Self-adaptive adjustment method for supporting position of crankshaft clamping center frame

Abstract

The invention discloses a self-adaptive adjustment method for the supporting position of a crankshaft clamping center frame, and the method comprises the steps: S1, building a crankshaft clamping finite element model, and obtaining crankshaft deformation data; s2, acquiring a crankshaft simulation data set; s3, establishing an agent model of a center frame supporting position-crank arm distance difference; and S4, optimizing and adjusting the supporting position of the center frame. The problems that the crankshaft deforms in the clamping process of grinding machining and the adjusting efficiency is low are solved, and the effect that the arm distance difference of the crankshaft meets the follow-up machining condition under the condition that the position of the center frame is adjusted for a small number of times is achieved. By means of the method, the stress field and deformation, affected by the position of the center frame, of the crankshaft in the clamping process can be analyzed, and the stress concentration area, affected by clamping, of the crankshaft and the change rule of crankshaft turning gear deformation within one circle of rotation of the crankshaft are obtained. The method is suitable for fast adjustment of the position of the center frame in the grinding and clamping process of crankshafts of other models, and control over the arm distance difference of the crankshafts is achieved.

Description

Technical Field

[0001] The present invention relates to an adaptive adjustment method for the support position of a crankshaft clamping steady rest, and belongs to the field of machining. Background Art

[0002] The crankshaft is a key component in a diesel engine. Its overall structure is relatively complex and asymmetric. Among the key components of a diesel engine, the machining quality of the crankshaft is the most difficult to guarantee. In actual grinding machining, due to the action of its own gravity and the clamping force of the grinding machine fixture, the crankshaft is prone to complex elastic deformation, seriously affecting the grinding accuracy of the crankshaft. In order to reduce the elastic deformation of the crankshaft and ensure the coaxiality of the main journals, a steady rest is required to support the main journals of the crankshaft during grinding, and the support position of the steady rest is adjusted according to the deformation of the crankshaft.

[0003] Regarding the problem of clamping deformation and adjustment of the crankshaft during grinding machining, in the actual grinding machining of large marine crankshafts, the deformation of the crankshaft is usually measured by the magnitude of the arm deflection difference. A dial gauge is used to measure the arm deflection difference of the crankshaft, and then the support position of the steady rest is repeatedly adjusted according to the production experience of workers so that the arm deflection difference is controlled within the processing requirement range. However, the existing method requires a large amount of manpower, and the adjustment efficiency is too low to achieve rapid adjustment of the steady rest position. Summary of the Invention

[0004] Object of the Invention: Aiming at the deficiencies existing in the prior art, the present invention provides an adaptive adjustment method for the support position of a crankshaft clamping steady rest. Through the optimization and iterative adjustment of the steady rest position, the present invention realizes the rapid adjustment of the support position of the crankshaft clamping steady rest, controls the deformation generated during the clamping of the crankshaft within the processing requirement range, and improves the production efficiency at the same time.

[0005] Technical Solution: The adaptive adjustment method for the support position of a crankshaft clamping steady rest includes the following steps:

[0006] S1. Establish a finite element model of crankshaft clamping: According to the size structure of the crankshaft and the structure of the steady rest, establish a finite element model of crankshaft clamping, analyze the deformation of the crankshaft during the clamping adjustment process and the steady rest support position adjustment process, and obtain crankshaft deformation data, that is, the deformation amount at each crank web of each crank throw of the crankshaft;

[0007] S2. Obtain a crankshaft simulation data set: Process the deformation data in S1 to obtain the arm deflection difference data of each crank throw of the crankshaft and the corresponding steady rest support position data, and establish a simulation data set;

[0008] S3. Establish a surrogate model for the center support position - crankshaft arm difference: Establish an arm difference prediction model based on the BP neural network, set the hyperparameters of the arm difference prediction model, and train the arm difference prediction model with the simulation data set in S2. Use the center support position as the input and the crank throw arm difference as the output to obtain a surrogate model for the center support position - crankshaft arm difference for predicting the crankshaft arm difference.

[0009] S4. Optimization and adjustment of the center support position: Based on the surrogate model of the center support position - crankshaft arm difference trained in S3, use the NSGA-II multi-objective genetic algorithm to establish an inverse solution optimization algorithm for the center support position, and iteratively adjust the support position of the center support based on the measured crank throw arm difference data of the crankshaft using this inverse solution optimization algorithm.

[0010] Preferred option, where S1 includes:

[0011] S101. Analyze the structural characteristics of the crankshaft, establish a finite element model for the crankshaft clamping according to the crankshaft size structure and the center support structure, and simplify the center support structure into a V-block structure for convenient simulation analysis;

[0012] S102. Set the material parameters of the crankshaft clamping finite element model: the density, Young's modulus, Poisson's ratio, and yield strength of the crankshaft;

[0013] S103. Perform mesh division on the crankshaft clamping finite element model;

[0014] S104. Set the boundary conditions of the crankshaft according to the actual clamping situation of the crankshaft to obtain the initial crankshaft clamping finite element model.

