Turbine disc die forging process reasoning method and system

Abstract

The invention discloses a turbine disc die forging process reasoning method and system. The method comprises the following steps that (1) key points, needing to be subjected to organization control, of a turbine disc are selected; (2) through finite element simulation in the whole forging process of the turbine disc, key point deformation information is extracted; (3) establishing a target optimization function, respectively proposing requirements for the average grain size and recrystallization fraction of the key points, and setting a corresponding weight for the final structure target of each key point; (4) setting a boundary value and a penalty function of process parameters according to the actual forging condition, and starting to execute the penalty function when the process obtained by reasoning exceeds the boundary; and (5) embedding the target optimization function and the penalty function into a multi-target optimization algorithm, and reasoning a plurality of forging process parameters at the same time. According to the method, the effect of optimizing the multi-region grain structure of the forge piece can be achieved, the die is effectively protected, and a theoretical basis and technical support are provided for high-quality forging.

Description

Technical Field

[0001] The invention belongs to the technical field of forging, and relates to a turbine disc die forging process reasoning method and system. Background Art

[0002] The multi-regional performance of modern aero-engine turbine disks means that different regions of the parts correspond to different microstructure distributions, which puts higher requirements on its forming process. To this end, combining the model proposed by the above-mentioned research institute with the mature research results of many scholars, using the existing multi-objective optimization methods, an effective method is established to optimize the die forging process parameters of turbine disks, effectively control the evolution of microstructures in multiple key areas, so that the final forgings can obtain multiple properties, improve the reliability and service life of parts, and meet the urgent quality and output requirements in the engineering field.

[0003] In view of the above difficulties, the present invention proposes a process reasoning method based on the target microstructure of the key area for the die forging process of the turbine disk, and builds a system for executing the method. The method establishes a target optimization function and a penalty function that describe the microstructure target according to the target organization requirements of the key area, and infers the process parameters of the turbine disk die forging process based on the multi-objective optimization algorithm. Finally, the process parameters that can be adjusted according to the target microstructure of the key area are inferred. The invention and application of this method have important technical guiding significance for the rational formulation of the die forging process.

[0004] The patent specification with publication number CN118735067A discloses a method for predicting the quality of turbine disk die forging based on feature fusion and ensemble learning. This method can only predict the quality performance parameters of turbine disk die forging based on forging data, but cannot infer more appropriate process parameters of the forging process.

[0005] The patent specification with publication number CN118734701A discloses a method for optimizing process parameters of turbine disk die forging based on multi-objective optimization problem modeling. This method lacks full-process simulation of forging and an optimization penalty mechanism, and cannot fully consider problems that may occur during the forging process.

[0006] The patent specification with publication number CN108515132B discloses a hot die forging process for a nickel-based deformed high-temperature alloy dual-performance turbine disk. This method only involves nickel-based deformed high-temperature alloy materials and cannot meet the versatility of the process. Summary of the invention

[0007] The purpose of the present invention is to provide a die forging process trajectory planning method and system. The method can reasonably plan the process trajectory in the die forging process, achieve the effect of refining the grain structure of the forging, effectively improve the quality of the forging, solve the problem that the existing die forging process is difficult to meet the requirements of modern industrial production, and build a system for executing the method to facilitate the implementation of the method.

[0008] In order to achieve the above object, the technical solution adopted by the present invention is:

[0009] 1. A process reasoning method and system for a turbine disk die forging, characterized in that the method can reasonably reason the die forging process parameters to achieve the effect of grain structure control in the key area of ​​the forging, and effectively improve the quality of the forging, and the method comprises the following steps:

[0010] Step 1: Analyze and select the key points of the turbine disc that need to be organized and controlled;

[0011] Step 2: Extract the deformation information of key points through finite element simulation of the whole process of turbine disc forging;

[0012] Step 3: Establish a target optimization function, propose requirements for the average grain size and recrystallization fraction of key points, and set corresponding weights for the final organizational goals of each key point;

[0013] Step 4: Set the boundary value and penalty function of the process parameters according to the actual forging situation. When the inferred process exceeds the boundary, the penalty function starts to execute;

[0014] Step 5: Embed the objective optimization function and penalty function into the multi-objective optimization algorithm and infer multiple forging process parameters simultaneously.

