A method for optimizing multi-process CNC grinding machines to balance accuracy and energy consumption

Abstract

This invention discloses a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption, relating to the field of CNC grinding machine process optimization. The method establishes a first relational model characterizing the relationship between process parameters and axial error. This first relational model reflects the relationship between process parameters and the axial error of the CNC grinding machine during actual machining under those process parameters, enabling high-precision prediction of axial error in the actual machining process. A second relational model is also established to reflect the relationship between process parameters and grinding machine energy consumption. Furthermore, multi-objective optimization is performed with axial error and grinding machine energy consumption as objectives, achieving the goal of suppressing spindle axial error while improving resource utilization efficiency and reducing production costs.

Description

Technical Field

[0001] This invention relates to the field of CNC grinding machine process optimization, and in particular to a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption. Background Technology

[0002] With the emergence of a new wave of industrial revolution, intelligent manufacturing technology is leading the progress and development of industrial manufacturing. CNC grinding machines, with their high technological content and precision, are widely used in the manufacturing of precision parts in aerospace, defense, and other fields. However, as users' requirements for product quality continue to increase and their emphasis on low-energy manufacturing concepts grows, improving the machining accuracy and performance of CNC grinding machines has become extremely urgent. Among the many factors affecting the machining accuracy of CNC grinding machines, the spindle axial error caused by factors such as bearing friction heat accounts for a very high proportion; and among the various performance indicators of CNC grinding machines, effective control of machining power is key to achieving low-power manufacturing. Therefore, improving the existing CNC grinding machine process parameters to suppress spindle axial elongation while controlling energy consumption is crucial to improving the machining accuracy of CNC grinding machines and achieving precision low-energy manufacturing.

[0003] Currently, there is relatively little research on suppressing spindle axial elongation by improving CNC grinding machine process parameters. The common method is to pre-establish a mapping relationship between several temperature measurement points and axial error, and collect temperature data near the spindle during the actual machining process to predict and compensate for the spindle axial elongation, thereby improving machining accuracy. This method has the following problems: (1) The pre-established mapping relationship between temperature and spindle axial error is often carried out under the no-load condition of the CNC grinding machine, ignoring the influence of grinding force during the machining process, which is significantly different from the actual machining process. It is difficult to guarantee the accuracy of axial error prediction during the actual machining process, which affects the machining quality; (2) Compensation is usually carried out based on the predicted axial error value after a batch of workpieces is machined, which may affect the machining quality due to error accumulation; (3) While suppressing spindle axial error, there is a lack of effective control over energy consumption, which reduces resource utilization efficiency and increases production costs. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption. This method can improve the accuracy of axial error prediction in actual machining processes and can improve resource utilization efficiency and reduce production costs while suppressing spindle axial errors.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for optimizing multi-process CNC grinding machine operations while balancing accuracy and energy consumption, the method comprising:

[0007] A first relational model is established for each process of the CNC grinding machine; the first relational model is used to characterize the correlation between process parameters and the axial error of the CNC grinding machine when it performs actual machining under the process parameters; the process parameters include grinding wheel linear speed, grinding depth and workpiece feed speed;

[0008] A second relational model is constructed for each process of the CNC grinding machine; the second relational model is used to characterize the relationship between the process parameters of the CNC grinding machine and the energy consumption of the grinding machine.

[0009] Based on the first and second relational models of each process, and taking the axial error and grinding energy consumption of each process as optimization objectives, a multi-objective evolutionary algorithm guided by reference vectors is used to solve the Pareto optimal solution set of each process; the Pareto optimal solution set of a process includes at least one set of optimal process parameters for that process.

[0010] Under the constraint of constant total grinding amount, the process parameters in the Pareto optimal solution set of each process are dynamically adjusted to obtain the global optimal solution set; the global optimal solution set includes at least one optimal process parameter set, and the optimal process parameter set includes the optimal process parameters of each process.

