A multi-functional cutter numerical control machining parameter setting method based on deep learning

Abstract

The application relates to the technical field of cutter machining, in particular to a multifunctional cutter numerical control machining parameter setting method based on deep learning. The method comprises the following steps: based on a deep learning algorithm, a machining threshold value model and a numerical control machining parameter setting model are established; machining threshold values corresponding to cutter basic information are obtained from a training data set to train the machining threshold value model; meanwhile, a part model, cutter machining parameters and a part overhang ratio are obtained to train the numerical control machining parameter setting model; cutter basic information and the part model input by a user are respectively input into the trained model to obtain the machining threshold values, the cutter machining parameters and the part overhang ratio; after a part is machined according to the parameters, a machining part number-blade edge arc radius curve is constructed; the cutter wear state is judged through the curve slope; and the machining threshold values and the part overhang ratio are corrected according to the cutter wear state. The application provides a method for quickly generating multifunctional cutter numerical control machining parameters, and solves the problem of machining precision decline caused by cutter wear.

Description

Technical Field

[0001] This invention relates to the field of cutting tool processing technology, and in particular to a method for setting parameters for multifunctional CNC cutting tools based on deep learning. Background Technology

[0002] Currently, the setting of tool machining parameters for multi-station grinding machines mainly relies on manual experience and experimental methods, which are not only time-consuming and labor-intensive but also inefficient. Therefore, many methods for optimizing machining parameters have been proposed in recent years. However, in actual machining processes, besides the setting of machining parameters, tool wear is also a significant influencing factor. With use, tools gradually wear down, leading to a gradual decrease in the accuracy of machined parts. This wear not only affects machining quality but may also shorten tool life, increasing production costs. Therefore, there is an urgent need for a method that can dynamically optimize machining parameters based on automatic tool CNC machining parameter setting, ensuring rapid parameter setting and guaranteeing the accuracy and quality of machined parts.

[0003] Chinese patent CN104268339B proposes a method and system for selecting cutting parameters and automatically setting parameters based on machining features. The method includes: classifying machining features by categorizing typical workpiece machining features; simulating the cutting process for each feature to obtain the cutting temperature corresponding to the minimum tool wear rate; fitting the relationship between cutting temperature, cutting parameters, and machining feature parameters; and obtaining the cutting speed calculation formula at the cutting temperature that minimizes tool wear. The calculation formulas for cutting parameters corresponding to different machining features are stored in a cutting parameter recommendation system as underlying data. The cutting parameter recommendation system is integrated into CNC programming software, enabling parameter setting within the software. While this patent solves the problem of parameter setting during CNC machining, it fails to fully consider the dynamic changes in tool condition during actual machining. Tools gradually wear during use, but this patent only uses the minimum tool wear rate to represent the actual wear state, ignoring the complexity of tool wear during machining. This may lead to increasingly larger errors in parts machined using this method over time, failing to meet the requirements of high-precision machining. Therefore, there is an urgent need for a method that can output tool machining parameters according to user machining requirements while effectively avoiding the problem of decreased accuracy due to tool wear. This would allow for further optimization of tool machining accuracy based on the tool machining parameter settings, ensuring the stability and reliability of machining quality. Summary of the Invention

[0004] To address this, the present invention provides a method for setting CNC machining parameters for multi-functional cutting tools based on deep learning, which overcomes the problem that the setting of cutting tool machining parameters for multi-station grinding machines in the prior art is highly dependent on experience and does not take into account the real-time changes in tool wear state, which leads to a decrease in the accuracy of machined parts.

[0005] To achieve the above objectives, this invention provides a method for setting CNC machining parameters for multi-functional cutting tools based on deep learning, comprising:

[0006] A machining threshold model and a CNC machining parameter setting model are established based on deep learning algorithms;

[0007] Obtain a training dataset, obtain the machining threshold corresponding to the basic information of the cutting tool from the training dataset, train the machining threshold model based on several machining thresholds, and obtain the cutting tool machining parameters and part overhang ratio corresponding to the part model and machining threshold from the training dataset, and train the CNC machining parameter setting model based on several cutting tool machining parameters and part overhang ratio.

[0008] The basic information of the cutting tool is obtained and input into the trained machining threshold model to obtain several machining thresholds. The several machining thresholds and the part model are then input into the CNC machining parameter setting model to obtain model output parameters, including several cutting tool machining parameters and part overhang ratio.