[0015] Step 6: Based on the above-mentioned turbine disk forging process reasoning method, in order to better realize the present invention, a turbine disk forging process reasoning system is further proposed to execute the above-mentioned turbine disk forging process reasoning method; receiving module, target optimization and boundary setting module, and reasoning module.

[0016] 2. According to the above method, the target optimization function established in step 3 can adapt to the grain structure target requirements of each key point by changing the weights. Therefore, there are two main optimization targets for each key point: recrystallization fraction and average grain size target. Therefore, the target optimization function can be expressed by formula (1):

[0017]

[0018] Among them, f i Used to characterize the degree of recrystallization of the i-th key region, s iUsed to characterize the average grain size of the i-th critical area, m i is the penalty weight, c i is the penalized function term, and the first two terms f and s are the standard values ​​after normalization of the recrystallization fraction term and the average grain size term, and have no units.

[0019] 3. According to the above method, the penalty function described in step 4 will take effect when the optimization parameter exceeds the upper and lower limits. It will make the value of the target optimization function larger and the fitness worse, thereby limiting the optimization range to within or near the feasible domain. In this study, the upper and lower limits of the optimized parameters x should be set first. u and x l , and then set the specific form of the penalty function. The optimization process is shown in formula (2).

[0020]

[0021] Where A(x) is the penalty function and m is the penalty factor. When the optimized parameter is within the boundary, the penalty function does not work; when the optimized parameter is not within the boundary, the penalty function will increase as the degree of exceeding the upper and lower limits of the boundary increases.

[0022] 4. Based on the above, a turbine disk die forging process reasoning method is proposed. In order to better realize the present invention, a turbine disk die forging process reasoning system is further proposed to execute the above-mentioned turbine disk die forging process reasoning method; receiving module, target optimization and boundary setting module, and reasoning module.

[0023] The receiving module is used to receive the input turbine disk position information and pass it to the subsequent processing link to provide basic data for the entire system.

[0024] The target optimization and boundary setting module is used to establish a target optimization function and set the boundary value and penalty function of the process parameters according to the actual situation of the forging process.

[0025] The inference module is used to screen the relationship between the process parameters and the parameters to be inferred, and use these relationships to derive the parameters to be inferred corresponding to the input process parameters. These parameters to be inferred will be used in the subsequent processing process to achieve optimization and quality control.

[0026] 5. According to the turbine disk forging process reasoning system, the feature is that in the receiving module, the following target data attributes are selected as data features used by the model:

[0027] (1) Turbine disk position information, including the specific position data of the turbine disk during the forging process, in order to analyze its stress and deformation.

[0028] (2) Simulation information, including extracting deformation information of key points through finite element simulation of the entire turbine disc forging process

[0029] 6. According to the above-mentioned turbine disk die forging process reasoning system, it is characterized in that in the target optimization and boundary setting module, the boundary value and penalty function of the process parameters are set according to the actual forging situation, and when the inferred process exceeds the boundary, the penalty function starts to execute.

[0030] 7. According to the above-mentioned turbine disk die forging process reasoning system, it is characterized in that the target optimization function and the penalty function are embedded in the multi-objective optimization algorithm in the reasoning module, and multiple forging process parameters are inferred simultaneously.