[0015] S105. Based on the initial crankshaft clamping finite element model established in S104, parameterize the model and solve it to obtain the crankshaft deformation data under different center support positions of the crankshaft.

[0016] Preferred option, where S103 is specifically:

[0017] Adopt different mesh density division strategies for different regions. Both the crankshaft and the center support adopt tetrahedral element meshes, and set the corresponding node sets for extracting the crankshaft deformation data:

[0018] The regions that need to adopt mesh density division include the surface region where the center support contacts the crankshaft, the region where the center support does not contact the crankshaft, and the region where the crankshaft does not contact the center support;

[0019] The mesh size of the surface region where the center support contacts the crankshaft is 0.5 - 1 mm, the mesh size of the region where the center support does not contact the crankshaft is 10 - 20 mm, and the mesh size of the region where the crankshaft does not contact the center support is 20 - 30 mm;

[0020] After the meshing is completed, each region contains a number of elements, and each element contains a number of nodes. The midpoints of the lowest edges of all the crank arms of the crankshaft are taken to form a node set among the nodes. By processing the displacement changes of the nodes in the node set, the deformation difference at each crank arm of the crankshaft is obtained.

[0021] Preferred option, the specific setting of the boundary conditions of the crankshaft according to the actual clamping situation of the crankshaft in S104 is as follows:

[0022] S1041. Apply a fully fixed constraint condition to the center hole at one end of the crankshaft, and apply an axial tightening force to the other end;

[0023] S1042. Move the center rest to the set position, apply a fully fixed constraint condition to the center rest, and adjust the tightening force applied to the crankshaft;

[0024] S1043. Constrain the movement and rotation of the center holes at both ends of the crankshaft, and rotate the crankshaft one week;

[0025] The gravity condition is applied to the above actual clamping process of the crankshaft, that is, the gravitational acceleration is 9.8m / s 2 .

[0026] Preferred option, the S105 includes:

[0027] S1051. Determine the adjustment range of the support position of the center rest. According to the settings of S101 - S104, perform a clamping simulation analysis on the crankshaft without support to obtain the crankshaft simulation deformation result;

[0028] S1052. Extract the deformation amount of the crankshaft at the required support position of the center rest in the simulation result, and use this as the lower limit H min of the support position of the center rest, and the upper limit of the support position of the center rest is set to H max = 0;

[0029] S1053. Obtain different center rest support positions from the actual clamping situation of the crankshaft. Within the range of the center rest support position, that is, H min ≤ H ≤ H max , select different combinations of center rest support position schemes that meet the range, complete the parameterization of the initial crankshaft clamping finite element model, and solve to obtain the crankshaft deformation data under different center rest support positions of the crankshaft.

[0030] Preferred option, the S1053 is specifically: select the adjustment amount δh of the center rest support position < 0.1H max, ensure that the simulation results can reflect the influence of the deviation of the supporting height of different steady rests on the crankshaft arm deflection. The Latin Hypercube Sampling (LHS) method is used to select different combinations of steady rest support positions. By equally dividing the value range of each variable into non - overlapping intervals with equal probability and randomly selecting samples in each interval, the formula is as follows

[0031]

[0032] In the formula, i=(1, 2, …n) is the number of variables, that is, the number of steady rests; j=(1, 2, …m) is the sampling dimension, that is, the j - th sampling; P i -1 is the probability distribution function; α ij is a random number within the range of [0, 1]; N is the number of samples, that is, the number of different combinations of steady rest support positions;

[0033] Finally, N different combinations of steady rest support positions are selected to complete the parameterization of the initial crankshaft clamping finite - element model.

[0034] Preferred option, specifically, S2 is as follows:

[0035] Extract the crankshaft deformation data obtained from the S1 simulation. Extract the axial displacement U3 of each node in the S1 node set. By processing the node data at different rotation angles, obtain the arm deflection of each crank of the crankshaft. Organize the arm deflection data of each crank and the corresponding steady rest adjustment position data under different actual crankshaft clamping conditions into a simulation data set.

[0036] Preferred option, specifically, S3 is as follows:

[0037] Establish a BP neural network model. Take the number of steady rests and the number of crankshaft cranks as the input and output of the BP neural network model respectively. Randomly select data from the simulation data set in a ratio of 7:1:2 as the training set, validation set, and test set for evaluating the steady rest support position - crankshaft arm deflection surrogate model of the BP neural network model. Set the hyperparameters of the BP neural network model. The hyperparameters include the maximum number of iterations, learning rate, and minimum training error. At the same time, set the structure of the BP neural network model, train the training set. The trained model is used for predicting the crankshaft arm deflection and provides a basis for solving the optimal position of the steady rest. The trained steady rest support position - crankshaft arm deflection surrogate model is:

[0038]

[0039] is the predicted crankshaft arm deflection, net is the trained BP neural network model, and H is the steady rest support position.