[0011] Optionally, establish the first relational model for each process of the CNC grinding machine, specifically including:

[0012] The experiment obtained the normal grinding force and tangential grinding force of the CNC grinding machine under different combinations of experimental data, and established normal grinding force mapping models and tangential grinding force mapping models; the different combinations of experimental data were combinations when the grinding wheel linear speed, grinding depth and workpiece feed speed were taken with different values;

[0013] Based on the processing procedure, the range of values ​​for each process parameter under the i-th process is defined, the factor level is determined using the Taguchi method, and an experimental orthogonal array is designed for the i-th process; i = 1, 2, ..., I, where I represents the number of processes in the CNC grinding process;

[0014] Based on the normal grinding force mapping model and the tangential grinding force mapping model, the normal grinding force and tangential grinding force corresponding to the experimental data combination in the experimental orthogonal array of the i-th process are determined; the experimental data combination consists of the experimental data of each process parameter.

[0015] The control parameters of the spindle loading experimental platform are set according to the normal grinding force and tangential grinding force corresponding to each experimental data combination in the experimental orthogonal table of the i-th process. The process parameter values ​​of the CNC grinding machine are set according to each experimental data combination in the i-th process. The loading experiment is carried out to obtain the axial error corresponding to each experimental data combination in the experimental orthogonal table of the i-th process.

[0016] Each experimental data combination and its corresponding axial error in the experimental orthogonal array of the i-th process are used as a mapping relationship sample to construct the initial sample set of the i-th process;

[0017] Based on the initial sample set of the i-th process, the first relational model of the i-th process of the CNC grinding machine is constructed using the frost-ice element generation algorithm.

[0018] Optionally, the frost-ice meta-generation algorithm includes a dynamically generated adversarial unit, a meta-learning unit, and a frost-ice optimization unit;

[0019] The dynamic generative adversarial unit is used to generate samples from the original samples to obtain generated samples; the original samples are the mapping relationship samples in the initial sample set.

[0020] The meta-learning unit is used to fuse the original samples and generated samples to obtain a fused sample set, and divide the samples in the fused sample set into multiple sub-tasks to construct an axial error prediction model for each sub-task.

[0021] The frost optimization unit is used to perform optimization iterations using soft frost search, hard frost puncture, and greedy selection mechanisms to construct the first relational model.

[0022] Optionally, the dynamically generated adversarial unit includes a dynamic generator and a discriminator, wherein the dynamic generator is equipped with a dynamic adjustment mechanism.

[0023] Optionally, the trigger condition for the dynamic adjustment mechanism is: the FID distance reaches a preset threshold, and the formula for calculating the FID distance is:

[0024] FID = ||μ r -μ g || 2 +Tr(Σ r +Σ g -2(Σ r Σ g ) 1 / 2 );

[0025] Where FID represents the FID distance, μ r and Σ r Let μ be the mean and covariance matrix of the original sample, respectively. g and Σ g are the mean and covariance matrices of the generated samples, respectively, and Tr is the trace of the matrix.

[0026] Optionally, the dynamic adjustment mechanism is used to adjust the noise level and / or the regularization intensity.

[0027] Optionally, a second relational model is constructed for each operation of the CNC grinding machine, specifically including:

[0028] Identify the energy-consuming modules related to process parameters during CNC grinding as target energy-consuming modules. The target energy-consuming modules include spindle system rotation, feed system motion, and material grinding.

[0029] Determine the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process; i = 1, 2, ..., I, where I represents the number of processes in the CNC grinding machine machining process;

[0030] Based on the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process, construct a relationship model between the process parameters and energy consumption power of each target energy-consuming module in the i-th process.

[0031] By integrating the relationship model between the process parameters and energy consumption power of each target energy-consuming module in the i-th process, a second relationship model for the i-th process is obtained.

[0032] Optionally, determine the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process, specifically including:

[0033] Under any process parameters, determine the stable operating power of the grinding machine, the power when the spindle system rotates alone, the power when the feed system moves alone, the no-load power before the grinding wheel contacts the workpiece, and the grinding power during grinding.

[0034] The difference between the power of the spindle system rotating independently and the power of the grinding machine during stable operation is calculated and used as the energy consumption power of the feed system motion;

[0035] The difference between the power of the feed system when it moves independently and the power of the grinding machine during stable operation is calculated and used as the energy consumption power of the feed system.

[0036] The difference between the grinding power during grinding and the no-load power before the grinding wheel contacts the workpiece is calculated as the energy consumption power for material grinding.