[0031] The innovation and beneficial effects of the present invention are as follows: the method can effectively plan the optimal process trajectory of die forging, so that the grain structure of the forging meets the target requirements, and provides a theoretical basis and technical support for achieving high-quality manufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the reasoning of the turbine disc die forging process in the preferred embodiment of the present invention;

[0033] Figure 2 The key point position distribution of the turbine disk in the preferred embodiment of the present invention;

[0034] Figure 3 The strain-displacement relationship of point P1 in the preferred embodiment of the present invention;

[0035] Figure 4 It is the multi-objective optimization process of the "TLBO" algorithm in the preferred embodiment of the present invention;

[0036] Figure 5 The following are cloud diagrams of recrystallization fraction and grain size distribution in the preferred embodiment of the present invention: (a) cloud diagram of recrystallization fraction of the control group; (b) cloud diagram of average grain size of the control group; (c) cloud diagram of recrystallization fraction of the optimized group; (d) cloud diagram of average grain size of the optimized group;

[0037] Figure 6 The final temperature and equivalent stress of the turbine disk forging in the preferred embodiment of the present invention are as follows: (a) temperature distribution cloud diagram; (b) equivalent stress distribution cloud diagram. DETAILED DESCRIPTION

[0038] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] The present invention is a die forging process reasoning method and system. Taking the die forging process reasoning of a certain type of aircraft engine high-pressure turbine disk as an example, the implementation details of the die forging process reasoning method and system involved in the present invention are introduced in detail.

[0040] This method mainly includes the following steps:

[0041] Step 1: In actual use, the stress field of the turbine disk changes with time, but only a few areas often fail. Therefore, it is not necessary to strictly control the microstructure of all areas on the entire turbine disk. Some key points can be selected and the microstructure corresponding to their performance can be used as the control target.

[0042] The rim and the wheel core may cause a small amount of plastic deformation during each start and stop process, and the key to fatigue failure lies in the accumulation of plastic deformation and the expansion of cracks, so key points P1 and P6 should be set on the rim and the wheel core. The web of the turbine disc is the main part of the turbine disc. The area close to the rim has a higher temperature. Due to the interaction between thermal stress and centrifugal force, fracture failure and yield may occur, so the key point P5 is set. In fact, point P5 also belongs to the rim area. In addition, at the connection between the web and the hub, certain stress concentration may occur there, so the connection between the web and the upper and lower hubs should also be set as key points, namely P2 and P3. The key point P4 represents the general position of the web to ensure the overall uniformity of the forging. The distribution positions of the key points are as follows: Figure 1 . As shown.

[0043] Step 2: To conduct process reasoning, it is first necessary to establish a finite element model of the turbine disk forging and extract the deformation information of key points. Figure 2 The figure shows the structure of a certain type of aircraft engine high-pressure turbine disk, the base material of which is GH4169 alloy. The initial parameter settings of the finite element model are shown in Table 1.

[0044] Table 1 Initial parameter settings of the finite element model

[0045]

[0046] During the forging process of turbine disks, the strain rate of each point inside the forging is not constant under ideal conditions, but changes with the specific shape of each area and the flow law of the metal. The time-varying strain rate makes the commonly used recrystallization kinetic model with constant strain rate unable to adapt to actual working conditions. The GH4169 kinetic model proposed by Chen can be used to apply to this working condition, as shown in formula (1).

[0047]

[0048] Among them, X drx is the recrystallization fraction, v drx is the recrystallization fractional differential term. c is the critical strain of dynamic recrystallization, ε 0.5is the strain when the recrystallization fraction reaches 50%, T is the deformation temperature (K), is the strain rate. According to the above formula, after obtaining the deformation history of the key points, the recrystallization fraction of each key point can be calculated.

[0049] For the recrystallization grain size model, the solid solution dynamic recrystallization grain size model proposed by Lin can be used, which can well describe the change of grain size of the alloy during deformation. The formula is as follows:

[0050]

[0051] Among them, R is the ideal gas constant (8.31 J / mol·K), and Z is the Zener-Hollomon parameter. The above-mentioned kinetic and average grain size models are also embedded in the DEFORM-2D finite element software by secondary development as the basic model for finite element simulation calculation.