[0040] Preferred option, S4 includes:

[0041] S401. Based on the surrogate model of the center support position - crankshaft arm distance difference trained in S3, bring it into the NSGA-II optimization algorithm, and add the condition to ensure that the sum of the distances between the actual support positions of the current center support and the crankshaft rotation axis is minimized, and complete the establishment of the reverse solution optimization algorithm for the center support position;

[0042] S402. Actually measure the arm distance difference of each crank of the crankshaft, and judge whether the actual crankshaft arm distance difference meets the processing conditions. If not, input the actual crankshaft arm distance difference into the reverse solution optimization algorithm for the center support position, obtain the actual support position of the current center support, adjust the actual support position of the current center support, measure the arm distance difference data of each crank again after adjustment. If it does not meet the processing conditions, continue to input it into the reverse solution optimization algorithm for iterative adjustment until the actual crankshaft arm distance difference meets the actual processing conditions.

[0043] Preferred option, S401 is specifically:

[0044] First, establish the mathematical problem of the center support position optimization problem, minimize the absolute error between the measured crankshaft arm distance difference and the predicted crankshaft arm distance difference, and at the same time minimize the sum of the absolute values of the center support positions corresponding to the predicted arm distance differences;

[0045] Next, set the constraint conditions, that is, set the upper and lower limits of the center support position adjustment range, and complete the establishment of the center support position optimization problem;

[0046] H min and H max are respectively the lower and upper limits of the center support position adjustment, D is the actually measured arm distance difference of the crankshaft, is the predicted crankshaft arm distance difference, H is the center support position, f 1 (D) is the sum of the crankshaft arm distance difference prediction errors, and f 2 (H) is the sum of the center support positions;

[0047] Finally, set the relevant parameters of the NSGA-II multi-objective genetic algorithm, including the population size, the maximum number of iterations, the crossover probability, and the mutation probability, and bring the center support position optimization problem into the NSGA-II multi-objective genetic algorithm to complete the establishment of the reverse solution optimization algorithm for the center support position.

[0048] Beneficial effects: The present invention solves the problems of deformation during the clamping of the crankshaft in grinding and low adjustment efficiency, and realizes that the arm distance difference of the crankshaft meets the subsequent processing conditions with fewer adjustments of the center rest position. By using this method, the stress field and deformation of the crankshaft affected by the center rest position during the clamping process can be analyzed, and the stress concentration area of the crankshaft affected by the clamping and the change law of the crankshaft deflection deformation within one rotation of the crankshaft can be obtained. The method proposed by the present invention is applicable to the rapid adjustment of the center rest position during the grinding clamping of other types of crankshafts, and realizes the control of the arm distance difference of the crankshaft. Description of the drawings

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to the provided drawings.

[0050] Figure 1 is the method flow chart of the present invention;

[0051] Figure 2 (a)-(b) of are respectively the three-dimensional model and main dimension diagram of the crankshaft in the present invention;

[0052] Figure 3 is the crankshaft axis deviation diagram in the present invention under the condition of no center rest support;

[0053] Figure 4 (a)-(i) of are the diagrams of the change of the arm distance values of crank throws 1 to 9 of the crankshaft at different support positions within one rotation in the present invention;

[0054] Figure 5 is the mean square error training result diagram of the crankshaft arm distance difference BP model in the present invention;

[0055] Figure 6 is the regression result diagram of the crankshaft arm distance difference BP model in the present invention;

[0056] Figure 7 is the Pareto optimal solution set diagram of the center rest position adjustment scheme obtained by the NSGA-II non-dominated sorting genetic algorithm in the present invention;

[0057] Figure 8 is the change diagram of the crankshaft arm distance difference within 3 adjustments of the center rest position in the present invention. Detailed implementation manners

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0059] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0060] In the present invention, unless otherwise clearly defined and limited, the first feature being "on" or "under" the second feature may include the direct contact between the first and second features, or may include the situation where the first and second features are not in direct contact but in contact through other features therebetween. Moreover, the first feature being "above", "over" and "on" the second feature includes that the first feature is directly above and obliquely above the second feature, or simply means that the horizontal height of the first feature is higher than that of the second feature. The first feature being "under", "beneath" and "under" the second feature includes that the first feature is directly below and obliquely below the second feature, or simply means that the horizontal height of the first feature is lower than that of the second feature.

[0061] As Figure 1 shown, an adaptive adjustment method for the support position of the crankshaft clamping center rest. This method is applied to the crankshaft model as Figure 2 shown, analyzes the deformation of the crankshaft during grinding clamping, and controls the deformation of the crankshaft within the machining range through the adjustment of the center rest. It includes the following steps:

[0062] S1. Establish a finite element model of crankshaft clamping: Establish a finite element model of crankshaft clamping according to the crankshaft size structure and the center rest structure, analyze the deformation of the crankshaft during the clamping adjustment process and the center rest support position adjustment process, and obtain the crankshaft deformation data, that is, the deformation amount at each crank throw crank arm of the crankshaft.

