Digital Precision Machining Method for Coordinating Temperature and Formability of Integrally Forged Shaft of Large Rotor

Abstract

The present invention discloses a collaborative digital precision machining method for warm forming of a large rotor integral forging shaft. In the present invention, a training data set and a verification data set obtained through cutting simulation experiments by a cutting simulation model are used to train and verify a geometric error transfer law model. During machining, the geometric errors of the workpiece are predicted based on various machining parameters, the temperature field of the machining area detected in real time, and the geometric error transfer law model. When the prediction results do not meet the preset requirements, the machining parameters are optimized. The key geometric structures of the workpiece are obtained through calculation and analysis of the training data set and the verification data set. The dynamic balance performance simulation experiment is carried out on the workpiece geometric model introducing the geometric errors of the key geometric structures after workpiece machining, so as to realize the prediction of the dynamic balance performance after workpiece machining. The present invention can realize the in-machine prediction and control of the machining geometric errors of the large rotor integral forging shaft, and can realize the prediction of the dynamic balance performance of the large rotor integral forging shaft.

Description

Technical Field

[0001] The present invention belongs to the technical field of precision machining of large rotor rotating parts, and specifically relates to a temperature-shape-property collaborative digital precision machining method for large rotor integral forging shafts. Background Art

[0002] High-end turbomachinery equipment such as industrial steam turbines, gas turbines, and aero-engines all achieve energy conversion through large rotating parts (such as integral forging shafts up to 6-8 m in length and turbine disks with a maximum diameter of 800 mm). The industry unanimously believes that the machining accuracy of such parts will directly affect the energy conversion efficiency and the safety of equipment operation. Therefore, the design side often puts forward extremely high dimensional accuracy requirements for such parts. Among them, during the cutting process of large rotor integral forging shafts, especially during the machining of the outer cylindrical surface of large rotor integral forging shafts, the workpiece machining surface has a thermal expansion problem due to uneven temperature, resulting in a non-linear change in the geometric error on the machining surface, and further causing the problem of unstable dynamic balance performance of the workpiece. Therefore, how to effectively control the influence of temperature on the machining deformation amount and achieve precise guarantee of the dynamic balance performance of large rotor integral forging shafts is particularly important.

[0003] There are some relatively mature methods in the field of controlling machining thermal deformation at present. However, there is currently no method for realizing digital precision machining through the integrated control of temperature, shape, and property during the machining of large rotating parts of turbomachinery. In order to effectively predict and detect the thermal deformation amount of the outer circle surface of large rotating parts during machining and effectively correlate the dimensional error of the outer circle surface with the dynamic balance performance, there is an urgent need for a machining method that can meet the high-performance machining requirements of the outer circle of large rotating parts. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and propose a temperature-shape-property collaborative digital precision machining method for large rotor integral forging shafts.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions:

[0006] The temperature-shape-property collaborative digital precision machining method for large rotor integral forging shafts of the present invention is as follows:

[0007] Step 1: Obtain the crystallographic data of the workpiece sample with inhomogeneous and discontinuous characteristics.

[0008] Step 2: Establish a crystal plasticity model of the workpiece with inhomogeneous and discontinuous characteristics.

[0009] Step 3: Establish a finite element model for turning machining.

[0010] Step 4: Establish a cutting simulation model of force-thermal-fluid coupling.

[0011] Step 5: Conduct a cutting simulation experiment using the cutting simulation model to obtain a training dataset and a validation dataset composed of workpiece surface temperature field data, workpiece geometric error data, and various machining parameters; among them, the various machining parameters include tool wear, tool geometric parameters, feed rate, and cutting fluid flow rate, and the geometric errors include shape error, position error, and dimensional error.

[0012] Step 6: Establish a geometric error transfer law model with the various machining parameters and the workpiece surface temperature field data varying with cutting time as input data and the geometric error of the workpiece as output data.

[0013] Step 7: Normalize the input data of the training dataset and the validation dataset, and use the normalized training dataset and validation dataset to train and validate the geometric error transfer law model.

[0014] Step 8: Identify the key geometric structures on the workpiece.

[0015] Step 9: The CNC machine tool processes the workpiece according to the set various machining parameters, and the machining process control system combines the various machining parameters, the temperature field of the machining area detected in real time, and the geometric error transfer law model to optimize the feed rate and the cutting fluid flow rate, so that the workpiece surface temperature field meets the preset temperature requirements and the geometric accuracy at each position on the workpiece meets the accuracy requirements. If optimizing the feed rate and the cutting fluid flow rate cannot meet the temperature and accuracy requirements, replace the tool to change the tool wear or the tool geometric parameters.

[0016] Step 10: After the workpiece is processed, measure the geometric errors at the positions of the key geometric structures on the workpiece.

[0017] Step 11: Predict the dynamic balance performance of the processed workpiece based on the geometric errors measured at the positions of the key geometric structures in Step 10.