[0052] Before the process reasoning is performed, a simulation of the isothermal die forging process of GH4169 alloy is performed with a benchmark finite element simulation condition, and point tracking is performed on six key point areas to extract the strain rate-displacement and strain-displacement process relationships of each key point from P1 to P6. The strain-displacement relationship of point P1 is as follows: Figure 3 The benchmark working conditions are as follows: the deformation temperature of the die and the blank is 980℃, and the pressing rate is 1mm / s. The simulated final forging microstructure distribution will also be used as a control observation group for comparison with the subsequent process optimization group.

[0053] Step 3: Establish a target optimization function, propose requirements for the average grain size and recrystallization fraction of key points, and set corresponding weights for the final organizational goals of each key point;

[0054] In the current optimization problem of near-isothermal forging of turbine disks, the goal is to meet the microstructure requirements corresponding to the service performance of each key point without damaging the die. Therefore, there are two main optimization goals for each key point: recrystallization fraction and average grain size target.

[0055] Therefore, the 12 parallel optimization objectives corresponding to the 6 key points can be expressed by formula (3).

[0056]

[0057] Among them, f i Used to characterize the degree of recrystallization of the i-th key region, s i Used to characterize the average grain size of the i-th critical area, m i is the penalty weight, c iis the penalized function. The first two terms f and s are the standard values ​​of the normalized recrystallization fraction term and the average grain size term, and have no units.

[0058] As mentioned above, the wheel core and hub areas must have good low-cycle fatigue resistance. According to research, fine grains can improve high-temperature low-cycle fatigue deformation behavior. Therefore, the microstructure of points P1, P5, and P6 requires strict uniformity and complete refinement. Therefore, the value of the weight coefficient should be relatively large. The main failure type of points P2 and P3 is fracture caused by stress concentration. For most alloys, cracks are prone to occur when the structure is not uniform. Therefore, the microstructure of points P2 and P3 must be uniform, regardless of whether it is coarse or fine grained. That is, the recrystallization fraction volume of P2 and P3 is the key criterion, and the grain size is not an absolute requirement. Therefore, according to the deformation conditions of P2 and P3, the value of w2 should be set larger and the sign of w3 is opposite to that of w2 to ensure that P2 and P3 achieve complete recrystallization or no recrystallization. The values ​​of r2 and r3 can be set to zero. Point P4 is a general point, giving general recrystallization fraction and grain size requirements, so w4 and r4 are not very large. The weights of each key point are shown in Table 2.

[0059] Table 2 Fitness function weights of each key point

[0060]

[0061] Step 4: In general product design or process design, restrictions and penalty functions are often added to ensure that the results are suitable for practical applications. When the optimization parameters exceed the upper and lower limits, the penalty function takes effect, which will increase the value of the target optimization function and deteriorate the fitness, thereby limiting the optimization range to within or near the feasible domain. There are many forms of penalty functions, and different types of penalty functions can be set according to different problems. In this study, a penalty function with a fixed penalty factor is used. First, the upper and lower limits x of the optimized parameters should be set. u and x l , that is, D = [x l ,x u ]; then, the specific form of the penalty function is set, as shown in formula (4).

[0062]

[0063] Where A(x) is the penalty function and m is the penalty factor. When the optimized parameter is within the boundary, the penalty function does not work; when the optimized parameter is not within the boundary, the penalty function will increase as the degree of exceeding the upper and lower limits of the boundary increases.

[0064] For the current isothermal die forging process of GH4169 alloy turbine disk, the main safety issues are die safety and life. Therefore, the penalty function should be designed for the two aspects of temperature and stress.

[0065] First, a penalty function for deformation heat generation is established. During the die forging process of turbine discs, a large amount of deformation heat is easily generated, which causes different temperatures in different areas of the forging, and eventually leads to a decrease in die strength, and may even cause overburning or damage the die. The deformation heat of GH4169 alloy has a direct relationship with the flow stress, so the flow stress of the key point at each moment during the deformation process is calculated first, and then the deformation heat generation under each strain is calculated. Finally, the total deformation heat is accumulated to obtain the final total deformation heat, as shown in formula (5).