[0063] S101. Analyze the structural characteristics of the crankshaft, establish a finite element model of crankshaft clamping according to the crankshaft size structure and the center rest structure, and simplify the center rest structure into a V-block structure for simulation analysis; in this embodiment, the minute structural characteristics of the crankshaft are removed, such as chamfers, fillets, small holes and oil holes, etc., so as to improve the simulation efficiency and the accuracy of the simulation results.

[0064] S102. Set the material parameters of the finite element model for crankshaft clamping: the density, Young's modulus, yield strength, and Poisson's ratio of the crankshaft;

[0065] Table 1 Mechanical and Physical Property Parameters of 42CrMo Alloy Steel

[0066]

[0067] S103. Mesh the finite element model of crankshaft clamping:

[0068] Adopt different mesh density division strategies for different regions. Tetrahedral element meshes are used for both the crankshaft and the center rest, the element type is set to C3D10, and corresponding node sets are set for extracting the deformation data of the crankshaft:

[0069] The regions that need to adopt mesh density division include the surface region where the center rest contacts the crankshaft, the region where the center rest does not contact the crankshaft, and the region where the crankshaft does not contact the center rest;

[0070] The mesh size of the surface region where the center rest contacts the crankshaft is 0.5 - 1 mm, the mesh size of the region where the center rest does not contact the crankshaft is 10 - 20 mm, and the mesh size of the region where the crankshaft does not contact the center rest is 20 - 30 mm;

[0071] After the meshing is completed, each region contains several elements, and each element contains several nodes. The midpoints of the lowest edges of all crank throws and crank arms in several nodes are used to form a node set. By processing the displacement changes of the nodes in the node set, the deformation difference at each crank throw and crank arm of the crankshaft is obtained.

[0072] S104. Set the boundary conditions of the crankshaft according to the actual clamping situation of the crankshaft to obtain the initial finite element model of crankshaft clamping.

[0073] The specific setting of the boundary conditions of the crankshaft according to the actual clamping situation of the crankshaft in S104 is as follows:

[0074] S1041. Apply a fully fixed constraint condition to the center hole at one end of the crankshaft, and apply an axial tightening force of 20 kN to the other end;

[0075] S1042. Move the center rest to the set position, apply a fully fixed constraint condition to the center rest, and adjust the tightening force applied to the crankshaft to 7.5 kN;

[0076] S1043. Constrain the movement in the X, Y, and Z directions and the rotation in the X and Y directions at the center holes at both ends of the crankshaft, and rotate the crankshaft one week;

[0077] Apply the gravity condition to the above actual clamping process of the crankshaft, that is, the gravitational acceleration is 9.8 m / s 2The boundary conditions are fixed constraint conditions, clamping force, crankshaft rotation, and movement.

[0078] S105. Based on the initial finite element model of the crankshaft clamping established in S104, parameterize the model and solve it to obtain the crankshaft deformation data under different center rest support positions of the crankshaft.

[0079] S1051. Determine the adjustment range of the center rest support position. Conduct a clamping simulation analysis of the crankshaft without support according to the settings in S101 - S104 to obtain the crankshaft simulation deformation results.

[0080] S1052. Extract the deformation amount of the crankshaft at the required support positions of the center rest from the simulation results, and use this as the lower limit H of the center rest support position. min The upper limit of the center rest support position is set to H max = 0;

[0081] Conduct a simulation analysis of the crankshaft's force - induced deformation without the center rest support. By processing the simulation results, extract the deformation amounts of 24 equally - spaced nodes in the gravity direction along the crankshaft axis. As Figure 3 shown, when there is no center rest support, the middle part of the crankshaft deforms the most under the action of gravity, and the axis is bent into a downward arc. The deformation amounts H min at the four support positions where the center rests are - 0.314, - 0.668, - 0.672, and - 0.328 mm respectively, and the deformation is approximately symmetrically distributed about the mid - point of the crankshaft.

