determined by the data obtained through physical twinning, and at the same time, the “defect-mechanical properties” quantitative relationship model of the material in the defective area under the condition of defects is constructed based on the data obtained through physical twinning and digital twinning. Combining machine learning with the material “process-solidification cooling rate and sequence”, “solidification cooling rate and sequence-mechanical properties”, and “defect-mechanical properties” quantitative relationship models, the global mechanical property numerical intelligence prediction model of the casting in the non-defective and defective areas is intelligently constructed.

[0030] S2, based on the characteristic standard casting, the global comprehensive mechanical property high-precision numerical intelligence prediction model of the casting is verified for tensile strength: In this embodiment, a characteristic standard casting of Al6.95Si0.25Mg0.1Fe0.04Mn0.15Ti is prepared, which has a structure design with a cold speed gradient distribution of 0.5-30 K / s, and contains both defect-free zones and different levels of defect zones. During the preparation of the alloy, argon degassing (gas flow 15 L / min, time 25 min) and 750 rpm vortex stirring are combined, and Al3Ti3B grain refiner is added to improve the uniformity of the structure. Based on the established high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting, the solidification cooling rate and solidification sequence information of the three defect-free regions and the three multi-level defect regions of the casting digital twin model are extracted, and a prediction cloud map of the tensile strength of the casting is generated. Six measurement regions are selected on the casting, which are the same as the prediction regions of the casting digital twin model, three of which are located in the defect-free region to verify the prediction accuracy of the tensile strength in the numerical and intelligent prediction model of the comprehensive mechanical properties of the casting; three are located in the multi-level defect region to verify the prediction accuracy of the tensile strength in the numerical and intelligent prediction model of the comprehensive mechanical properties of the casting with defects. By comparing the predicted and measured values of the tensile strength of each measurement region, and combining machine learning for error analysis and dynamic correction of model coefficients, the model is continuously iterated and optimized.

[0031] S3, Real-time prediction of the tensile strength of the product casting based on the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting: In this embodiment, an Al6.95Si0.25Mg0.1Fe0.04Mn0.15Ti product casting is prepared, and the selected product casting has a complex structure and a significant change in wall thickness. First, the digital twin model of the product casting is established, and then the physical twin of the product casting is set up with the following initial process parameters: melting temperature 725°C, mold temperature lower mold 205°C, upper mold 225°C, working pressure 545 mbar, holding time 28 s pressure increase + 42 s solidification, degassing time 18 min, vortex stirring speed 690 rpm. During the casting process, the solidification cooling rate and sequence and defect data of the product casting are collected in real time by K-type thermocouples and X-ray real-time imagers. Based on the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting, the above digital data is input into the model to predict the tensile strength of the product casting in real time, and the prediction results are shown in Figure 10 Six measurement regions are selected on the product casting for tensile strength testing, and the measurement regions are selected as shown in Figure 8 . Example 3

[0032] S1, Establishment of a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting coupled with multiple factors: In this embodiment, the selected alloy is Al7.0Si0.25Mg0.1Fe0.04Mn0.15Ti, which is melted in a 12 kg capacity graphite crucible using an electric resistance furnace, with the melting temperature controlled at 750°C, and the single furnace melt mass being 8 kg. After the melt is homogenized for 1 hour, Al-10wt.% Sr intermediate alloy is added, and pure argon gas is used for rotary jet degassing (rotating speed 350 rpm, time 4 minutes). After degassing, commercial granular flux is applied, and the casting process is carried out after 10 minutes of standing. During casting, multiple groups of sensors are arranged to monitor the temperature field evolution, solidification behavior, and defect formation of the casting under different process conditions in real time. After casting is completed, the mechanical properties of the casting are tested, and the “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” digital twin quantitative relationship models of the material under the condition of no defects are constructed based on digital twinning. The quantitative relationship between the elongation of the casting and the cooling rate is as shown in Figure 11 The formula coefficients in the “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” digital twin quantitative relationship models of the material under the condition of no defects are determined by the data obtained through physical twinning. At the same time, based on the data obtained through physical twinning and digital twinning, the “defect-mechanical properties” quantitative relationship model of the material with a defective region under the condition of defects is constructed, and the global mechanical property numerical intelligence prediction model of the casting with no defects and with a defective region is intelligently constructed by coupling the material “process-solidification cooling rate and sequence”, “solidification cooling rate and sequence-mechanical properties”, and “defect-mechanical properties” quantitative relationship models through machine learning.