[0018] Preferably, the process of Step 1 is as follows: Cut the workpiece sample, prepare the workpiece sample into a workpiece specimen, detect the workpiece specimen through a scanning electron microscope, collect the crystallographic data of the workpiece specimen to obtain the crystallographic image of the workpiece specimen. Among them, when collecting the grain orientation data, analyze the EBSD data to obtain the grain orientation of each crystal on the workpiece specimen. When collecting the grain size data, select the grain sizes of each crystal within a preset range on the workpiece specimen. When collecting the dislocation density data, measure the microstructure distribution of the workpiece specimen to obtain the dislocation density of the crystals on the workpiece specimen; then use the MTEX plug-in in Matlab to generate a script, process the obtained crystallographic data of the workpiece specimen, analyze to obtain the grain orientation and position coordinates of each crystal on the workpiece specimen, draw the crystal orientation map, and export the crystallographic data of the workpiece specimen as a text-format Abaqus input file.

[0019] More preferably, the process of the second step is as follows: establish a workpiece simulation model through finite element simulation software, perform hexahedral mesh division on the workpiece simulation model, import the Abaqus input file into the finite element simulation software, and map the dislocation density of the crystals on the workpiece sample, the grain orientation and grain size of each crystal to each grid node of the workpiece simulation model through the subroutine of the finite element simulation software. Then, define the plastic properties of the crystals on the workpiece, and then conduct a quasi-static tensile experiment, and obtain the mechanical parameters of the grains and grain boundaries by combining molecular dynamics simulation. Finally, obtain the workpiece crystal plasticity model, where the mechanical parameters of the grains include the elastic modulus and yield strength of the grains, and the mechanical parameter of the grain boundary is the cohesive force model parameter of the grain boundary.

[0020] Preferably, the process of the third step is as follows: establish a tool simulation model, define the tool material properties after meshing, and then assemble the workpiece simulation model and the tool simulation model to obtain a finite element model for turning processing. During the assembly process, set the boundary conditions and load conditions, set the friction coefficient between the tool and the workpiece, and set the thermal conductivity of the tool and the workpiece; where the tool is defined as a rigid body, the thermal conductivity of the tool is a fixed value, and for different grain orientations of each crystal on the workpiece, there corresponds a thermal conductivity tensor K. Different thermal conductivities are assigned to each grain through the thermal conductivity tensor K. The thermal conductivity tensor K is a 3×3 matrix:

[0021]

[0022] In the formula, k xx 、k yy and k zz respectively represent the thermal conductivities along the x-axis direction, y-axis direction and z-axis direction, and k xy 、k xz 、k yx 、k yz 、k zx and k zy respectively represent the coupled thermal conductivities between the x-axis and y-axis, between the x-axis and z-axis, between the y-axis and x-axis, between the y-axis and z-axis, between the z-axis and x-axis, and between the z-axis and y-axis.

[0023] More preferably, the process of step four is as follows: First, an Euler domain model of the cooling process is constructed. The Euler domain model is divided into a cutting fluid region and an empty region. The cutting fluid material properties of the cutting fluid region are defined, and the Euler domain model is meshed. Then, the Euler domain model is assembled with the finite element model of turning machining to obtain a cutting simulation model. During the assembly process, the cutting fluid region is moved to the contact region between the tool and the workpiece, and the load conditions of the cutting fluid region are constrained to be consistent with the tool load conditions. The analysis step parameters and output parameters are set. The output parameters are the temperature field data of the workpiece machining surface changing with the cutting time during the cutting simulation process and the geometric error of the workpiece after the cutting simulation.

[0024] Preferably, the process of step five is as follows: Set the processing parameter ranges of the cutting and cooling strategies. The processing parameters of cutting include tool wear amount, tool geometry parameters, and feed rate. The processing parameter of the cooling strategy is the cutting fluid flow rate. Select multiple parameter nodes within each processing parameter range, perform an orthogonal design on the parameter nodes of each processing parameter, obtain a processing parameter combination set composed of combinations of different parameter nodes selected for each processing parameter, and divide the processing parameter combination set into a training set and a validation set; Through the cutting simulation model, use the processing parameter combinations of the training set and the validation set to conduct cutting simulation experiments, obtain the temperature field of the workpiece machining surface and the geometric error of the workpiece during the cutting simulation under the processing parameter combinations of the training set and the validation set, and then obtain a training data set and a validation data set composed of tool wear amount, tool geometry parameters, feed rate, cutting fluid flow rate, temperature field data, and the geometric error of the workpiece.

[0025] More preferably, the process of step eight is as follows: According to the training data set and the validation data set, calculate the first-order sensitivity index S i (Y j ) and the total sensitivity index S Ti (Y j ) of the shape error, position error, and dimensional error at each position on the workpiece. Among them, the calculation formula for the first-order sensitivity index S i (Y j ) is

[0026]

[0027]

[0028] In the formula, X i represents all the input data of the geometric error transfer law model corresponding to the i-th data point of the workpiece surface temperature field in the training data set and the validation data set. i = 1, 2, 3... n, where n is the total number of data points of the workpiece surface temperature field, and Y jDenote the shape error, position error, or dimensional error data corresponding to the workpiece surface temperature field in the training dataset and the validation dataset at the \(i\)-th data point. \(j = 1, 2,\) and \(3\) correspond to shape error, position error, and dimensional error respectively. \(Var(Y j |X i ) is the variance of all shape errors, position errors, or dimensional errors corresponding to the workpiece surface temperature field in the training dataset and the validation dataset at the \(i\)-th data point; \(Var(Y j ) is the total variance of all shape errors, position errors, or dimensional errors in the training dataset and the validation dataset, and \(E[Y j is the expected value of all shape errors, position errors, or dimensional errors in the training dataset and the validation dataset;