[0066]

[0067] When the deformation temperature of GH4169 alloy reaches above 1040℃, the recrystallized grains coarsen rapidly, and even carbides may appear, causing overburning and damaging the die surface, so the temperature should be limited below 1040℃. High-temperature alloys are difficult to deform. When the temperature is lower than 950℃, the dynamic recrystallization behavior of GH4169 alloy is extremely slow, and the plasticity is poor. It is not only easy to cause uneven deformation, but also easy to produce great rheological resistance, which will have an adverse effect on the normal use of the mold and the normal operation of the forging equipment. Therefore, the minimum temperature should be set above 950℃. In summary, the penalty function considering the temperature rise is shown in formula (6).

[0068]

[0069] Where c1 is the penalty function considering the effect of deformation heat generation on grain structure, m1 is the penalty factor, T' is the temperature after considering deformation heat, T is the original temperature, and ΔT is the temperature rise. It can be derived by derivation of the mapping relationship between the downward displacement of the die forging press and the strain of each key point. The value of the penalty factor is temporarily set to 1.

[0070] The yield strength of the mold material currently used at 1050°C is 700 MPa. According to experience, at the same temperature, the yield strength of the mold material should be about 3 times that of the forging material. Therefore, the flow stress of the forged GH4169 alloy should be less than or equal to 220 MPa, and the penalty function form is shown in formula (7).

[0071]

[0072] Where c2 is the penalty function that considers the influence of the deformation resistance of the forging material on the die performance, m2 is the penalty factor, and the flow stress σ used is obtained by formula (5). The value of the penalty factor is set to 1.

[0073] Step 5: For the current optimization problem, the main optimization objects are the forging starting temperature and the forging pressing speed. The goal is to coordinate the microstructure of each key point to obtain a multi-performance turbine disk part. The difficulty of this problem lies in the fact that the deformation process parameters or organizational goals of each key control point are not consistent. In order to coordinately optimize the goals of each key point, all goals need to be considered in parallel, making the constraints of the optimization process numerous and complex, increasing the difficulty of the optimization problem, and posing a great challenge to the rapid and correct convergence of the optimization algorithm.

[0074] The "Teaching and Learning" multi-objective optimization algorithm (TLBO) is a relatively new optimization method with the characteristics of concise logic, fast convergence speed and concise implementation code. It has been widely used in engineering and mathematics. For this optimization problem, the coupling relationship between the process parameters to be optimized and the objectives is relatively complex. The "Teaching and Learning" algorithm can adapt well to the characteristics of this problem and converge quickly and accurately to obtain the optimal result.

[0075] The “teaching and learning” algorithm imitates the classroom learning scenario in real life. For the current process optimization problem, the “students” are regarded as a population, and the process parameters that need to be optimized are regarded as “subjects”. The “students’” “grades” are calculated through the target optimization function and the penalty function (Formula (3)). The individuals with the best “grades” are called “teachers”, and the others are “students”. The task of the “teachers” is to improve the average grades of the “students” and gradually approach the level T of the “teachers”. That is, in the i-th iteration, the “teachers” try to make the average grades of the “students” M j Report to the teacher i Close to get the new "student" average score M new , and record the difference between the original average score and the new average score as DifferenceMean i , as shown in formula (8).

[0076] Difference Mean i =r i (M new -T F M j ) (8)

[0077] Among them, T F is the teaching constant, which determines the size of the average value; r i The random constant in the i-th iteration ranges between [0, 1], T F The value of is determined in the interval [1, 2]. They are used as heuristic process quantities and are usually determined in each iteration with equal probability distribution.

[0078] TF =round[1+rand(0,1){2-1}] (9)

[0079] According to DifferenceMean i The value of the optimization variable can be updated, as shown in formula (10).

[0080] X new,i =X old,i +Difference Mean i (10)

[0081] In addition to the teaching of the "teacher", the "students" can also improve their performance by learning from each other in the population, which is another important feature of the TLBO algorithm. The "student" improves his performance in the population through random influence. If the performance of other "students" is better than his, he will learn from other students, and vice versa, other students will learn from him, as shown in formula (11).