[0009] Based on the model output parameters, the parts are machined, the cutting edge radius of the tool is obtained, and a curve of the number of machined parts - cutting edge radius is constructed. The wear state of the tool is determined based on the slope at the corresponding node, and it is determined whether to correct the machining threshold for the tool.

[0010] In response to the processing threshold being adjusted to a critical value, it is determined whether to correct the part overhang ratio based on the wear state of the tool after completing the processing of a single part.

[0011] The processing thresholds include the maximum single radial movement speed and the maximum axial movement speed.

[0012] The overhang ratio of the part is the ratio of the length of the part extending out of the clamping tool to the total length of the part.

[0013] The radius of the cutting edge arc is the radius of the arc formed at the cutting edge of the tool.

[0014] Furthermore, the process of determining the tool wear state based on the slope at the node of the curve representing the number of machined parts versus the radius of the cutting edge includes:

[0015] Obtain the slope of a single part at a node in the curve of number of processed parts - radius of the cutting edge arc;

[0016] When the slope is greater than the first preset slope, the tool wear state is determined to be the first wear state, and the maximum single radial movement speed of the tool is adjusted based on the difference between the slope and the first preset slope.

[0017] Wherein, the maximum single radial movement speed of the tool is the maximum radial movement speed of the tool in one feed action.

[0018] Furthermore, the process of adjusting the maximum single radial traverse speed of the tool includes:

[0019] The slope difference is compared with the preset slope difference, and the maximum single radial movement speed of the tool is reduced based on the comparison result. The reduction in the maximum single radial movement speed of the tool is proportional to the slope difference.

[0020] The slope difference refers to the difference between the slope and the first preset slope.

[0021] Furthermore, after adjusting the maximum single radial traverse speed of the tool, the maximum axial traverse speed is increased based on the average depth of cut;

[0022] The increase in the maximum axial movement speed is inversely proportional to the average depth of cut.

[0023] Furthermore, after adjusting the maximum single radial movement speed of the tool, the maximum single radial movement speed of the tool is reduced in combination with the cumulative usage time of the tool, and the maximum single radial movement speed of the tool is inversely proportional to the cumulative usage time.

[0024] Furthermore, after adjusting the maximum single radial movement speed of the tool, the radius of the cutting edge arc after the tool completes machining the next part is re-determined and the number of machined parts - radius of the cutting edge arc is updated. Based on the updated curve, the slope at the current node is determined and the tool wear status is determined.

[0025] When the slope is in the first wear state, the maximum single radial movement speed of the tool is redetermined based on the difference between the slope and the first preset slope.

[0026] Furthermore, when the slope is in the first loss state and the maximum single radial traverse speed for the tool needs to be re-determined, if the corrected maximum single radial traverse speed is less than the set critical radial traverse speed, the real-time spindle speed is reduced based on the current tool temperature.

[0027] The real-time spindle speed is inversely proportional to the current tool temperature.

[0028] Furthermore, the process of determining whether the tool wear state belongs to the second wear state based on the slope at the node of the curve of the number of processed parts - the radius of the cutting edge includes:

[0029] Obtain the slope of a single part at a node in the curve of number of processed parts - radius of the cutting edge arc;

[0030] When the slope is greater than the set second preset slope, the tool wear state is determined to be the second wear state, and the tool state is adjusted based on the historical slope variance of the quantity.

[0031] If the historical slope variance of the quantity is greater than the preset slope variance, the overhang ratio of the part is adjusted based on the difference between the historical slope variance of the quantity and the preset slope variance.

[0032] If the historical slope variance of the stated quantity is less than or equal to the preset slope variance, then the tool should be replaced.

[0033] Furthermore, the process of adjusting the overhang ratio of the part includes:

[0034] The variance difference value is compared with the preset variance difference value, and the overhang ratio of the part is reduced based on the comparison result;

[0035] The reduction in the overhang ratio of the part is proportional to the variance difference.

[0036] The variance difference refers to the difference between the historical slope variance and the preset slope variance.

[0037] Furthermore, after the overhang ratio of the part is adjusted, the reduction in the overhang ratio of the part is corrected based on the average dimensional deviation of the historical parts that have been processed.

[0038] The reduction in the overhang ratio of the part is proportional to the average dimensional deviation of the historical parts.