[0037] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0038] A method for optimizing the multi-process operation of a CNC grinding machine to balance accuracy and energy consumption includes: establishing a first relational model for each process of the CNC grinding machine; the first relational model characterizes the correlation between process parameters and the axial error of the CNC grinding machine during actual machining under the process parameters; the process parameters include grinding wheel linear speed, grinding depth, and workpiece feed speed; constructing a second relational model for each process of the CNC grinding machine; the second relational model characterizes the correlation between the process parameters of the CNC grinding machine and the grinding machine energy consumption; based on the first and second relational models for each process, using the axial error and grinding machine energy consumption of each process as optimization objectives, employing a reference vector-guided multi-objective evolutionary algorithm to solve for the Pareto optimal solution set of each process; the Pareto optimal solution set of each process includes at least one set of optimal process parameters for that process; dynamically adjusting the process parameters in the Pareto optimal solution set of each process under the constraint of constant total grinding amount to obtain a global optimal solution set; the global optimal solution set includes at least one set of optimal process parameters, and the set of optimal process parameters includes the optimal process parameters for each process. This invention establishes a first relational model characterizing the relationship between process parameters and axial error. This first relational model can reflect the relationship between process parameters and the axial error of the CNC grinding machine during actual machining under the process parameters, and can achieve high-precision prediction of axial error in the actual machining process. This invention also establishes a second relational model to reflect the relationship between process parameters and grinding machine energy consumption. Furthermore, with axial error and grinding machine energy consumption as objectives, multi-objective optimization is performed to solve the problem, which can improve resource utilization efficiency and reduce production costs while suppressing spindle axial error. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A flowchart illustrating a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption, provided in an embodiment of the present invention;

[0041] Figure 2 A schematic diagram illustrating a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption, provided in an embodiment of the present invention.

[0042] Figure 3 This is a schematic diagram of the force sensor arrangement on a CNC grinding machine provided in an embodiment of the present invention;

[0043] Figure 4This is a force diagram of CNC grinding wheel grinding provided in an embodiment of the present invention;

[0044] Figure 5 This is a schematic diagram of the workpiece before and after tooth profile formation provided in an embodiment of the present invention;

[0045] Figure 6 This invention provides a normal / tangential grinding force mapping model before / after workpiece tooth profile formation in an embodiment of the invention.

[0046] Figure 7 A simplified diagram of the spindle loading experimental platform provided in an embodiment of the present invention;

[0047] Figure 8 This is a schematic diagram of the connection between the spindle and the bearing provided in an embodiment of the present invention;

[0048] Figure 9 This is a schematic diagram of the pressure sensor and servo electric cylinder push rod of the spindle loading experimental platform provided in an embodiment of the present invention.

[0049] Explanation of reference numerals in the attached figures:

[0050] 1. Triaxial force sensor; 2. Test workpiece; 3. Worktable; 4. Grinding wheel; 5. Normal servo cylinder; 6. Tangential servo cylinder; 7. Bearing unit; 8. Spindle body; 9. Inner ring of bearing unit; 10. Outer ring of bearing unit; 11. Spindle front end; 12. Normal pressure sensor; 13. Tangential pressure sensor. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] The purpose of this invention is to provide a multi-process optimization method for CNC grinding machines that balances accuracy and energy consumption. This method can improve the accuracy of axial error prediction in actual machining processes and can improve resource utilization efficiency and reduce production costs while suppressing spindle axial errors.

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] In one embodiment, the present invention provides a method for optimizing multi-process CNC grinding machines that balances accuracy and energy consumption, such as... Figure 1 and Figure 2 As shown, the method includes:

[0055] Step 101: Establish a first relational model for each process of the CNC grinding machine; the first relational model is used to characterize the correlation between process parameters and the axial error of the CNC grinding machine when it performs actual machining under the process parameters; the process parameters include grinding wheel linear speed, grinding depth and workpiece feed speed.

[0056] For example, step 101 in this embodiment specifically includes:

[0057] S1: Fix the triaxial force sensor 1 between the test workpiece 2 and the worktable 3 of the CNC grinding machine. Sequentially change the grinding wheel linear speed, grinding depth, and workpiece feed speed, i.e., set different process parameters to obtain multiple sets of test data. Record the normal / tangential grinding forces under different combinations of process parameters, and establish normal / tangential grinding force mapping models for the workpiece before and after tooth formation. In this step, the normal grinding force mapping model includes normal grinding force mapping models for the stage before and after workpiece tooth formation, and the tangential grinding force mapping model includes tangential grinding force mapping models for the stage before and after workpiece tooth formation.