[0082]

[0083] If f(X new,i ) is greater than f(X old,i ), then accept the learning process; otherwise, let X new,i =X old,i The process of TLBO algorithm is as follows Figure 4 As shown in the figure, after 50 iterations, the algorithm results have converged, which takes about 5 minutes. The feasible domain of the optimized process is: the initial temperature range is 982℃~988℃, and the pressing rate range is 0.240mm / s~0.390mm / s.

[0084] In order to conduct subsequent process verification, the initial temperature of the optimization group was set to 985°C and the pressing rate was set to 0.315 mm / s. The optimized process parameters (985°C, 0.315 mm / s) were imported into the DEFORM 2D finite element simulation software for simulation verification and compared with the control group (980°C, 1 mm / s) to verify the reliability of the results. The simulation results are shown in the figure below. Figure 5 shown.

[0085] like Figure 5As shown in the figure, overall, the turbine disk forgings with optimized process parameters have a higher overall recrystallization fraction, a smaller average grain size, and a more uniform grain size distribution. From the key point of view, the recrystallization fraction of the optimized group forgings at the wheel core is complete with that of the control group, and the optimized group has a larger area of ​​complete recrystallization, which can better meet the requirements; from the rim, the optimized group has basically undergone sufficient recrystallization in the entire rim, and the grains are also sufficiently refined, while only about half of the control group has undergone complete recrystallization, and the organization is also in a mixed crystal state; from the key points of the two hubs, the lower hub area of ​​the optimized group has undergone sufficient recrystallization, and the upper hub has also undergone partial recrystallization, but the degree is not high, and relatively coarse grains are still retained; for general points, the turbine disk of the optimized group has also undergone recrystallization from the outside to the inside, and the grains are refined overall.

[0086] Table 3. Tissue distribution of key points in the optimization group and the control group.

[0087]

[0088] As shown in Table 3, the recrystallization volume fractions of points P1, P2, and P6 in the optimization group are 0.992, 0.975, and 0.982, respectively, which are higher than those in the control group, while the average grain sizes of P1 and P6 in the optimization groups are smaller than those in the control group, indicating that the low-cycle fatigue and processing properties of these optimized forging areas are better than those in the control group. For point P5, the recrystallization fraction of the control group is slightly higher than that of the optimization group. The recrystallization behavior of point P3 in both the optimization and control groups is still restricted, but the recrystallization level of group P4 in the optimization group is much higher than that of the initial group. This means that the general area of ​​the turbine disk can undergo a more adequate recrystallization process, and the optimized processing parameters can satisfactorily meet the key requirements on a global scale.

[0089] Figure 6 (a) shows the temperature distribution of the turbine disk forging in the final forming stage. It can be found that the temperature in most areas is below 1000°C, which shows that the optimized process parameters can effectively control the generation of deformation heat without excessive growth of recrystallized grains. Figure 6 (b) shows the distribution of equivalent stress. For most forging areas, the effective stress is at a low level with an average value of 126MPa. This indirectly reveals that the load between the die and the forging is relatively small, which is of great significance for extending the service life of the die. On the other hand, low effective stress means better forming performance under this forging condition, which can improve filling capacity and ensure suitable material flow lines. Therefore, the inferred forging process parameters can ensure good billet plasticity and die safety during die forging.

[0090] Step 6: Based on all the items in the above steps 1 to 5, this step proposes a turbine disk forging process reasoning system for executing the above-mentioned turbine disk forging process reasoning method; it includes a receiving module, a target optimization and boundary setting module, and a reasoning module.

[0091] The receiving module is used to receive the input turbine disk position information and pass it to the subsequent processing link to provide basic data for the entire system.

[0092] The target optimization and boundary setting module is used to establish a target optimization function and set the boundary value and penalty function of the process parameters according to the actual situation of the forging process.