[0029] The total sensitivity index \(S\) of shape error, position error, or dimensional error Ti (Y j ) is calculated by the formula

[0030]

[0031] where, is the variance of all the remaining shape errors, position errors, or dimensional errors in the training dataset and the validation dataset except those corresponding to \(X i ;

[0032] Next, calculate the weighted comprehensive index \(I\) of the first-order sensitivity index and the total sensitivity index of shape error, position error, and dimensional error at each position on the workpiece i :

[0033]

[0034] where, \(\alpha\) and \(\beta\) are sensitivity weight coefficients, and \(\alpha+\beta = 1\), \(\alpha\lt\beta\); \(\omega 1 \), \(\omega 2 and \(\omega 3 are error weight coefficients, and \(\omega 1 +\omega 2 +\omega 3 = 1;

[0035] Then, sort the values of \(I i at each position on the workpiece from high to low, and select the three positions on the workpiece corresponding to the top three ranked values as the key geometric structures.

[0036] More preferably, the process of step nine is as follows: The numerical control machine tool processes the workpiece according to the set machining parameters. During the machining process, the infrared temperature sensor placed above the tool rest is used to detect the temperature field of the machining area on the surface of the tool and the workpiece in real time, and the detected temperature field data is transmitted to the control system. The control system predicts the geometric error at the machining position of the workpiece under the current machining parameter combination through the geometric error transfer law model based on the machining parameters and the temperature field data detected in real time. If all the temperature values in the temperature field data detected in real time do not exceed the preset temperature and the predicted geometric error at the machining position meets the accuracy requirements, the numerical control machine tool continues to process the workpiece according to the current machining parameter combination. Otherwise, the feed rate and the cutting fluid flow rate are optimized (usually by reducing the feed rate and increasing the cutting fluid flow rate) so that all the temperature values in the temperature field data detected in real time do not exceed the preset temperature and the geometric error at the machining position predicted by the geometric error transfer law model based on the optimized machining parameter combination and the temperature field data detected in real time meets the accuracy requirements. When optimizing the feed rate and the cutting fluid flow rate cannot meet the temperature and accuracy requirements, the tool is replaced to change the tool wear amount or the tool geometric parameters to avoid excessive tool wear or unsuitable tool geometric parameters for the current cutting.

[0037] Preferably, in step ten, the geometric errors of the key geometric structures of the workpiece are measured by a ruby probe.

[0038] Preferably, the process of step eleven is as follows: A geometric model of the machined workpiece is constructed, and the measured geometric error data of the key geometric structures is imported into the finite element software. Unbalanced masses are applied at the positions of the key geometric structures of the workpiece geometric model to form an unbalanced moment. Among them, the geometric model of the machined workpiece is constructed by using the process of establishing the crystal plasticity model of the workpiece in step two. Then, boundary conditions are applied to the geometric model of the workpiece. The support conditions are defined at the bearing supports of the workpiece, the analysis step is set as transient dynamic analysis, the rotational speed curve of the workpiece geometric model is defined, and the time step is set. Then, a dynamic balance performance simulation experiment is carried out, and key parameters are output. The dynamic balance performance of the workpiece is evaluated according to the key parameters. If all the key parameters obtained from the simulation are within the corresponding preset acceptable ranges, the machining is completed. Otherwise, the key geometric structures are further machined, and the unbalanced masses at the positions of the key geometric structures of the workpiece geometric model after the finish machining are updated, and the dynamic balance performance simulation experiment is carried out again until the dynamic balance performance of the workpiece meets the requirements. Among them, the key parameters include the displacement responses of the key geometric structures, the reaction forces and moments at the bearing supports, the vibration spectrum of the workpiece, and the stress and strain distributions of the key geometric structures.

[0039] The present invention has the following beneficial effects:

[0040] 1. The present invention can realize in - machine prediction and control of the geometric errors in the machining of large - scale rotor integral forging shafts, and can accurately predict the dynamic balance performance of large - scale rotor integral forging shafts. Specifically, the present invention establishes a workpiece simulation model, a tool simulation model, and an Euler domain model for the cooling process, assembles the workpiece simulation model, the tool simulation model, and the Euler domain model, and combines the plastic characteristics of the crystals on the workpiece to obtain a force - heat - fluid coupled cutting simulation model. Through the cutting simulation model, cutting simulation experiments are carried out to obtain a training data set and a verification data set composed of workpiece surface temperature field data, workpiece geometric error data, and various machining parameters. By using the training data set and the verification data set to train and verify the established geometric error transfer law model, during the workpiece machining process, the control system combines various machining parameters, the temperature field of the machining area between the tool and the workpiece surface detected by the infrared temperature sensor in real - time, and the geometric error transfer law model to realize in - machine prediction of the geometric errors of the workpiece. And when the prediction result does not meet the requirements, the control system optimizes the corresponding machining parameters to make the prediction result meet the requirements, thus realizing in - machine prediction and control of the geometric errors of the workpiece. Further, the present invention uses the training data set and the verification data set to perform computational analysis to obtain the key geometric structures on the workpiece. By detecting the geometric errors of each key geometric structure on the workpiece after machining and introducing the detection results into the geometric model of the workpiece after machining, the influence of unbalanced mass is introduced into the geometric model of the workpiece after machining, and then a dynamic balance performance simulation experiment is carried out, thereby realizing accurate prediction of the dynamic balance performance of the workpiece after machining.