[0058] S1 specifically refers to:

[0059] S11: Fix the SBT 301 triaxial force sensor 1 between the test workpiece 2 and the worktable 3 of the CNC grinding machine (e.g., Figure 3 As shown in the figure), the sampling interval was set to 1 second, and the real-time data was imported into the computer for display through the MCC-1608G data acquisition card. Considering that the actual processing of the workpiece usually includes a rough grinding stage to remove most of the blank material and a fine grinding stage to ensure machining accuracy, a forming grinding experiment with two stages, before and after tooth profile formation, was designed to correspond to the rough and fine grinding processes, respectively. The grinding wheel linear speed, grinding depth, and workpiece feed speed were changed sequentially (as shown in Table 1), and the normal grinding force, tangential grinding force, and axial grinding force under different combinations of process parameters were recorded. Among them, when the grinding wheel 4 is in contact with the workpiece, the axial grinding force is small and can be ignored (e.g., when the grinding wheel 4 is in contact with the workpiece, the axial grinding force is small and can be ignored). Figure 4 (As shown).

[0060] Table 1. Experimental Parameter Settings for Form Grinding

[0061] parameter Parameter value Grinding wheel linear velocity (m / s) 26；28；30 Grinding depth (mm) 0.05；0.1；0.2；0.3；0.4；0.5 Workpiece feed rate (mm / min) 160；200；240；280

[0062] S12: The empirical formulas for solving the normal grinding force and the tangential grinding force are as follows:

[0063]

[0064] Where F represents the normal grinding force F n Or tangential grinding force F tv1 represents the grinding wheel linear velocity, v2 represents the workpiece feed rate, h represents the grinding depth, and C and x1 to x3 represent weighting constants. After logarithmic transformation and simplification, we have:

[0065] F = C + x1v1 + x2h + x3lnv2

[0066] Before tooth profile formation, as the grinding depth increases, more abrasive grains participate in the grinding process. However, once the tooth profile is formed, the number of abrasive grains participating in the grinding process no longer changes (e.g., ...). Figure 5 (As shown). Therefore, based on experimental data and formulas, tangential grinding force mapping models are established for the stages before and after workpiece tooth profile formation, as well as for the stages before and after workpiece tooth profile formation. Figure 6 As shown.

[0067] S2: Build a spindle loading experimental platform, arrange two servo electric cylinders perpendicular to the normal / tangential direction of the front end 11 of the spindle, and arrange displacement sensors directly opposite the front end 11 of the spindle.

[0068] S2 specifically refers to:

[0069] A spindle loading experimental platform was constructed, and two servo cylinders (including a normal servo cylinder 5 and a tangential servo cylinder 6) were vertically arranged at the front end 11 of the spindle to simulate normal / tangential loads (such as...). Figure 7 (As shown). The servo electric cylinder applies load to the spindle body 8 via the bearing unit 7. The inner ring 9 of the bearing unit is tightly fitted to the front end 11 of the spindle and rotates with the spindle body 8. The outer ring 10 of the bearing unit is fixed and does not rotate, and is used to bear the normal / tangential load of the servo electric cylinder (e.g., ...). Figure 8 (As shown); the pressure sensors (including normal pressure sensor 12 and tangential pressure sensor 13) are fixed to the front end of the servo cylinder push rod, receive pressure signals and feed them back to the servo cylinder through the control box, thereby realizing the control of the pressure magnitude (e.g. Figure 9 (As shown); The displacement sensor is arranged directly opposite the front end 11 of the spindle, and the change in the distance between the front end 11 of the spindle and the displacement sensor reflects the axial error of the spindle.

[0070] S3: Based on the actual processing of the workpiece, define the range of values ​​for each process parameter under each process step, use the Taguchi method to determine the factor level and design an experimental orthogonal array for each process step, analyze and obtain characteristic parameters such as rotation speed and grinding force for each group of experiments, use a servo electric cylinder to simulate the corresponding load conditions, and determine the axial error of the spindle body 8 running for N hours (e.g., 2 hours) under each combination of process parameters.