[0039] Compared with existing technologies, the advantages of this invention are as follows: Firstly, by establishing a machining threshold model and a CNC machining parameter setting model based on deep learning algorithms, this invention can automatically learn the complex relationship between tool parameters and machining conditions from a large amount of training data. When the user inputs basic tool information and part model, the model can quickly output the corresponding machining threshold and tool machining parameters, eliminating the need for tedious manual parameter adjustments and greatly improving the efficiency of tool parameter setting. Secondly, this invention constructs a curve representing the number of machined parts versus the radius of the cutting edge. After machining a single part, the slope of this curve at that node determines the tool wear state. This real-time monitoring method detects changes in tool wear and promptly corrects the machining threshold when the tool wear reaches a certain level, adapting to the worn state and reducing the adverse effects of tool wear on machining accuracy. Furthermore, after correcting the machining threshold, continuous monitoring and determination of whether to adjust the overhang ratio of the part based on the tool wear state after machining a single part further optimizes the machining process, ensuring that the accuracy of the machined parts remains within a controllable range. In summary, this invention automatically sets tool parameters through a deep learning model, improving parameter setting efficiency and reducing the time and effort required for manual debugging. Simultaneously, by monitoring tool wear in real time and promptly correcting machining parameters, it effectively avoids the problem of decreased machining accuracy caused by tool wear, thereby improving the quality and stability of machined parts.

[0040] Furthermore, determining the tool wear state by the slope at the node of the curve of the number of machined parts - the radius of the cutting edge provides a quantitative and objective method for tool wear assessment. Adjusting the maximum single radial movement speed of the tool based on the difference between the slope and the preset slope ensures that the current adjustment can meet the actual wear condition of the tool, thereby improving the adaptability and stability of the machining process and effectively avoiding machining quality problems caused by tool wear.

[0041] Furthermore, this solution adjusts the maximum single radial movement speed of the tool by adjusting the slope difference, which helps to ensure that the amount of material removed from the machined part remains stable even when the tool wear is accelerated, thereby ensuring machining accuracy. On the other hand, by setting the adjustment range to be proportional to the slope difference, the machining parameters are always adapted to the current tool condition, thus optimizing the machining process.

[0042] Furthermore, to avoid the impact of the maximum single radial traverse speed correction on the maximum single radial traverse speed, this invention adjusts the maximum axial traverse speed by means of the average cutting depth, making the adjustment of tool wear more precise. This effectively avoids the problem of reduced side cutting amount that may be caused by the correction of the maximum single radial traverse speed, thereby effectively solving the problem of decreased machining accuracy caused by tool wear and improving machining accuracy.

[0043] Furthermore, by reducing the maximum single radial traverse speed in conjunction with the cumulative usage time of the tool, this adjustment method takes into account the factor of tool wear over time, thereby further improving the adaptability and stability of the machining process and effectively avoiding machining quality problems caused by tool wear.

[0044] Furthermore, by redetermining the cutting edge radius after the tool completes machining the next part and updating the machining part number-cutting edge radius curve, this dynamic update mechanism can reflect the tool wear in a timely manner and redetermine the tool wear state and machining parameters based on the updated curve. This further improves the adaptability and stability of the machining process, avoids dependence on initial data, makes adjustments more precise, and improves machining accuracy.

[0045] Furthermore, when the corrected maximum single radial traverse speed is greater than the critical radial traverse speed, reducing the real-time spindle speed based on the current tool temperature helps to balance machining accuracy and efficiency, avoiding potential expansion deformation and efficiency reduction issues during tool wear correction. This further improves the adaptability and stability of the machining process, ensuring the quality and efficiency of the machined parts.

[0046] Furthermore, the second wear state is determined by the slope at the node of the curve of number of processed parts - radius of the cutting edge. Based on the historical slope variance, the cause of the second wear state is accurately judged. This allows the present invention to make targeted corrections for different causes of tool wear, which not only improves the adaptability of the present invention, but also specifically corrects the impact of tool wear, improves the machining accuracy, and effectively avoids machining quality problems caused by tool wear.

[0047] Furthermore, the overhang ratio of the part is an important factor affecting machining accuracy. By adjusting the overhang ratio of the part based on the variance difference, the problem of tool vibration or breakage caused by mismatch of the overhang ratio of the part is effectively solved. This not only eliminates the problem of decreased machining accuracy caused by tool wear due to tool vibration, but also protects the tool and extends its service life.