[0041] 2. In the process of establishing the workpiece model, the present invention considers the inhomogeneous and discontinuous characteristics of the workpiece material. By collecting the crystal attributes of the workpiece material and introducing the crystal attributes of the workpiece material into the workpiece simulation model, the accuracy of the prediction result is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is the flow chart of the present invention.

[0043] Figure 2 is the flow chart of establishing the force - heat - fluid coupled cutting simulation model in the present invention.

[0044] Figure 3 is the flow chart of establishing the geometric error transfer law model in the present invention.

[0045] Figure 4 is the flow chart of realizing the prediction of the dynamic balance performance of the workpiece in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0046] The present invention will be further described below with reference to the drawings.

[0047] Such as Figure 1 、 Figure 2, Figure 3 and Figure 4 As shown in and

[0048] , the large-rotor integral-forged shaft warm forming and property synergy digital precision machining method of the present invention is as follows:

[0048] Step 1. Obtain crystallographic data of a workpiece sample with heterogeneous and discontinuous characteristics: Cut a workpiece sample with dimensions of 10 mm × 10 mm × 5 mm. After polishing the surface of the workpiece sample, perform ultrasonic cleaning, etching, cleaning, and drying in sequence to prepare the workpiece sample. Then, detect the workpiece sample through a scanning electron microscope, collect the crystallographic data on the workpiece sample, and obtain the crystallographic image of the workpiece sample. Among them, the crystallographic data includes grain orientation (Euler angle), dislocation density, and grain size. When collecting grain orientation data, analyze through EBSD data to obtain the grain orientation of each crystal on the workpiece sample. When collecting grain size data, select the grain sizes of each crystal within the range of 10 μm to 30 μm on the workpiece sample. When collecting dislocation density data, measure the microstructure distribution of the workpiece sample to obtain the dislocation density of the crystals on the workpiece sample. Then, use the MTEX plug-in in Matlab to generate a script, process the obtained crystallographic data of the workpiece sample, analyze to obtain the grain orientation and position coordinates of each crystal on the workpiece sample, draw a crystal orientation map, and export the crystallographic data of the workpiece sample as a text-format Abaqus input file.

[0049] Step 2. Establish a crystal plasticity model of the workpiece with heterogeneous and discontinuous characteristics: Establish a workpiece simulation model through finite element simulation software, and perform hexahedral mesh division on the workpiece simulation model. To ensure that the mesh in the cutting area of the workpiece is fine enough to accurately capture the influence of the microstructure of the workpiece on the cutting force and deformation behavior, in this embodiment, the minimum mesh size is set to 0.5 μm. Import the Abaqus input file into the finite element simulation software, and map the dislocation density of the crystals on the workpiece sample, the grain orientation and grain size of each crystal to each mesh node of the workpiece simulation model through the subroutine of the finite element simulation software to ensure that the influence of the microstructure of the workpiece on the mechanical behavior can be accurately reflected during simulation. Then, define the plastic properties of the crystals on the workpiece, including integrating the dislocation density of the crystals on the workpiece sample, the grain orientation and grain size of each crystal into the Johnson-Cook constitutive model of the workpiece and setting a fracture threshold. In this embodiment, the fracture threshold is set to 0.15. Perform a quasi-static tensile experiment, and combine molecular dynamics simulation to obtain the mechanical parameters of the grains and grain boundaries. Finally, obtain the crystal plasticity model of the workpiece. Among them, the mechanical parameters of the grains include the elastic modulus and yield strength of the grains, and the mechanical parameter of the grain boundary is the cohesive force model parameter of the grain boundary.

[0050] Step 3. Create a finite element model for turning machining: Establish a tool simulation model, define the tool material properties after meshing. Among them, define the tool as a rigid body in the simulation to reduce the computational complexity. Then, assemble the workpiece simulation model and the tool simulation model to obtain a finite element model for turning machining. During the assembly process, set the boundary conditions and load conditions, set the friction coefficient between the tool and the workpiece, and set the thermal conductivity of the tool and the workpiece. Among them, define the thermal conductivity of the tool as a constant value. For different grain orientations of each crystal on the workpiece, there corresponds a thermal conductivity tensor K. Different thermal conductivity tensors K can be assigned to each grain with different thermal conductivities by writing a subroutine of the finite element simulation software. The thermal conductivity tensor K can be expressed as a 3×3 matrix:

[0051]

[0052] In the formula, k xx 、k yy and k zz represent the thermal conductivities in the x-axis direction, y-axis direction, and z-axis direction respectively. k xy 、k xz 、k yx 、k yz 、k zx and k zy represent the coupled thermal conductivities between the x-axis and y-axis, between the x-axis and z-axis, between the y-axis and x-axis (the same as k xy ), between the y-axis and z-axis, between the z-axis and x-axis (the same as k xz ), and between the z-axis and y-axis (the same as k yz ) respectively.