[0071] S3 specifically refers to:

[0072] S31: Taking a circular gear rack workpiece as an example, its grinding process mainly includes rough grinding and fine grinding stages. The main parameters for the rough grinding stage are v1 = 26 m / s, h = 8 mm, and v2 = 150 mm / min; the main parameters for the fine grinding stage are v1 = 28 m / s, h = 0.1 mm, and v2 = 800 mm / min. Based on the existing processing parameters, the value ranges of each process parameter for rough grinding and fine grinding are defined respectively, and the experimental orthogonal arrays for each process are designed using the Taguchi method, as shown in Tables 2-4.

[0073] Table 2. Adjustment range of Taguchi parameters and setting of control factor levels during the coarse grinding stage.

[0074]

[0075] Table 3. Adjustment range of Taguchi parameters and setting of control factor levels during the fine grinding stage.

[0076]

[0077] Table 4 Taguchi Orthogonal Table L 25

[0078]

[0079]

[0080] S32: Using the grinding load force model established in S1, the normal / tangential grinding forces of each group of experiments shown in the table are analyzed. Two servo electric cylinders are used to simulate the corresponding load conditions to determine the axial error of the spindle body 8 after running for 2 hours under each combination of process parameters.

[0081] S4: Based on a small number of process parameter-axial error mapping relationship samples in S3, the sample capacity is expanded by using the dynamic generative adversarial unit in the frost-ice meta-generation algorithm. After dividing the samples into multiple sub-tasks, an axial error model for each sub-task is established through the meta-learning unit. The performance of each sub-task in other tasks is evaluated, and the frost-ice optimization unit is introduced to iteratively optimize and establish a more accurate process parameter-axial error correlation model as the first relationship model mentioned above.

[0082] S4 specifically refers to:

[0083] S41: The Frost Ice Meta-Generation Algorithm mainly includes a Dynamic Generative Adversarial Unit (GRAP), a Meta-Learning Unit (MEL), and a Frost Ice Optimization Unit. The Dynamic GRAP consists of a dynamic generator (D) and a discriminator (G). To prevent the generated samples from having too little difference from the original samples, the quality indicators of the generated samples, such as the FID distance, are periodically monitored during training. When these indicators reach a predetermined threshold, the generator's dynamic adjustment mechanism is triggered. The formula for calculating the FID distance is:

[0084] FID = ||μ r -μg || 2 +Tr(Σ r +Σ g -2(Σ r Σ g ) 1 / 2 )

[0085] Here, it is assumed that the multidimensional Gaussian distributions of the original sample and the generated sample are N(μ r ,Σ r ), N(μg,Σ g ), μ r and Σ r Let μ represent the mean and covariance matrix of the original sample, respectively. g and Σ g Let represent the mean and covariance matrices of the generated samples, respectively, and Tr be the trace of the matrix.

[0086] The dynamic adjustment mechanism mainly includes two methods: noise level adjustment and regularization intensity adjustment, to ensure a balance between the diversity and quality of generated samples. The dynamic generator D and the discriminator G engage in a game-like interaction, ultimately reaching a Nash equilibrium where, given a fixed opponent's decision, the generator will always make the optimal judgment. The final objective function of the dynamic generative adversarial unit is:

[0087]

[0088] Where x represents the original sample, This indicates that the true sample distribution p data In the expression (x), the random variable x is subjected to expectation operation. p(z), G(·), and D(·) represent the prior part of the random noise vector z in the latent space, the generating function, and the discriminator function with an output span of [0,1], respectively. When D(x) is 0, the discriminator considers this sample to be a generated sample; if it does not, this value is 1.

[0089] S42: The generated samples are fused with the original samples. The samples are then divided into multiple sub-tasks using a meta-learning unit, and an axial error prediction model is built for each sub-task. The performance of each sub-task on other tasks is evaluated. Specifically, the samples in the fused sample set are divided into multiple subsets, each corresponding to a sub-task. An axial error prediction model is built based on the samples of each subset. The axial error prediction model for each sub-task is iteratively trained and selected. During training, samples from other subsets are used as test samples to test the performance of the axial error prediction model during training, and the test results are used as the objective function value. See S43 for specific iterative training steps.

[0090] S43: Optimize and iterate according to the soft frost search, hard frost puncture and greedy selection mechanism in the frost optimization unit.