[0082] S1053. Obtain different center rest support positions according to the actual clamping situation of the crankshaft. Within the range of the center rest support position, that is, H min ≤H≤H max , select different combinations of center rest support position schemes that meet the range. Generate simulation inp files for multiple combinations of different center rest positions by writing a python script to complete the parameterization of the initial finite element model of the crankshaft clamping, and solve to obtain the crankshaft deformation data under different center rest support positions of the crankshaft:

[0083] Select the center rest support position adjustment amount δh < 0.1H max , ensure that the simulation results can reflect the influence of different center rest support height deviations on the crankshaft arm difference. Select different combinations of center rest support position schemes through the Latin Hypercube Sampling (LHS) method. By equally - probabilistically dividing the value range of each variable into non - overlapping intervals and randomly selecting samples in each interval, the formula is as follows

[0084]

[0085] where \(i=(1,2,\cdots,n)\) is the number of variables, i.e., the number of steady rests; \(j=(1,2,\cdots,m)\) is the sampling dimension, i.e., the \(j\)-th sampling; \(P\) i -1 is the probability distribution function; \(\alpha\) ij is a random number within the value range \([0,1]\); \(N\) is the number of samples, i.e., the number of combination schemes of the support positions of different steady rests;

[0086] Finally, \(N\) groups of combination schemes of the support positions of different steady rests are selected to parameterize the initial finite element model of the crankshaft clamping.

[0087] S2. Obtain the crankshaft simulation data set: By establishing a Python script to process the deformation data in S1, the arm distance difference data of each crank of the crankshaft and the corresponding support position data of the steady rest are obtained, and a simulation data set is established:

[0088] Through the secondary development content of ABAQUS, a python post-processing script is written to extract the crankshaft deformation data in the ODB result file simulated in S1, extract the axial displacement \(U3\) of each node in the S1 node set, and calculate the arm distance difference of each crank of the crankshaft by taking the difference between the nodes on the corresponding two crank arms of the same crank. The arm distance difference data of each crank and the corresponding center rest adjustment position data under different actual clamping conditions of the crankshaft are sorted into a simulation data set.

[0089] In this embodiment, the crankshaft simulation deformation results under 3 groups of different steady rest support schemes are selected for processing. As Figure 4 shown, Figures (a)-(i) sequentially show the changes in the arm distance values of crank 1 to crank 9 within one revolution under 3 support schemes. The arm distance value of the crankshaft changes the most without the support of the steady rest; the arm distance value of the crankshaft changes less when the support positions of the 4 steady rests deviate from the crankshaft axis by 0.25, 0.5, 0.5, and 0.25 mm in sequence; the arm distance value of the crankshaft changes the least when it deviates from the crankshaft axis by 0.1, 0.2, 0.2, and 0.1 mm, indicating that the crankshaft axis deviation is positively correlated with the deformation of each crank.

[0090] Among them, crank 1 and 5 are most affected by the axis deviation and the arm distance value changes up to 0.12 mm without the support of the steady rest; the arm distance value changes of crank 6 and 7 are close to 0.09 mm; the arm distance value changes of crank 3, 4, 8, and 9 are within 0.03 - 0.08 mm; the arm distance value change of crank 2 is the smallest, within 0.03 mm. It can be seen from the figure that within one revolution, the change in the arm distance value of the crankshaft is similar to a sine wave. Due to the angle between each crank, there is a certain phase angle deviation in the change of the arm distance value of different cranks.

[0091] The difference between the maximum and minimum values of the crankshaft arm distance within one revolution is the arm distance difference. Process the arm distance values of the crankshaft under 200 sets of center rest support schemes obtained, calculate the corresponding crankshaft arm distance difference data, and organize the data into a data set for subsequent prediction of the crankshaft arm distance difference and adaptive adjustment of the center rest.

[0092] S3. Establish a surrogate model of the center rest support position - crankshaft arm distance difference: Establish a prediction model of the arm distance difference based on the BP neural network, set the hyperparameters of the arm distance difference prediction model, and train the arm distance difference prediction model in combination with the simulation data set in S2. Take the center rest support position as the input and the crank throw arm distance difference as the output to obtain a surrogate model of the center rest support position - crankshaft arm distance difference for predicting the crankshaft arm distance difference:

[0093] Establish a BP neural network model, and use 4 center rests and 9 crank throws of the crankshaft as the number of neurons in the input and output layers of the BP neural network model respectively.

[0094] Divide the 200 sets of data constructed by simulation into three parts: training, validation, and testing. The training samples account for 70% of the samples in the simulation data set, and 140 sets are used for the initial training and tuning of the network model. The validation samples are 20 sets, accounting for 10%, and are used for optimizing the model parameters and adjusting the hyperparameters during the training process. The test samples include 40 sets and are used to evaluate the prediction performance of the model. The maximum number of iterations of the BP neural network model is 1000 times, the learning rate is 0.001, and the minimum training error is set to 1e - 6 to ensure the convergence and stability of the training.