[0053] Step 4. Establish a coupled force-thermal-fluid cutting simulation model; construct an Euler domain model for the cooling process, divide the Euler domain model into a cutting fluid region and an empty region, define the cutting fluid material properties of the cutting fluid region, and mesh the Euler domain model. Then, assemble the Euler domain model and the finite element model for turning machining to obtain a cutting simulation model. During the assembly process, move the cutting fluid region to the contact region between the tool and the workpiece, and constrain the load conditions of the cutting fluid region to be consistent with the tool load conditions. Set the analysis step parameters and output parameters. Among them, the output parameters are the temperature field data of the workpiece machining surface changing with the cutting time during the cutting simulation process and the geometric errors of the workpiece after the cutting simulation. The geometric errors include shape errors, position errors, and dimensional errors.

[0054] Step 5. Geometric error simulation experiment: Set the processing parameter ranges of the cutting and cooling strategies. The processing parameters of cutting include tool wear, tool geometry parameters, and feed rate, and the processing parameter of the cooling strategy is the cutting fluid flow rate. Select multiple parameter nodes within each processing parameter range, conduct an orthogonal design for the parameter nodes of each processing parameter, obtain a set of processing parameter combinations composed of combinations of different parameter nodes selected for each processing parameter, and divide the set of processing parameter combinations into a training set and a validation set. In this embodiment, the parameter nodes selected within each processing parameter range are shown in Table 1; Set the workpiece rotation speed to be constant, and through the cutting simulation model, use the processing parameter combinations of the training set and the validation set to conduct cutting simulation experiments under force-thermal-fluid coupling to obtain the temperature field of the workpiece machining surface and the geometric error of the workpiece during the cutting simulation process for each processing parameter combination of the training set and the validation set, and further obtain a training data set and a validation data set composed of tool wear, tool geometry parameters, feed rate, cutting fluid flow rate, temperature field data, and the geometric error of the workpiece.

[0055] Table 1 Orthogonal simulation experiment parameter table

[0056]

[0057] Step 6. Establish a geometric error transfer law model; Use tool wear, tool geometry parameters, feed rate, cutting fluid flow rate, and workpiece surface temperature field data that changes with cutting time as input data, and use the shape error, position error, and size error of the workpiece as output data to establish a geometric error transfer law model. Among them, the geometric error transfer law model is built using a feedforward neural network model of the Pytorch framework. The structure of the feedforward neural network model is: the number of neurons in the input layer is 5, the hidden layer is set to two layers to extract complex non-linear features in the data, the activation function uses the ReLU function, and the number of neurons in the output layer is 3.

[0058] Step 7. Normalize each input data of the training data set and the validation data set, and use the normalized training data set and validation data set to train and validate the geometric error transfer law model established in Step 6, which can accurately predict the law of the workpiece geometric error and realize the in-machine prediction of the workpiece geometric error during the machining process.

[0059] Step 8. Use the variance decomposition method to identify the key geometric structures on the workpiece according to the training data set and the validation data set: Calculate the first-order sensitivity index S i (Y j ) and the total sensitivity index S Ti (Y j ) of the shape error, position error, and size error at each position on the workpiece, where the first-order sensitivity index S i (Yj ) The calculation formula is

[0060]

[0061]

[0062] Wherein, X i represents all the input data of the geometric error transfer law model corresponding to the workpiece surface temperature field at the i-th data point in the training data set and the validation data set (i.e., tool wear, tool geometric parameters, feed rate, cutting fluid flow rate, and workpiece surface temperature field data), and i = 1, 2, 3... n, where n is the total number of data points of the workpiece surface temperature field, and Y j represents the shape error, position error, or dimensional error data corresponding to the workpiece surface temperature field at the i-th data point in the training data set and the validation data set, and j = 1, 2, and 3 correspond to shape error, position error, and dimensional error respectively. Var(E[Y j ∣X i ) is the variance of all shape errors, position errors, or dimensional errors corresponding to the workpiece surface temperature field at the i-th data point in the training data set and the validation data set; Var(Y j ) is the total variance of all shape errors, position errors, or dimensional errors in the training data set and the validation data set, and E[Y j is the expected value of all shape errors, each position error, or each dimensional error in the training data set and the validation data set;

[0063] The total sensitivity index S Ti (Y j ) The calculation formula is

[0064]

[0065] Wherein, is the variance of all the remaining shape errors, position errors, or dimensional errors in the training data set and the validation data set except those corresponding to X i .

[0066] Then calculate the weighted comprehensive index I of the first-order sensitivity index and the total sensitivity index of the shape error, position error, and dimensional error at each position on the workpiece i :

[0067]

[0068] Wherein, α and β are sensitivity weight coefficients used to adjust the relative importance of S i and S Ti , and α + β = 1, α < β. In this embodiment, α = 0.3 and β = 0.7 are selected; ω1 , ω 2 and ω 3 are error weight coefficients, and ω 1 + ω 2 + ω 3 = 1. In this embodiment, ω 1 = ω 2 = ω 3 = 1 / 3.

[0069] Then, sort the I i values at each position on the workpiece from high to low, and select three positions on the workpiece corresponding to the top three values as the key geometric structures.