[0048] Furthermore, after the overhang ratio of the part is adjusted, the overhang ratio of the part is corrected based on the average dimensional deviation of the historical parts that have been processed. This correction mechanism based on historical data helps to ensure that the processed parts are closer to the design dimensions, thereby improving the processing accuracy of the parts and enhancing the adaptability of the corrected overhang ratio of the parts to the processing process and processing parameters, thus avoiding the impact of tool wear and tool wear. Attached Figure Description

[0049] Figure 1This is a flowchart illustrating a method for setting CNC machining parameters for multi-functional cutting tools based on deep learning, as described in an embodiment of the present invention.

[0050] Figure 2 This is a logic diagram for determining whether the tool wear state belongs to the first wear state based on the slope of the curve node of the number of processed parts - the radius of the cutting edge arc in an embodiment of the present invention.

[0051] Figure 3 This is a logic diagram for determining whether the maximum single radial movement speed needs to be corrected based on the critical radial movement speed in an embodiment of the present invention.

[0052] Figure 4 This is a logic diagram for determining whether the tool wear state belongs to the second wear state based on the slope at the node of the curve of number of processed parts - radius of the cutting edge in an embodiment of the present invention. Detailed Implementation

[0053] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0054] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0055] Please see Figure 1 , Figure 2 , Figure 3 as well as Figure 4 The diagrams shown are, respectively, a flowchart of a deep learning-based CNC machining parameter setting method for multi-functional cutting tools in an embodiment of the present invention; a logic determination diagram for determining whether the tool wear state belongs to the first wear state based on the slope at the node of the curve of number of machined parts - radius of the cutting edge; a logic determination diagram for determining whether the maximum single radial traverse speed needs to be corrected based on the critical radial traverse speed; and a logic determination diagram for determining whether the tool wear state belongs to the second wear state based on the slope at the node of the curve of number of machined parts - radius of the cutting edge. The deep learning-based CNC machining parameter setting method in an embodiment of the present invention includes:

[0056] S1: Establish a machining threshold model and a CNC machining parameter setting model based on deep learning algorithms;

[0057] S2: Obtain a training dataset, obtain the machining threshold corresponding to the basic tool information from the training dataset, train the machining threshold model based on several machining thresholds, and obtain the tool machining parameters and part overhang ratio corresponding to the part model and machining threshold from the training dataset, and train the CNC machining parameter setting model based on several tool machining parameters and part overhang ratio.

[0058] S3: Input the basic tool information output by the user into the trained machining threshold model to obtain several machining thresholds;

[0059] S4: Input several processing thresholds and the part model output by the user into the trained CNC machining parameter setting model to obtain several tool processing parameters and part overhang ratio;

[0060] S5: Machining parts based on several tool machining parameters and part overhang ratio output by the CNC machining parameter setting model;

[0061] S6: After machining is completed, obtain the cutting edge radius of the tool and construct the curve of the number of machined parts - cutting edge radius;

[0062] S7: After completing the machining of a single part, determine the tool wear state based on the slope of the obtained curve at that node, and determine whether to correct the machining threshold for the tool based on the determined state.

[0063] S8: After completing the correction of the processing threshold, continue to monitor and determine whether to correct the overhang ratio of the part based on the wear status of the tool after completing the processing of a single part.

[0064] The processing thresholds include the maximum single radial movement speed and the maximum axial movement speed. The range of the maximum single radial movement speed is not limited in principle, and can be selected from any value between 100 mm / min and 800 mm / min. This will not be elaborated further. The range of the maximum axial movement speed is not limited in principle, and can be selected from any value between 200 mm / min and 3000 mm / min. This will not be elaborated further.

[0065] The overhang ratio of the part is the ratio of the length of the part extending from the clamping tool to the total length of the part. The range of values ​​for the overhang ratio is not limited in principle. Technicians can set the corresponding overhang ratio according to their needs, such as 1 / 10, 1 / 5, 1 / 3, 1 / 12, 2 / 7, which will not be elaborated further.

[0066] The radius of the cutting edge arc is the radius of the arc formed at the cutting edge of the tool. The range of the value of the cutting edge arc radius is not limited in principle. For example, it can be selected from any value between 0.015mm and 5mm. This will not be elaborated further.

[0067] Specifically, deep learning algorithms are machine learning methods based on artificial neural networks. They construct multi-layered neural network models by simulating the structure and function of neurons in the human brain. In principle, there are no restrictions on the types of deep learning algorithms. They can be any algorithm among fully connected neural networks, convolutional neural networks, recurrent neural networks, long short-term memory networks, generative adversarial networks, autoencoders, attention mechanisms, and reinforcement learning networks. This will not be elaborated further.