[0091] Inspired by the characteristics of soft frost growth—its strong randomness, ability to freely cover most of an object's surface, and slow growth in the same direction—the soft frost search strategy enables the algorithm to quickly cover the entire search space in early iterations, making it less prone to getting trapped in local optima. The expression for soft frost search is:

[0092]

[0093] in, This refers to updating the new position of the optimal solution, where i and j represent the j-th parameter of the i-th solution. R best,j It is the j-th parameter of the optimal solution among all solutions. Parameter r1 is a random number in the range (-1, 1), used to adjust the parameter update control of the optimization algorithm; cosθ varies with the number of iterations; β is an environmental factor that varies with the number of iterations to simulate the influence of the external environment, used to ensure the convergence of the optimization algorithm; h is the adhesion, which fluctuates randomly within (0, 1), used to adjust the relative distance between solutions in the optimization algorithm. Furthermore, the expressions for θ, β, and E are as follows:

[0094]

[0095]

[0096]

[0097] Where t is the current iteration number, β is a step function, and [] indicates rounding; ω has a default value of 5 and is used to control the number of steps in the step function. E is the adhesion coefficient, which increases with the number of iterations.

[0098] Hard rime ice formed in strong winds often exhibits the same growth direction, making it prone to cross-piercing, and the piercing probability increases with better growth conditions. The hard rime ice piercing mechanism, through particle exchange, effectively improves the algorithm's convergence and ability to escape local optima. The expression for hard rime ice piercing is:

[0099]

[0100] Among them, F normr (S i ) represents the normalized value of the current fitness, and represents the chance of the i-th solution being selected. r3 is a random number in the range (-1, 1) used to adjust the update probability of solutions in the optimization algorithm.

[0101] Subsequently, based on the greedy selection mechanism, the fitness values ​​before and after the update are compared, and the error model parameters are iteratively updated to establish a more accurate process parameter-axial error correlation model.

[0102] Step 102: Construct a second relational model for each operation of the CNC grinding machine; the second relational model is used to characterize the relationship between the process parameters of the CNC grinding machine and the energy consumption of the grinding machine, including the following steps:

[0103] S5: Determine the energy consumption sources related to the CNC grinding machine machining process and process parameters, such as the spindle system operation, feed system movement, and material grinding. Change each process parameter in sequence and use a power analyzer to determine the power consumption of the corresponding module. Establish the functional relationship between each energy consumption module of the CNC grinding machine and the process parameters. After merging, construct the process parameter-grinding machine energy consumption correlation model as the second relationship model mentioned above.

[0104] S5 specifically refers to:

[0105] The CNC grinding process mainly includes energy-consuming modules such as: stable operation of the grinding machine, spindle rotation, feed system movement, material grinding, and auxiliary systems. The energy consumption during stable operation refers to the energy consumed when the CNC grinding machine is running stably without any other operations. The energy consumption of the auxiliary systems refers to the energy consumed by lighting, cooling, and CNC control during the grinding process. The energy consumption during stable operation and the energy consumption of the auxiliary systems are independent of the settings of the CNC grinding machine's process parameters.

[0106] The energy consumption of the spindle system rotation and feed system motion is related to the grinding wheel linear velocity and workpiece feed speed, respectively. To establish the relationship between each energy-consuming module and process parameters, the CNC grinding machine first performs actions such as table 3 feed and spindle body 8 rotation independently. A power analyzer is used to characterize the energy consumption caused by the spindle system rotation and feed system motion by measuring the difference between the measured grinding machine power and the stable operating power. Power models for the energy consumption modules of the spindle system rotation and feed system motion are established separately.

[0107] The energy consumption of material grinding is comprehensively affected by multiple process parameters such as spindle speed, feed rate, and grinding depth. The energy consumption of material grinding is characterized by the difference between the grinding power of the CNC grinding machine during grinding and the no-load power before the grinding wheel 4 contacts the workpiece. The empirical power model for the material grinding energy consumption module is as follows:

[0108]

[0109] Where H and h represent the axial and radial grinding depths, respectively; v2 represents the workpiece feed rate; k c η represents the grinding force per unit area; η is the total efficiency of the grinding machine, which is usually taken as 1 without considering the no-load condition.

[0110] By integrating the power models of energy-consuming modules such as spindle rotation, feed system motion, and material grinding, a process parameter-grinding machine energy consumption correlation model is constructed, namely the second relationship model.

[0111] Step 103: Based on the first and second relational models of each process, and taking the axial error and grinding energy consumption of each process as optimization objectives, a multi-objective evolutionary algorithm guided by reference vectors is used to solve the Pareto optimal solution set of each process; the Pareto optimal solution set of a process includes at least one set of optimal process parameters for that process.