[0070] Step Nine: The CNC machine tool processes the workpiece according to the set machining parameters. During the machining process, an infrared temperature sensor placed above the tool rest is used to detect the temperature field of the machining area on the surface of the tool and the workpiece in real time, and the detected temperature field data is transmitted to the control system. At the same time, the control system predicts the geometric error at the machining position of the workpiece under the current machining parameter combination through the geometric error transfer law model based on the machining parameters and the real-time detected temperature field data. Among them, before machining, the wear amount of the currently used tool is measured by a laser tool setter, and the wear amount of the tool is transmitted to the control system; during the machining of the workpiece, if the temperature values in the temperature field data detected in real time by the infrared temperature sensor on the workpiece do not exceed the preset temperature and the predicted geometric error at the machining position meets the accuracy requirements, the CNC machine tool continues to machine the workpiece according to the current machining parameter combination. Otherwise, the adjustment mechanism of the control system is triggered. The control system optimizes the feed rate and the cutting fluid flow rate so that the temperature values in the real-time detected temperature field data do not exceed the preset temperature and the geometric error at the machining position predicted by the geometric error transfer law model based on the optimized machining parameter combination and the real-time detected temperature field data meets the accuracy requirements, avoiding problems such as thermal expansion and deformation caused by uneven temperature, and ensuring that the geometric accuracy of the workpiece meets the accuracy requirements. If optimizing the feed rate and the cutting fluid flow rate cannot meet the temperature and accuracy requirements, replace the tool to change the tool wear amount or the tool geometric parameters.

[0071] Step Ten: After the workpiece is machined, replace the tool with a ruby probe. The CNC machine tool controls the ruby probe to accurately move to the positions of the key geometric structures of the workpiece, and precisely measures the geometric errors of the key geometric structures through the ruby probe. After the measurement is completed, the measurement data is sent to the control system.

[0072] Step Eleven: Precise Prediction of Dynamic Balance Performance: Construct a geometric model of the machined workpiece, import the geometric error data of the measured key geometric structures into the finite element software, apply the unbalanced mass at the positions of the key geometric structures of the workpiece geometric model to form an unbalanced moment, and introduce the influence of the unbalanced mass into the workpiece geometric model. Among them, the geometric model of the machined workpiece is constructed by using the process of establishing the crystal plasticity model of the workpiece in Step Two; then, boundary conditions are applied to the geometric model of the workpiece, the support conditions are defined at the bearing supports of the workpiece to simulate the rolling support during actual operation, the analysis step is set as transient dynamic analysis, the rotational speed curve of the workpiece geometric model is defined to simulate the process of the workpiece reaching the working speed of 2500 rmp from rest, and the time step size is set to capture the transient response caused by the unbalanced moment during the rotation of the workpiece; then, a dynamic balance performance simulation experiment is carried out, and key parameters are output to evaluate the dynamic balance performance of the workpiece. The key parameters include the displacement response of each key geometric structure, the reaction force and moment at the bearing support, the vibration spectrum of the workpiece, and the stress and strain distributions of each key geometric structure. Among them, the radial displacement and its change of the key geometric structure during rotation can be analyzed through the displacement response of each key geometric structure, the change of the additional load caused by the unbalanced moment can be evaluated through the reaction force and moment at the bearing support, whether the working frequency of the workpiece is close to the natural frequency of the mechanical system during operation can be judged through the vibration spectrum, the vibration frequency and resonance risk of the workpiece at different rotational speeds can be analyzed, and whether there is a risk of fatigue failure in the key geometric structure can be analyzed through the stress and strain distributions of each key geometric structure; if all the key parameters obtained from the simulation are within the corresponding preset acceptable ranges, it means that the dynamic balance performance of the machined workpiece meets the requirements and the machining is completed. Otherwise, it means that the dynamic balance performance of the machined workpiece does not meet the requirements, and further finish machining is required for each key geometric structure, and the unbalanced mass at the positions of the key geometric structures after finish machining of the workpiece geometric model is updated, and the dynamic balance performance simulation experiment is carried out again until the dynamic balance performance of the workpiece meets the requirements; among them, in this embodiment, the acceptable range of the displacement of the key geometric structure is 0~3 mm, the acceptable range of the reaction force at the bearing support is 0~10 kN, the acceptable range of the moment is 0~500 N·m, the acceptable range of the vibration spectrum of the workpiece is that the workpiece frequency is outside the range of plus and minus 5 Hz of the natural frequency of the mechanical system and its multiple frequencies, the acceptable range of the stress of the key geometric structure does not exceed 0.8 of the material yield strength, and the acceptable range of the total strain does not exceed 0.001.