[0095] At the same time, set the structure of the BP neural network model. In this embodiment, set the number of layers of the BP neural network model to 4 layers. The first layer is the input layer, which is used to receive the data of the center rest support position. The number of neurons in this layer is equal to the number of center rests. The second and third layers are hidden layers, and the number of neurons is set in the range of 13 to 21. Find the combination of the number of neurons with the smallest training error within this range, such as 15 - 19. The fourth layer is the output layer, which is used to output the predicted crank throw arm distance difference data. The number of neurons in this layer is equal to the number of crank throws of the crankshaft;

[0096] Train the training set. The trained model is used for predicting the crankshaft arm distance difference and provides a basis for solving the optimal position of the center rest at the same time. Use the test set and select the mean square error as the main index to measure the deviation between the predicted value and the actual observed value. Figure 5The training results are shown, presenting the error changes at different iteration stages, as well as the validation and test performances. As the number of model iterations increases, the MSE corresponding to the validation curve continuously decreases. After training, when the number of training times is 29, the mean square error of the validation set reaches the minimum, with a value of 0.013213, and the training stops. Regression analysis is used to evaluate the correlation between the training results of the BP neural network and the target values. The correlation coefficient of the BP neural network model on the training samples is 0.9963, on the validation samples is 0.98371, on the test samples is 0.98705, and the overall correlation coefficient is 0.99334, indicating that the training model has very good effects. The results are shown in Figure 6 。

[0097] The surrogate model of the center rest support position - crankshaft arm distance difference after training is:

[0098]

[0099] is the predicted crankshaft arm distance difference, net is the trained BP neural network model, and H is the center rest support position.

[0100] S4. Optimization and adjustment of the center rest support position: Based on the surrogate model of the center rest support position - crankshaft arm distance difference trained in S3, use the NSGA-II multi-objective genetic algorithm to establish an inverse solution optimization algorithm for the center rest support position, and based on this inverse solution optimization algorithm, iteratively adjust the support position of the center rest through the measured crankshaft crank arm distance difference data;

[0101] S401. Based on the surrogate model of the center rest support position - crankshaft arm distance difference trained in S3, bring it into the NSGA-II optimization algorithm, and add the condition of ensuring the minimum sum of the distances between the actual support position of the current center rest and the crankshaft rotation axis to prevent over-adjustment of the center rest, and complete the establishment of the inverse solution optimization algorithm for the center rest support position:

[0102] First, establish the mathematical problem of the center rest support position optimization problem, minimize the absolute error between the measured crankshaft arm distance difference and the predicted crankshaft arm distance difference, and at the same time minimize the sum of the absolute values of the center rest positions corresponding to the predicted arm distance differences;

[0103] Next, set the constraint conditions, that is, set the upper and lower limits of the center rest position adjustment range, and complete the establishment of the center rest support position optimization problem;

[0104] H min and H max are respectively the lower and upper limits of the center rest position adjustment, D is the actually measured arm distance difference of the crankshaft, is the predicted crankshaft arm deflection, H is the support position of the steady rest, f 1 (D) is the sum of the prediction errors of the crankshaft arm deflections, f 2 (H) is the sum of the support positions of the steady rest;

[0105] Finally, set the relevant parameters of the NSGA-II multi-objective genetic algorithm. The model parameter settings include a population size of 200, a maximum number of iterations of 300, a crossover probability set to 0.9, and a mutation probability set to 0.05. Then, bring the steady rest support position optimization problem into the NSGA-II multi-objective genetic algorithm to complete the establishment of the reverse solution optimization algorithm for the steady rest support position.

[0106] Verify the reverse solution optimization algorithm for the steady rest support position through simulation analysis:

[0107] First, randomly initialize a combination of steady rest support positions, and obtain the crankshaft arm deflections under the current steady rest support position scheme through simulation analysis. Then, input the obtained crankshaft arm deflections into the optimization algorithm, and use the algorithm to solve for the support positions of the steady rest when the crankshaft is in the current deformed state. Finally, find a set of solutions in the obtained Pareto solution set to adjust the steady rest positions. This set of solutions needs to meet the following conditions: (1) The error between the predicted crankshaft arm deflections under this set of steady rest support schemes and the actually measured crankshaft arm deflections is the smallest; (2) The absolute value of the sum of the distances from this set of steady rest support positions to the crankshaft axis is the smallest, that is, the adjustment increment required for the steady rest positions is the smallest.

[0108] S402. Actually measure the arm deflections of each crank of the crankshaft, and judge whether the actual crankshaft arm deflections meet the machining conditions. If not, input the actual crankshaft arm deflections into the reverse solution optimization algorithm for the steady rest support position to obtain the actual support positions of the current steady rest, adjust the actual support positions of the current steady rest, and then measure the arm deflection data of each crank again after adjustment. If it still does not meet the machining conditions, continue to input it into the reverse solution optimization algorithm for iterative adjustment until the actual crankshaft arm deflections meet the actual machining conditions.

[0109] Through the above algorithm process, measure and adjust the deformations under a given set of steady rest support schemes. After 3 adjustments in all cases, the deviation of the crankshaft axis is adjusted within 0.01 mm. As Figure 7 shown, the Pareto solution set of 3 adjustments is obtained using the NSGA-II algorithm. As the number of adjustments increases, the Pareto solution set gradually decreases and approaches 0, reflecting that the deviation of the support position of the steady rest from the crankshaft axis becomes smaller and smaller during the continuous adjustment process.