[0033] S2, verification of the elongation of the casting global comprehensive mechanical property high-precision numerical intelligence prediction model based on the feature standard casting: In this embodiment, a characteristic standard casting of Al7.0Si0.25Mg0.1Fe0.04Mn0.15Ti is prepared, which has a structure design with a cold speed gradient distribution of 0.5-30 K / s, and contains both defect-free zones and multi-level defect zones. During the alloy preparation process, argon degassing (gas flow 15 L / min, time 25 minutes) and 750 rpm vortex stirring are combined, and Al3Ti3B grain refiner is added to improve the uniformity of the structure. Based on the established high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting, the solidification cooling rate and solidification sequence information of the three defect-free regions and the three different level defect regions of the casting digital twin model are extracted, and the casting global elongation prediction cloud map is generated. Select six measurement areas on the casting that are the same as the prediction areas of the casting digital twin model, three of which are located in the defect-free zone to verify the prediction accuracy of the elongation in the numerical intelligence prediction model of the global mechanical properties of the casting defect-free zone, and three of which are located in the multi-level defect zone to verify the prediction accuracy of the elongation in the numerical intelligence prediction model of the global mechanical properties of the casting with defects. By comparing the predicted value and the measured value of the elongation of each measurement area, and combining machine learning for error analysis and dynamic correction of model coefficients, the model is continuously iterated and optimized.

[0034] S3, Real-time prediction of the global tensile strength of the product casting based on the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting: In this embodiment, an Al7.0Si0.25Mg0.1Fe0.04Mn0.15Ti product casting is prepared, and the selected product casting has a complex structure and a significant change in wall thickness. First, the digital twin model of the product casting is established, and then the physical twin of the product casting is set up with the following initial process parameters: melting temperature 725°C, mold temperature lower mold 205°C, upper mold 225°C, working pressure 545 mbar, holding time 28 s pressure increase + 42 s solidification, degassing time 18 min, vortex stirring speed 690 rpm. During the casting process, the solidification cooling rate and sequence and defect data of the product casting are collected in real time by K-type thermocouples and X-ray real-time imagers. Based on the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting, the above digital data is input into the model to predict the elongation of the product casting in real time, and the prediction results are shown in Figure 12 As shown in Figure 8 As shown in Embodiment 4

[0035] S1, High-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting coupled with multiple factors: In this embodiment, the selected alloy is Al7.1Si0.25Mg0.1Fe0.04Mn0.15Ti, which is melted in a 12 kg capacity graphite crucible using an electric resistance furnace, with the melting temperature controlled at 750°C, and the single furnace melt mass being 8 kg. After the melt is homogenized for 1 hour, Al-10wt.% Sr intermediate alloy is added, and pure argon gas is used for rotary jet degassing (rotating speed 350 rpm, time 4 minutes). After degassing, commercial granular flux is applied, and the casting process is carried out after 10 minutes of standing. During the casting process, multiple groups of sensors are arranged to monitor the temperature field evolution, solidification behavior, and defect formation of the casting under different process conditions in real time. After the casting is completed, the mechanical properties of the casting are tested, and the quantitative relationship models of “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” of the material under the condition of no defects are constructed based on digital twinning, wherein the quantitative relationship of the Young's modulus and the microcomponent difference of the casting material is as shown in Figure 13 The formula coefficients in the quantitative relationship models of “process-solidification cooling rate and sequence” and “solidification cooling rate and sequence-mechanical properties” of the material under the condition of no defects are determined by the data obtained through physical twinning. At the same time, based on the data obtained through physical twinning and digital twinning, the quantitative relationship model of “defect-mechanical properties” of the material in the defective area under the condition of defects is constructed, and the global mechanical property numerical intelligence prediction model of the casting in the non-defective and defective areas is intelligently constructed by coupling the material “process-solidification cooling rate and sequence”, “solidification cooling rate and sequence-mechanical properties”, and “defect-mechanical properties” quantitative relationship models through machine learning.