Claims

1. A digital precision machining method for temperature and shape coordination of large rotor integral forging shafts, characterized by: The details are as follows: Step 1, obtaining crystallographic data of workpiece samples with inhomogeneous and discontinuous characteristics, the specific process is: cutting the workpiece sample, preparing the workpiece sample into a workpiece sample, detecting the workpiece sample through a scanning electron microscope, collecting the crystallographic data of the workpiece sample, and obtaining a crystallographic image of the workpiece sample, wherein, when collecting grain orientation data, the grain orientation of each crystal on the workpiece sample is obtained through EBSD data analysis, when collecting grain size data, the grain size of each crystal within a preset range on the workpiece sample is selected, and when collecting dislocation density data, the microstructure distribution of the workpiece sample is measured to obtain the dislocation density of the crystal on the workpiece sample; then, the MTEX ​​plug-in in Matlab is used to generate a script, process the obtained crystallographic data of the workpiece sample, analyze and obtain the grain orientation and position coordinates of each crystal on the workpiece sample, draw a crystal orientation map, and export the crystallographic data of the workpiece sample as an Abaqus input file in text format; Step 2, establish a workpiece crystal shaping model with non-homogeneous and non-continuous characteristics. The specific process is: establish a workpiece simulation model through finite element simulation software, divide the workpiece simulation model into hexahedral grids, import the Abaqus input file into the finite element simulation software, and map the dislocation density of the crystals on the workpiece sample, the grain orientation and grain size of each crystal to each grid node of the workpiece simulation model through the subroutine of the finite element simulation software, then define the plastic properties of the crystals on the workpiece, and then conduct a quasi-static tensile test, and combine molecular dynamics simulation to obtain the mechanical parameters of the grains and grain boundaries, and finally obtain the workpiece crystal shaping model, wherein the mechanical parameters of the grains include the elastic modulus and yield strength of the grains, and the mechanical parameters of the grain boundaries are the cohesive force model parameters of the grain boundaries; Step 3: Establish a finite element model for turning processing; Step 4: Establish a force-heat-liquid coupled cutting simulation model; Step 5: Use the cutting simulation model to conduct a cutting simulation experiment to obtain a training data set and a verification data set consisting of workpiece surface temperature field data, workpiece geometric error data, and various processing parameters; wherein the various processing parameters include tool wear, tool geometric parameters, feed rate, and cutting fluid flow rate, and the geometric error includes shape error, position error, and size error; Step 6: Taking the processing parameters and the workpiece surface temperature field data changing with cutting time as input data and the workpiece geometric error as output data, a geometric error transmission law model is established; Step 7: normalize the input data of the training data set and the verification data set, and use the normalized training data set and the verification data set to train and verify the geometric error transfer law model; Step 8: Identify key geometric structures on the workpiece; Step 9, the CNC machine tool processes the workpiece according to the set processing parameters, and the processing control system combines the processing parameters, the real-time detected processing area temperature field and the geometric error transmission law model to optimize the feed rate and cutting fluid flow rate, so that the surface temperature field of the workpiece meets the preset temperature requirements, and the geometric accuracy of each position on the workpiece meets the accuracy requirements. If the optimized feed rate and cutting fluid flow rate cannot meet the temperature and accuracy requirements, the tool is replaced to change the tool wear or tool geometric parameters; Step 10: After the workpiece is processed, measure the geometric errors at key geometric structure positions on the workpiece; Step 11: Based on the geometric errors at the key geometric structure positions measured in step 10, the dynamic balancing performance of the workpiece after processing is predicted. The specific process is as follows: construct a geometric model of the workpiece after processing, and import the measured geometric error data of the key geometric structures into the finite element software, and apply the unbalanced mass to the key geometric structure positions of the workpiece geometric model to form an unbalanced torque, wherein the process of establishing the workpiece crystal shaping model in step 2 is used to construct the geometric model of the workpiece after processing; then, boundary conditions are applied to the geometric model of the workpiece, support conditions are defined at the bearing support of the workpiece, the analysis step is set to transient dynamic analysis, and the rotation speed of the workpiece geometric model is defined. curve, and set the time step; then carry out a dynamic balancing performance simulation experiment, and output the key parameters, and evaluate the dynamic balancing performance of the workpiece according to the key parameters; if the key parameters obtained by simulation are within the corresponding preset acceptable range, the processing is completed, otherwise the key geometric structures are further refined, and the unbalanced mass of the workpiece geometric model at the key geometric structure position after fine machining is updated, and the dynamic balancing performance simulation experiment is carried out again until the dynamic balancing performance of the workpiece meets the requirements; among which, the key parameters include the displacement response of each key geometric structure, the reaction force and torque at the bearing support, the vibration spectrum of the workpiece, and the stress and strain distribution of each key geometric structure.

2. The method for digital precision machining of large rotor integral forging shaft by temperature and shape coordination according to claim 1 is characterized in that: The process of step three is: establish a tool simulation model, define the tool material properties after dividing the grid, then assemble the workpiece simulation model and the tool simulation model to obtain a turning processing finite element model, and set boundary conditions and load conditions during the assembly process, set the friction coefficient between the tool and the workpiece, and set the thermal conductivity of the tool and the workpiece; wherein the thermal conductivity of the tool is defined as a constant value, and different grain orientations of each crystal on the workpiece correspond to a thermal conductivity tensor K, and different thermal conductivity is given to each grain through the thermal conductivity tensor K, and the thermal conductivity tensor K is a 3×3 matrix: In the formula, k xx , k yy and k zz They represent the thermal conductivity along the x-axis, y-axis and z-axis respectively, k xy , k xz , k yx , k yz , k zx and k zy They represent the coupled thermal conductivity between the x-axis and the y-axis, between the x-axis and the z-axis, between the y-axis and the x-axis, between the y-axis and the z-axis, between the z-axis and the x-axis, and between the z-axis and the y-axis, respectively.