[0110] As Figure 8As shown, the adjustment process of one set of the center rest support positions is selected for display. After selecting the optimal adjustment scheme, the position of the center rest is adjusted, and the arm distance difference of the 9 crank pins of the crankshaft is reduced to within 0.02 mm within 3 iterative adjustments and remains stable in subsequent adjustments. The specific changes in the arm distance difference of each crank pin are shown in Table 2. After the first adjustment, only the arm distance difference of crank pin 8 does not meet the subsequent processing conditions. After the second adjustment, the arm distance differences of all crank pins meet the requirements. In addition, it can be seen from the third adjustment that the arm distance difference of the crankshaft tends to be stable, and the maximum change is 0.4 μm, which reflects the feasibility of the proposed method.

[0111] Table 2. Arm distance differences of the crankshaft during iterative adjustment

[0112]

[0113] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The same or similar parts among the embodiments can be referred to each other. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple. For the relevant parts, please refer to the description of the method part.

[0114] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for adaptively adjusting the support position of a crankshaft clamping center frame, characterized in that: The following steps are involved: S1. Establishing a finite element model for crankshaft clamping: establishing a finite element model for crankshaft clamping according to the crankshaft size structure and the center frame structure, analyzing the crankshaft deformation during the clamping adjustment process and the center frame support position adjustment process, and obtaining the crankshaft deformation data, that is, the deformation amount of each crank arm of the crankshaft; S2. Acquire a crankshaft simulation data set: process the deformation data in S1, obtain the arm distance difference data of each crankshaft and the corresponding center frame support position data, and establish a simulation data set; S3, establishing a proxy model of center frame support position-crankshaft arm distance difference: establishing an arm distance difference prediction model based on BP neural network, setting hyper parameters of the arm distance difference prediction model, training the arm distance difference prediction model in combination with the simulation data set in S2, taking the center frame support position as input and the crank arm distance difference as output, and obtaining a proxy model of center frame support position-crankshaft arm distance difference for crankshaft arm distance difference prediction; S4. Optimization and adjustment of the center frame support position: Based on the proxy model of the center frame support position-crankshaft arm distance difference trained in S3, the NSGA-II multi-objective genetic algorithm is used to establish a reverse solution optimization algorithm for the center frame support position, and based on the reverse solution optimization algorithm, the support position of the center frame is iteratively adjusted through the measured crankshaft arm distance difference data.

2. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 1, characterized in that: The S1 includes: S101, analyzing the structural characteristics of the crankshaft, establishing a crankshaft clamping finite element model according to the crankshaft size structure and the center frame structure, and simplifying the center frame structure into a V-shaped block structure for easy simulation analysis; S102, setting material parameters of the crankshaft clamping finite element model: density, Young's modulus, Poisson's ratio, and yield strength of the crankshaft; S103, meshing the crankshaft clamping finite element model; S104, setting the boundary conditions of the crankshaft according to the actual clamping condition of the crankshaft to obtain an initial crankshaft clamping finite element model; S105. According to the initial crankshaft clamping finite element model established in S104, the model is parameterized and solved to obtain the crankshaft deformation data under different center frame support positions of the crankshaft.

3. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 2, characterized in that: The S103 is specifically: Different mesh density division strategies are adopted for different areas. Tetrahedral unit meshes are used for both the crankshaft and the center frame, and corresponding node sets are set for extracting the crankshaft deformation data: The areas that need to be divided by grid density include the surface area where the center frame contacts the crankshaft, the area where the center frame does not contact the crankshaft, and the area where the crankshaft does not contact the center frame; The mesh size of the surface area where the center frame contacts the crankshaft is 0.5-1 mm, the mesh size of the area where the center frame does not contact the crankshaft is 10-20 mm, and the mesh size of the area where the crankshaft does not contact the center frame is 20-30 mm; After the mesh division is completed, each area contains several units, each unit contains several nodes, and the midpoints of the lowest edges of all crank arms among the several nodes are taken to form a node set. By processing the displacement changes of the nodes in the node set, the deformation difference at each crank arm of the crankshaft is obtained.

4. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 2, characterized in that: In S104, the boundary conditions of the crankshaft are set according to the actual clamping condition of the crankshaft as follows: S1041. Apply a complete fixing constraint to the center hole at one end of the crankshaft, and apply a tightening force in the axial direction to the other end; S1042, moving the center frame to a set position, applying a completely fixed constraint condition to the center frame, and adjusting the tightening force applied to the crankshaft; S1043, restricting the movement and rotation of the center holes at both ends of the crankshaft, and the crankshaft rotates one circle; The actual clamping process of the above crankshafts is subjected to gravity conditions, that is, the gravity acceleration is 9.8m / s 2 .

5. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 2, characterized in that: The S105 includes: S1051, determining the adjustment range of the center frame support position, and performing a clamping simulation analysis on the crankshaft without support according to the settings of S101-S104 to obtain a crankshaft simulation deformation result; S1052, extract the deformation of the crankshaft at the required support position of the center frame in the simulation result, and use it as the lower limit H of the support position of the center frame min , the upper limit of the center frame support position is set to H max =0; S1053, obtain different center frame support positions according to the actual clamping conditions of the crankshaft, within the range of the center frame support position, i.e. H min ≤H≤H max , select different center frame support position combination schemes that meet the range, complete the parameterization of the initial crankshaft clamping finite element model, and solve the crankshaft deformation data under different center frame support positions.

6. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 5, characterized in that: The S1053 is specifically as follows: select the center frame support position adjustment amount δh<0.1H max To ensure that the simulation results can reflect the influence of different center frame support height deviations on the crankshaft arm distance difference, different center frame support position combination schemes are selected by Latin hypercube sampling LHS method. The value range of each variable is divided into non-intersecting intervals with equal probability, and samples are randomly selected in each interval. The formula is as follows Where i = (1, 2, ... n) is the number of variables, that is, the number of central frames; j = (1, 2, ... m) is the sampling dimension, that is, the jth sampling; P i -1 is the probability distribution function; α ij is a random number in the range [0,1]; N is the number of samples, that is, the number of different center frame support position combination schemes; Finally, the selection of N groups of different center frame support position combination schemes is completed to complete the parameterization of the initial crankshaft clamping finite element model.

7. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 3, characterized in that: The S2 is specifically: The crankshaft deformation data obtained by S1 simulation is extracted, and the axial displacement U3 of each node in the S1 node set is extracted. The arm distance difference of each crank of the crankshaft is obtained by processing the node data at different rotation angles. The arm distance difference data of each crank of different actual crankshaft clamping conditions and the corresponding center frame adjustment position data are organized into a simulation data set.

8. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 1, characterized in that: The S3 is specifically: A BP neural network model was established, and the number of center frames and the number of crankshafts were used as the input and output of the BP neural network model respectively. Data were randomly selected in the simulation data set at a ratio of 7:1:2 as the training set, validation set and test set for evaluating the center frame support position-crankshaft arm distance difference proxy model of the BP neural network model. The hyperparameters of the BP neural network model were set, including the maximum number of iterations, learning rate and minimum training error. At the same time, the structure of the BP neural network model was set, and the training set was trained. The trained model was used to predict the crankshaft arm distance difference and provide a basis for solving the optimal position of the center frame. The proxy model of the center frame support position-crankshaft arm distance difference after training was: is the predicted crankshaft arm distance difference, net is the trained BP neural network model, and H is the center frame support position.

9. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 1, characterized in that: The S4 includes: S401, based on the proxy model of the center frame support position-crankshaft arm distance difference trained in S3, bring it into the NSGA-Ⅱ optimization algorithm, and add the condition of ensuring that the sum of the distances between the actual support position of the current center frame and the crankshaft rotation axis is minimized, and complete the establishment of the reverse solution optimization algorithm for the center frame support position; S402. Actual measurement is performed on the arm distance difference of each crankshaft to determine whether the actual crankshaft arm distance difference meets the processing conditions. If not, the actual crankshaft arm distance difference is input into the reverse solution optimization algorithm of the center frame support position to obtain the actual support position of the current center frame, and the actual support position of the current center frame is adjusted. After the adjustment, the arm distance difference data of each crank is measured again. If it does not meet the processing conditions, it continues to be input into the reverse solution optimization algorithm for iterative adjustment until the actual crankshaft arm distance difference meets the actual processing conditions.

10. The method for adaptively adjusting the support position of the crankshaft clamping center frame according to claim 8, characterized in that: The S401 is specifically as follows: Firstly, the mathematical problem of the optimization of the center frame support position is established to minimize the absolute error between the measured crankshaft arm distance difference and the predicted crankshaft arm distance difference, and at the same time minimize the sum of the absolute values ​​of the center frame positions corresponding to the predicted arm distance difference; Next, set the constraint conditions, that is, set the upper and lower limits of the center frame position adjustment range, and complete the establishment of the center frame support position optimization problem; H min and H max are the lower and upper limits of the center frame position adjustment, respectively, and D is the actual measured arm distance difference of the crankshaft. is the predicted crankshaft arm distance difference, H is the center frame support position, f1(D) is the crankshaft arm distance difference prediction error and, f2(H) is the center frame support position and; Finally, the relevant parameters of the NSGA-II multi-objective genetic algorithm are set, including population size, maximum number of iterations, crossover probability and mutation probability, and the center frame support position optimization problem is introduced into the NSGA-II multi-objective genetic algorithm to complete the establishment of the reverse solution optimization algorithm of the center frame support position.

Images