[0036] S2, verification of the Young's modulus of the high-precision numerical intelligence prediction model of the global comprehensive mechanical properties of the casting based on the coupling of the feature standard casting and multiple factors In this embodiment, a characteristic standard casting of Al7.1Si0.30Mg0.11Fe0.05Mn0.15Ti is prepared, which has a structure design with a cold speed gradient distribution of 0.5-30 K / s, and contains both defect-free zones and multi-level defect zones. During the alloy preparation process, argon degassing (gas flow 15 L / min, time 25 minutes) and 750 rpm vortex stirring are combined, and Al3Ti3B grain refiner is added to improve the uniformity of the structure. Based on the established high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting, the solidification cooling rate and solidification sequence information of the three defect-free zones and three different level defect zones of the casting digital twin model are extracted, and the prediction cloud map of the Young's modulus of the casting is generated. Six measurement areas are selected on the casting, which are the same as the prediction areas of the casting digital twin model, three of which are located in the defect-free zone to verify the prediction accuracy of the Young's modulus in the numerical intelligence prediction model of the comprehensive mechanical properties of the casting defect-free zone, and the other three are located in the multi-level defect zone to verify the prediction accuracy of the Young's modulus in the numerical intelligence prediction model of the comprehensive mechanical properties of the casting with defects. By comparing the predicted and measured values of the Young's modulus of each measurement area, and combining machine learning for error analysis and dynamic correction of model coefficients, the model is continuously iteratively optimized.

[0037] S3, Real-time prediction of the tensile strength of the product casting based on the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting In this embodiment, a product casting of Al7.1Si0.25Mg0.1Fe0.04Mn0.15Ti is prepared, and the selected product casting has a complex structure and a significant change in wall thickness. First, the digital twin model of the product casting is established, and then the physical twin of the product casting is set up with the following initial process parameters: melting temperature 725°C, mold temperature lower mold 205°C, upper mold 225°C, working pressure 545 mbar, holding time 28 s pressure increase + 42 s solidification, degassing time 18 min, vortex stirring speed 690 rpm. During the casting process, the solidification cooling rate and sequence and defect data of the product casting are collected in real time by K-type thermocouples and X-ray real-time imaging instruments. Based on the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting, the above digital data is input into the model to predict the Young's modulus of the product casting in real time, and the prediction results are shown in Figure 14 As shown in Figure 8 As shown in

[0038] Table 1. Comparison of mechanical property data of defect-free zones and multi-level defect zones obtained by high-precision numerical intelligence prediction model of comprehensive mechanical properties of casting and physical twin of characteristic standard casting in Examples 1-4 ;

[0039] Table 2. Percentage error of predicted and actual values of mechanical properties of feature standard castings in Examples 1-4 ; Among them, the measurement area 1-4 is the mechanical property prediction and measured value data of the feature standard casting defect-free area, and the measurement area 4-6 is the mechanical property prediction and measured value data of the feature standard casting multi-level defect area.

[0040] According to the results shown in Table 1 and Table 2, the yield strength prediction value and actual value deviation range of the feature standard casting of the present application is 0.29-4.78 MPa, the prediction error is ≤2%; the tensile strength prediction value and actual value deviation range is 0.41-13.2 MPa, the prediction error is ≤4%; the elongation prediction absolute value and actual absolute value deviation range is 0.01-0.3%, the prediction error is ≤6%; the Young's modulus prediction value and actual value range is 0.7-2.5 GPa, the prediction error is ≤2%, all within the error range, confirming the high precision and reliability of the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting.

[0041] Table 3. Comparison results of predicted values of high-precision numerical intelligence prediction model of comprehensive mechanical properties of casting of product castings in Examples 1-4 and physical twin measured values of six measurement areas ; Table 4. Percentage error of predicted and actual values of mechanical properties of product castings in Examples 1-4

[0042] According to the results shown in Table 3 and Table 4, the yield strength prediction value and actual value deviation range of the product casting of the present application is 0.6-4.1 MPa, the prediction error is ≤2%; the tensile strength prediction value and actual value deviation range is 0.6-12.3 MPa, the prediction error is ≤4%; the elongation prediction absolute value and actual absolute value deviation range is 0.01-0.55%, the prediction error is ≤6%; the Young's modulus prediction value and actual value range is 0.1-0.8 GPa, the prediction error is ≤2%, all within the error range, verifying the high precision and reliability of the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the casting proposed by the present application, which can realize high-reliability, fast-lightweight and high-precision numerical intelligence design of the casting.

[0043] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples without contradiction.

[0044] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be specifically embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch instructions from the instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instruction execution systems, apparatus or devices.

[0045] The above embodiments have been described in detail, and the principles and embodiments of the present application have been described herein by applying specific examples. The above examples are only used to help understand the method of the present application and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific embodiments and application scope will be changed; in view of the above, the content of the specification should not be understood as a limitation of the present application.