[0093] The inference module is used to screen the relationship between the process parameters and the parameters to be inferred, and use these relationships to derive the parameters to be inferred corresponding to the input process parameters. These parameters to be inferred will be used in the subsequent processing process to achieve optimization and quality control.

Claims

1. A process reasoning method and system for turbine disk die forgings, characterized in that By reasonably reasoning the die forging process parameters through the method, the effect of grain structure control in the key area of ​​the forging can be achieved, and the quality of the forging can be effectively improved. The method and system construction include the following steps: Step 1: Analyze and select the key points of the turbine disc that need to be organized and controlled; Step 2: Extract the deformation information of key points through finite element simulation of the whole process of turbine disc forging; Step 3: Establish a target optimization function, propose requirements for the average grain size and recrystallization fraction of key points, and set corresponding weights for the final organizational goals of each key point; Step 4: Set the boundary value and penalty function of the process parameters according to the actual forging situation. When the inferred process exceeds the boundary, the penalty function starts to execute; Step 5: Embed the objective optimization function and penalty function into the multi-objective optimization algorithm and infer multiple forging process parameters simultaneously. Step 6: Based on the turbine disk forging process reasoning method proposed above, in order to better realize the present invention, a turbine disk forging process reasoning system is further proposed, which is used to execute the turbine disk forging process reasoning method receiving module, target optimization and boundary setting module, and reasoning module.

2. The method according to claim 1, characterized in that: The target optimization function established in step 3 can adapt to the grain structure target requirements of each key point by changing the weights. Therefore, there are two main optimization targets for each key point: recrystallization fraction and average grain size target. Therefore, the target optimization function can be expressed by formula (1): Among them, f i Used to characterize the degree of recrystallization of the i-th key region, s i Used to characterize the average grain size of the i-th critical area, m i is the penalty weight, c i is the penalized function term, and the first two terms f and s are the standard values ​​after normalization of the recrystallization fraction term and the average grain size term, and have no units.

3. The method according to claim 1, characterized in that: The penalty function described in step 4 will take effect when the optimization parameters exceed the upper and lower limits of the boundaries. It will increase the value of the target optimization function and deteriorate the fitness, thereby limiting the optimization range to within or near the feasible domain. In this study, we should first set the upper and lower limits x of the optimized parameters. u and x l , and then set the specific form of the penalty function. The optimization process is shown in formula (2). Where A(x) is the penalty function and m is the penalty factor. When the optimized parameter is within the boundary, the penalty function does not work; when the optimized parameter is not within the boundary, the penalty function will increase as the degree of exceeding the upper and lower limits of the boundary increases.

4. The method according to claim 1, a turbine disk forging process reasoning system, characterized in that: include: The receiving module is used to receive the input turbine disk position information and pass it to the subsequent processing link to provide basic data for the entire system; The target optimization and boundary setting module is used to establish the target optimization function and set the boundary value and penalty function of the process parameters according to the actual situation of the forging process; The inference module is used to screen the relationship between the process parameters and the parameters to be inferred, and use these relationships to derive the parameters to be inferred corresponding to the input process parameters. These parameters to be inferred will be used in the subsequent processing to achieve optimization and quality control.

5. A turbine disk forging process reasoning system according to claim 4, characterized in that: In the receiving module, the following target data attributes are selected as data features used by the model: (1) Turbine disk position information, including the specific position data of the turbine disk during the forging process, so as to analyze its stress and deformation; (2) Simulation information, including extracting key point deformation information through finite element simulation of the entire turbine disk forging process.

6. A turbine disk forging process reasoning system according to claim 4, characterized in that: In the target optimization and boundary setting module, the boundary values ​​and penalty functions of the process parameters are set according to the actual forging conditions. When the inferred process exceeds the boundary, the penalty function begins to execute.

7. A turbine disk forging process reasoning system according to claim 4, characterized in that: In the reasoning module, the objective optimization function and penalty function are embedded in the multi-objective optimization algorithm, and multiple forging process parameters are inferred simultaneously.

Images