[0068] Specifically, the process of constructing a processing threshold model using a fully connected neural network algorithm in deep learning includes:

[0069] The number of neurons N in the input layer is set to be the same as the number of features of the tool's basic information.

[0070] The basic information of the cutting tool includes the tool type, tool diameter, total tool length, cutting edge length, number of cutting edges, tool helix angle, tool rake angle, tool clearance angle, cutting edge radius, tool material type, material hardness, material coating type, material coating thickness, initial wear degree, and tool service life.

[0071] The initial number of hidden layers is set to be equal to one-third of the number of features in the tool's basic information, rounded down.

[0072] Set each hidden layer to contain Y. i There are 1, 2, 3, ..., n neurons in the hidden layer, where the initial number of neurons Y1 is equal to the number of features of the basic information of the tool.

[0073] Set ReLU as the activation function;

[0074] Set the number of neurons M in the output layer to be equal to the number of features processed based on the threshold.

[0075] Specifically, the process of constructing a CNC machining parameter setting model using a fully connected neural network algorithm in deep learning includes:

[0076] Set the number of neurons in the input layer N' to be equal to the sum of the number of features for the part type and the processing threshold;

[0077] The initial number of hidden layers is set to equal the number of features based on the part model and processing threshold, rounded down to one-quarter.

[0078] Set each hidden layer to contain Y' i There are 1, 2, 3, ..., n neurons in the hidden layer, where the initial number of neurons Y1 is equal to the number of features of the basic information of the tool.

[0079] Set ReLU as the activation function;

[0080] The number of neurons in the output layer, M, is set to be equal to the sum of the tool machining parameters and the number of features related to the part overhang ratio;

[0081] The tool machining parameters refer to the cutting speed, feed rate, depth of cut, and spindle speed.

[0082] Specifically, the process of training the processing threshold model based on several processing thresholds includes:

[0083] Set the loss function and optimization algorithm. The loss function can be any function selected from root mean square error, mean absolute error, cross entropy, binary cross entropy, hinge loss, and KL divergence. The optimization algorithm can be any algorithm selected from stochastic gradient descent, momentum, Adagrad, RMSprop, and Adam. This will not be elaborated further.

[0084] Here, the loss function is set to root mean square error and the optimization algorithm is set to stochastic gradient descent.

[0085] Set the learning rate to [0.0005, 0.005];

[0086] Set the number of iterations to twenty times the number of features in the tool's basic information;

[0087] Set the batch size to equal the number of features in the tool's basic information;

[0088] Based on the machining thresholds corresponding to the basic tool information obtained from the training dataset, the machining threshold model is trained based on several machining thresholds.

[0089] Specifically, the process of training a CNC machining parameter setting model based on several tool machining parameters and part overhang ratio includes:

[0090] Set the loss function to mean absolute error and the optimization algorithm to stochastic gradient descent.

[0091] Set the learning rate to [0.0005, 0.005];

[0092] Set the number of iterations to twenty times the sum of the number of features for the part type and the processing threshold;

[0093] Set the batch size to equal the sum of the part type and the number of features based on the processing threshold;

[0094] Obtain the tool machining parameters and part overhang ratio corresponding to the part model and machining threshold from the training dataset, and train the CNC machining parameter setting model based on several tool machining parameters and part overhang ratios.

[0095] Input the basic tool information output by the user into the trained machining threshold model to obtain several machining thresholds. Input the several machining thresholds and the part model output by the user into the trained CNC machining parameter setting model to obtain several tool machining parameters and part overhang ratio.

[0096] The part model number output by the user refers to a specific code or name used to uniquely identify and distinguish different processing objects. In principle, there is no limitation on the part model number output by the user. Technicians can select different part models according to their needs to generate processing parameters for the corresponding part model.

[0097] Specifically, the process of determining the tool wear state based on the slope at the j-th node of the curve representing the number of machined parts versus the radius of the cutting edge includes:

[0098] Obtain the slope K of the j-th part at the node of the curve representing the number of processed parts minus the radius of the cutting edge. j , where j = 1, 2, ..., n;

[0099] The slope K j Compare with the first preset slope K1, where K1 = tan(k), k ∈ [10°, 40°], and k represents the angle between the tangent at the j-th node and the x-axis;

[0100] If the slope K j If the current blade wear is less than or equal to the first preset slope K1, it is considered that the current blade wear will not affect the machining accuracy of the part, and the machining of the next part continues.