Claims

1. A coupled multi-factor casting global synthesis force high-precision digital real-time prediction method, characterized in that, Comprise the following steps: S1, constructing a material process-solidification cooling rate and sequence digital twin quantitative relationship model under defect-free conditions and a material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model, and determining the formula coefficients in the two digital twin quantitative relationship models based on physical twinning, constructing a defect-mechanical property quantitative relationship model of a material with a defect area under defect conditions based on physical twinning and digital twinning, combining machine learning method to establish a high-precision numerical and intelligent prediction model of comprehensive mechanical properties of the corresponding material casting, and realizing the prediction of the mechanical properties of the casting, the mechanical properties of the casting including yield strength, tensile strength, elongation, Young's modulus; S2, designing and manufacturing a characteristic standard casting containing both a defect-free area and a multi-level defect area, comparing the mechanical property prediction obtained from the high-precision numerical and intelligent prediction model of comprehensive mechanical properties of the casting with the mechanical property test data obtained from the physical twinning, and correcting the high-precision numerical and intelligent prediction model of comprehensive mechanical properties of the casting; S3, establishing a product casting digital twin model, manufacturing a product casting, collecting the solidification cooling rate and sequence and defect data of the product casting in real time through physical twinning, and importing the collected data into the high-precision numerical and intelligent prediction model of comprehensive mechanical properties of the casting in real time to predict the comprehensive mechanical properties of the product casting in real time.

2. The method of claim 1, wherein the method is characterized by: The specific process of step S1 comprises the following steps: S11, constructing a material process-solidification cooling rate and sequence digital twin quantitative relationship model under defect-free conditions, collecting the solidification cooling rate and solidification sequence of the material, and the material process-solidification cooling rate and sequence digital twin quantitative relationship model is defined as: ； ； in, To determine the solidification cooling rate of materials in different regions, For the thermal conductivity of metals, ρ For metal density, for The second derivative, Specific heat capacity of the metal h The interfacial heat transfer coefficient, For characteristic length, For material temperature, T mold For mold temperature, An array sorted by the solidification cooling rate of materials within the region from largest to smallest; S12, constructing a material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model under defect-free conditions, obtaining the structure data and mechanical properties of the material under defect-free conditions, and the structure data includes microstructure and microcomponent difference; S13, determining the formula coefficients in the material process-solidification cooling rate and sequence digital twin quantitative relationship model and the material solidification cooling rate and sequence-mechanical property digital twin quantitative relationship model under defect-free conditions through the physical twinning of the process parameters, solidification cooling rate, structure data and mechanical properties of the material under defect-free conditions; S14, obtaining the quantitative relationship based on the data obtained from the physical twinning of the defect, tensile strength and elongation data of the material with a defect area under defect conditions, and then constructing a defect-mechanical property quantitative relationship model of a material with a defect area under defect conditions based on digital twinning, and obtaining the defect and mechanical property data of the material with a defect area under defect conditions based on the model constructed in steps S11-S13 and the yield strength and Young's modulus data, the defect data including shrinkage, shrinkage, hot crack and inclusion; S15, sort the importance of the structure data of the material and the defect data of the defect area on the comprehensive mechanical properties of the casting by combining a machine learning method, and intelligently build a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting under the defect-free condition of the corresponding material based on the sorting results and the quantitative relationship model of the material process-solidification cooling rate and sequence under the defect-free condition and the quantitative relationship model of the material solidification cooling rate and sequence-mechanical properties.

3. The method of claim 2, wherein the method is characterized by: The specific process of step S12 includes the following steps: S121, obtaining microstructure of the material, including grain size and secondary dendrite arm spacing : ； ； wherein is a proportionality constant, is the solidification cooling rate of the material, , is an exponential constant, is a proportionality constant characterizing the solidification properties of the material; S122, calculate the material composition concentration to obtain the micro-component difference of the material: ； in, For a certain solid phase position, the first i The concentration of the material components, For allocation coefficients, For the first i Initial concentration of the material This represents the total time it takes for the material temperature to rise from the liquidus to the solidus. For material coefficients, The liquidus temperature of the material. The solidus temperature of the material; S123, obtaining material yield strength : ； wherein, is the total number of component types in the material, is the influence coefficient of the content of the i component on the yield strength of the material matrix, is the contribution value of the standard content of the i component to the yield strength of the material matrix, is a material constant, is the damage coefficient of the secondary dendrite arm spacing on the yield strength, is the sensitivity coefficient of the secondary dendrite arm spacing on the yield strength. S124, obtaining tensile strength of the material : ； in, For the first i The influence coefficient of the content of a certain component on the tensile strength of the material matrix. For the first i The contribution of the standard content of each component to the tensile strength of the material matrix. For material constants, The secondary dendrite arm spacing is the damage coefficient against tensile strength. The secondary dendrite arm spacing is a sensitivity coefficient to tensile strength. S125, obtaining the elongation of the material : ； wherein, is the coefficient of the influence of the content of the first component on the elongation of the material matrix, i is the coefficient of the influence of the content of the first component on the elongation of the material matrix, is the contribution value of the standard content of the first component to the elongation of the material matrix, i is the contribution value of the standard content of the first component to the elongation of the material matrix, is the material constant, is the damage coefficient of the secondary dendrite arm spacing on the elongation, is the sensitivity coefficient of the secondary dendrite arm spacing on the elongation. S126, obtaining Young's modulus of the material : ； in, For the first i Young's modulus of a material composition. For material constants, The factor representing the damage to elongation caused by the secondary dendrite arm spacing is denoted as . This is the sensitivity coefficient of the secondary dendrite arm spacing to the elongation.