3. The method for temperature-shape coordinated digital precision machining of large rotor integral forging shaft according to claim 2 is characterized in that: The process of step four is as follows: first, construct an Euler domain model of the cooling process, divide the Euler domain model into a cutting fluid area and an empty area, define the cutting fluid material properties of the cutting fluid area, and mesh the Euler domain model, then assemble the Euler domain model with the turning processing finite element model to obtain a cutting simulation model, move the cutting fluid area to the contact area between the tool and the workpiece during the assembly process, constrain the load condition of the cutting fluid area to be consistent with the tool load condition, set the analysis step parameters and output parameters, wherein the output parameters are the temperature field data of the workpiece processing surface changing with the cutting time during the cutting simulation process and the geometric error of the workpiece after the cutting simulation.

4. The method for temperature-shape coordinated digital precision machining of large rotor integral forging shaft according to claim 1 is characterized in that: The process of step five is as follows: setting the range of machining parameters for cutting and cooling strategies, wherein the machining parameters for cutting include tool wear, tool geometric parameters and feed rate, and the machining parameter for cooling strategy is cutting fluid flow rate, selecting multiple parameter nodes within each machining parameter range, orthogonally designing the parameter nodes of each machining parameter, obtaining a machining parameter combination set consisting of a combination of different parameter nodes selected for each machining parameter, and dividing the machining parameter combination set into a training set and a verification set; performing a cutting simulation experiment using each machining parameter combination of the training set and the verification set through a cutting simulation model, obtaining the temperature field of the workpiece machining surface and the geometric error of the workpiece during the cutting simulation process under each machining parameter combination of the training set and the verification set, and then obtaining a training data set and a verification data set consisting of tool wear, tool geometric parameters, feed rate, cutting fluid flow rate, temperature field data and geometric error of the workpiece.

5. The method for temperature-shape coordinated digital precision machining of large rotor integral forging shaft according to claim 4 is characterized in that: The process of step eight is: according to the training data set and the verification data set, the first-order sensitivity index S of the shape error, position error and size error at each position on the workpiece is calculated. i (Y j ) and the overall sensitivity index S Ti (Y j ), where the first-order sensitivity index S of shape error, position error or size error is i (Y j ) is calculated as Where, X i represents all the input data of the geometric error transmission law model corresponding to the i-th data point of the workpiece surface temperature field in the training data set and the validation data set, i=1, 2, 3...n, n is the total number of data points of the workpiece surface temperature field, Y j represents the shape error, position error or size error data corresponding to the i-th data point of the workpiece surface temperature field in the training data set and the validation data set, j = 1, 2 and 3 correspond to the shape error, position error and size error respectively, Var(Y j ∣X i ) is the variance of all shape errors, position errors or size errors corresponding to the i-th data point of the workpiece surface temperature field in the training data set and the validation data set; Var(Y j ) is the total variance of all shape errors, position errors or size errors in the training data set and the validation data set, E[Y j ] is the expected value of all shape errors, position errors or size errors in the training dataset and the validation dataset; The total sensitivity index S of shape error, position error or size error Ti (Y j ) is calculated as In the formula, Divide X in the training and validation datasets i The variance of all other shape errors, position errors or size errors other than the corresponding ones; Then, the first-order sensitivity index of shape error, position error and size error at each position on the workpiece and the weighted comprehensive index of the total sensitivity index I are calculated. i : Where α and β are sensitivity weight coefficients, and α+β=1, α<β; ω1, ω2 and ω3 are error weight coefficients, and ω1+ω2+ω3=1; Then, I i The values ​​are sorted from high to low, and the three positions on the workpiece corresponding to the top three values ​​are selected as the key geometric structures.

6. The method for digital precision machining of large rotor integral forging shaft by temperature and shape coordination according to claim 5 is characterized in that: The process of step nine is as follows: the CNC machine tool processes the workpiece according to the set processing parameters, and during the processing, the temperature field of the tool and the workpiece surface processing area is detected in real time by an infrared temperature sensor placed above the tool holder, and the detected temperature field data is transmitted to the control system, and the control system predicts the geometric error at the processing position of the workpiece under the current processing parameter combination according to the processing parameters and the real-time detected temperature field data through the geometric error transfer law model; if the temperature values ​​in the real-time detected temperature field data do not exceed the preset temperature and the predicted geometric error at the processing position meets the accuracy requirement, the CNC machine tool continues to process the workpiece according to the current processing parameter combination, otherwise the feed rate and cutting fluid flow rate are optimized so that the temperature values ​​in the real-time detected temperature field data do not exceed the preset temperature and the geometric error at the processing position predicted by the geometric error transfer law model according to the optimized processing parameter combination and the real-time detected temperature field data meets the accuracy requirement; if the optimized feed rate and cutting fluid flow rate cannot meet the temperature and accuracy requirements, the tool is replaced to change the tool wear or tool geometric parameters.

7. The method for temperature-shape coordinated digital precision machining of large rotor integral forging shaft according to claim 1 is characterized in that: In the step 10, the geometric errors of the key geometric structures of the workpiece are measured by a ruby ​​probe.

Images