[0112] Step 104: Under the constraint of constant total grinding amount, dynamically adjust the process parameters in the Pareto optimal solution set of each process to obtain the global optimal solution set; the global optimal solution set includes at least one set of optimal process parameters, and the set of optimal process parameters includes the optimal process parameters for each process. Steps 103 and 104 in this embodiment include the following steps:

[0113] S6: Using a cross-process joint optimization algorithm, the axial error of a single process and the energy consumption of the grinding machine, as well as the Pareto optimal solution set of each process, are successively used as optimization objectives to obtain the global optimization solution of the CNC grinding machine process parameters, thereby achieving multi-process optimization of CNC grinding machines that balances accuracy and energy consumption.

[0114] S6 specifically refers to:

[0115] S61: Taking the workpiece described in S31 as an example, there are two processes: rough grinding and fine grinding, with a total grinding amount of 8.1 mm. When improving process parameters, time cost must be considered, i.e., ensuring the number of processes remains constant. The cross-process joint optimization algorithm first breaks down the manufacturing process into multiple process parameter optimization problems with independent optimization objectives (axial error, grinding machine energy consumption, etc.), and then uses a reference vector-guided multi-objective evolutionary algorithm to solve for the uniformly distributed Pareto optimal solution set for each process. The axial error and grinding machine energy consumption are determined based on the process parameter-axial error correlation model established in S4 and the process parameter-grinding machine energy consumption correlation model established in S5, respectively.

[0116] S62: At the global level, using the Pareto optimal solution set of each process as the optimization objective, a multi-objective evolutionary algorithm guided by reference vectors is applied again to dynamically adjust the coordination parameters between processes under the constraint of constant total grinding amount, thereby achieving balanced optimization among processes. Specifically, the process parameters in the Pareto optimal solution set of each process are combined to obtain multiple full-process process parameter combination results. Based on the total grinding amount and target grinding amount corresponding to each full-process process parameter combination result, the full-process process parameter combination result where the difference between the total grinding amount and the target grinding amount is less than a certain threshold is determined as the global optimal solution set.

[0117] As can be seen from this embodiment, the technical solution provided by the present invention has the following beneficial effects:

[0118] 1) This invention builds a spindle loading experimental platform to simulate the normal / tangential grinding force under various combinations of process parameters of CNC grinding machine, records the spindle axial error and establishes a process parameter-axial error correlation model, thereby realizing accurate prediction of axial error matching different combinations of process parameters.

[0119] 2) This invention slows down the accumulation process of spindle axial error from the perspective of process design, reduces the requirements for axial error prediction accuracy, and reduces the difficulty of error compensation.

[0120] 3) By improving the grinding process parameters, this invention balances the relationship between axial error and grinding machine energy consumption between each process and within a single process, thus realizing the precision and low-power manufacturing of CNC grinding machines.