4. The method of claim 3, wherein the method is characterized by: The specific process of step S14 includes: S141, the quantitative relationship model of the defect-mechanical properties of the material with the defect area under the defect condition, in which the yield strength and Young's modulus of the defect area are determined by the yield strength and Young's modulus of the material under the defect-free condition determined by the quantitative relationship model of the material solidification cooling rate and sequence-mechanical properties under the defect-free condition; S142, the quantitative relationship model of the defect-mechanical properties of the material with the defect area under the defect condition, in which the calculation method of the tensile strength and elongation of the defect area is: ； ； wherein, is the tensile strength of the material with the defect region, is the tensile strength of the material without the defect region, is an empirical coefficient, is the volume fraction of the defect, is the elongation of the material with the defect region, is the elongation of the material without the defect region, is the base of the natural logarithm, is an empirical coefficient.

5. The method of claim 2, wherein the method is characterized by: The specific process of step S15 includes the following steps: S151, sort the importance of the structure data of the material under the defect-free condition on the comprehensive mechanical properties of the casting by combining a machine learning method, and intelligently build a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting under the defect-free condition of the corresponding material based on the sorting results and the quantitative relationship model of the material process-solidification cooling rate and sequence under the defect-free condition and the quantitative relationship model of the material solidification cooling rate and sequence-mechanical properties under the defect-free condition; S152, sort the importance of the structure data and defect data of the material with the defect area under the defect condition on the comprehensive mechanical properties of the casting by combining a machine learning method, and intelligently build a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting with the defect area of the corresponding material based on the sorting results and the quantitative relationship model of the defect-mechanical properties of the material with the defect area under the defect condition; S153, intelligently build a high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting of the corresponding material based on the machine learning coupling of the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting under the defect-free condition of the corresponding material and the high-precision numerical and intelligent prediction model of the comprehensive mechanical properties of the casting with the defect area of the corresponding material.

6. The method of claim 1, wherein the method is a coupled multi-factor casting global synthesis force high-precision digital real-time prediction method. The specific process of step S2 includes the following steps: S21, build a U-shaped groove-shaped feature standard casting with a reinforcing rib structure, the feature standard casting has a cold speed gradient distribution covering 0.5-30K / s solidification cooling speed range, and three levels of defect areas are generated at the reinforcing rib part, including first-level defects with a defect size of 0-1mm, second-level defects with a defect size of 1-5mm, and third-level defects with a defect size of 5-30mm; S22, using the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the whole domain of the casting to predict the mechanical properties of the characteristic standard casting, and using physical twinning to obtain the measured data of the mechanical properties of the characteristic standard casting, calculating the prediction error of each mechanical property, if the prediction error of the yield strength is > 2%, the prediction error of the tensile strength is > 4%, the prediction error of the elongation is > 6% or the prediction error of the Young's modulus is > 2%, the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the whole domain of the casting is corrected.

7. The method of claim 1, wherein the method is a coupled multi-factor casting global synthesis force high-precision digital real-time prediction method. The specific process of step S3 includes: establishing a digital twin model of the product casting, constructing a product casting mold and manufacturing a product casting, arranging temperature sensors and defect sensors inside the product casting mold, collecting solidification cooling rate, sequence and defect data in the casting process of the product casting in real time online, and inputting the collected data into the high-precision numerical intelligence prediction model of the comprehensive mechanical properties of the whole domain of the casting in real time, and predicting the comprehensive mechanical properties of the whole domain of the product casting in real time online.

Images