[0121] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0122] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A precision and energy consumption balanced multi-process optimization method for a CNC grinding machine, characterized in that, The method includes: A first relational model is established for each process of the CNC grinding machine; the first relational model is used to characterize the correlation between process parameters and the axial error of the CNC grinding machine when it performs actual machining under the process parameters; the process parameters include grinding wheel linear speed, grinding depth and workpiece feed speed; A second relational model is constructed for each process of the CNC grinding machine; the second relational model is used to characterize the relationship between the process parameters of the CNC grinding machine and the energy consumption of the grinding machine. Based on the first and second relational models of each process, and taking the axial error and grinding energy consumption of each process as optimization objectives, a multi-objective evolutionary algorithm guided by reference vectors is used to solve the Pareto optimal solution set of each process; the Pareto optimal solution set of a process includes at least one set of optimal process parameters for that process. Under the constraint of constant total grinding amount, the process parameters in the Pareto optimal solution set of each process are dynamically adjusted to obtain the global optimal solution set; the global optimal solution set includes at least one optimal process parameter set, and the optimal process parameter set includes the optimal process parameters of each process. Establish the first relational model for each process of the CNC grinding machine, specifically including: The experiment obtained the normal grinding force and tangential grinding force of the CNC grinding machine under different combinations of experimental data, and established normal grinding force mapping models and tangential grinding force mapping models; the different combinations of experimental data were combinations when the grinding wheel linear speed, grinding depth and workpiece feed speed were taken with different values; Based on the processing procedure, the range of values ​​for each process parameter under the i-th process is defined, the factor level is determined using the Taguchi method, and an experimental orthogonal array is designed for the i-th process; i=1,2,...,I,I represents the number of processes in the CNC grinding process; Based on the normal grinding force mapping model and the tangential grinding force mapping model, the normal grinding force and tangential grinding force corresponding to the experimental data combination in the experimental orthogonal array of the i-th process are determined; the experimental data combination consists of the experimental data of each process parameter. The control parameters of the spindle loading experimental platform are set according to the normal grinding force and tangential grinding force corresponding to each experimental data combination in the experimental orthogonal table of the i-th process. The process parameter values ​​of the CNC grinding machine are set according to each experimental data combination in the i-th process. The loading experiment is carried out to obtain the axial error corresponding to each experimental data combination in the experimental orthogonal table of the i-th process. Each experimental data combination and its corresponding axial error in the experimental orthogonal array of the i-th process are used as a mapping relationship sample to construct the initial sample set of the i-th process; The first relational model of the i-th process of the CNC grinding machine is constructed using the frost-ice element generation algorithm based on the initial sample set of the i-th process. Constructing a second relational model for each operation of a CNC grinding machine, specifically including: Identify the energy-consuming modules related to process parameters during CNC grinding as target energy-consuming modules. The target energy-consuming modules include spindle system rotation, feed system motion, and material grinding. Determine the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process; i = 1, 2, ..., I, where I represents the number of processes in the CNC grinding machine machining process; Based on the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process, construct a relationship model between the process parameters and energy consumption power of each target energy-consuming module in the i-th process. By integrating the relationship model between the process parameters and energy consumption power of each target energy-consuming module in the i-th process, a second relationship model for the i-th process is obtained.

2. The method for optimizing the multi-process of CNC grinding machines to balance accuracy and energy consumption according to claim 1, characterized in that, The frost and ice meta-generation algorithm includes a dynamic generation adversarial unit, a meta-learning unit, and a frost and ice optimization unit. The dynamic generative adversarial unit is used to generate samples from the original samples to obtain generated samples; the original samples are the mapping relationship samples in the initial sample set. The meta-learning unit is used to fuse the original samples and generated samples to obtain a fused sample set, and divide the samples in the fused sample set into multiple sub-tasks to construct an axial error prediction model for each sub-task. The frost optimization unit is used to perform optimization iterations using soft frost search, hard frost puncture, and greedy selection mechanisms to construct the first relational model.

3. The method for optimizing the multi-process of CNC grinding machines to balance accuracy and energy consumption according to claim 2, characterized in that, The dynamic generation adversarial unit includes a dynamic generator and a discriminator, wherein the dynamic generator is equipped with a dynamic adjustment mechanism.

4. The method for optimizing the multi-process of CNC grinding machines to balance accuracy and energy consumption according to claim 3, characterized in that, The trigger condition for the dynamic adjustment mechanism is: the FID distance reaches a preset threshold, and the formula for calculating the FID distance is: ； in, Indicates FID distance, and These are the mean and covariance matrices of the original sample, respectively. and These are the mean and covariance matrices of the generated samples, respectively. Let be the trace of the matrix.

5. The method for optimizing the multi-process of CNC grinding machines to balance accuracy and energy consumption according to claim 3, characterized in that, The dynamic adjustment mechanism is used to adjust the noise level and / or the regularization intensity.

6. The method for optimizing the multi-process of CNC grinding machines to balance accuracy and energy consumption according to claim 1, characterized in that, Determine the energy consumption power of each target energy-consuming module corresponding to different process parameters of the i-th process, specifically including: Under any process parameters, determine the stable operating power of the grinding machine, the power when the spindle system rotates alone, the power when the feed system moves alone, the no-load power before the grinding wheel contacts the workpiece, and the grinding power during grinding. The difference between the power of the spindle system rotating independently and the power of the grinding machine during stable operation is calculated and used as the energy consumption power of the feed system motion; The difference between the power of the feed system when it moves independently and the power of the grinding machine during stable operation is calculated and used as the energy consumption power of the feed system. The difference between the grinding power during grinding and the no-load power before the grinding wheel contacts the workpiece is calculated as the energy consumption power for material grinding.

Images