[0101] If the slope K j If the slope is greater than the first preset slope K1, it indicates that the current degree of blade wear will affect the machining accuracy of the part. Therefore, combined with the slope K... j The difference between the slope and the first preset slope K1 is used to adjust the single radial movement speed of the tool.

[0102] In the number of machined parts - cutting edge radius curve, the horizontal axis represents the number of machined parts, reflecting the degree of tool use and the cumulative effect of wear, while the vertical axis represents the cutting edge radius, indicating the degree of wear on the tool's cutting edge. By constructing the number of machined parts - cutting edge radius curve, this embodiment of the invention can reflect the tool wear throughout the entire machining process based on the overall shape and trend of the curve, providing a reliable basis for scientifically judging the tool wear state and laying the foundation for subsequent development of highly adaptable modification schemes; and by obtaining the slope K of the j-th part at the node of the number of machined parts - cutting edge radius curve. j It can effectively measure the rate of tool wear, thereby making a judgment on the tool's usage status.

[0103] Specifically, the process of adjusting the maximum single radial movement speed of the tool includes:

[0104] The slope difference A is compared with the set first preset slope difference A1 and second preset slope difference A2, wherein the first preset slope difference A1 is one-third of the difference between the second preset slope K2 and the first preset slope K1, and the second preset slope difference A2 is one-half of the difference between the second preset slope K2 and the first preset slope K1.

[0105] If the slope difference A is less than or equal to the first preset slope difference A1, then the maximum single radial movement speed Vr is corrected by the first radial correction coefficient α1, wherein the corrected maximum single radial movement speed Vr' = Vr × α1, and the first radial correction coefficient α1 = 0.99.

[0106] If the slope difference A is greater than the first preset slope difference A1 and less than or equal to the second preset slope difference A2, then the maximum single radial movement speed Vr is corrected by the second radial correction coefficient α2, wherein the corrected maximum single radial movement speed Vr' = Vr × α2, and the second radial correction coefficient α2 = 0.97.

[0107] If the slope difference A is greater than the second preset slope difference A2, then the maximum single radial movement speed Vr is corrected by the third radial correction coefficient α3, wherein the corrected maximum single radial movement speed Vr' = Vr × α3, and the third radial correction coefficient α3 = 0.94.

[0108] When the tool is in its first wear state, the amount of material processed per unit time may decrease while the maximum single radial traverse speed remains unchanged. Therefore, it is necessary to reduce the maximum single radial traverse speed to ensure machining accuracy. This solution adjusts the maximum single radial traverse speed of the tool by using the slope difference. On the one hand, the slope difference can intuitively reflect the degree of deviation between the current wear state of the tool and the normal reference state. The larger the slope difference, the faster the tool wears. On the other hand, by making the reduction of the maximum single radial traverse speed of the tool proportional to the slope difference, it helps to adjust the maximum single radial traverse speed proportionally based on the comparison results, making the adjustment more scientific and reasonable, and avoiding excessive reduction that affects machining efficiency or insufficient reduction that fails to effectively protect the tool.

[0109] After adjusting the maximum single radial traverse speed for the tool, the process of increasing the maximum axial traverse speed based on the average depth of cut includes:

[0110] The average cutting depth H is compared with the first preset depth H1 and the second preset depth H2, wherein the first preset depth H1 is 1.1 times the average cutting depth output by the CNC machining parameter setting model, and the second preset depth H2 is 1.3 times the average cutting depth output by the CNC machining parameter setting model.

[0111] If the average cutting depth H is less than or equal to the first preset depth H1, then the maximum axial movement speed Va is corrected by the first axial correction coefficient β1, wherein the corrected maximum axial movement speed Va' = Va × β1, and the first radial correction coefficient β1 = 1.09.

[0112] If the average cutting depth H is greater than the first preset depth H1 and less than or equal to the second preset depth H2, then the maximum axial movement speed Va is corrected by the second axial correction coefficient β2, wherein the corrected maximum axial movement speed Va' = Va × β2, and the second radial correction coefficient β2 = 1.13.

[0113] If the average cutting depth H is greater than the second preset depth H2, then the maximum axial movement speed Va is corrected by the third axial correction coefficient β3, wherein the corrected maximum axial movement speed Va' = Va × β3, and the first radial correction coefficient β3 = 1.14.

[0114] Adjusting the maximum single radial traverse speed may result in a deeper cut distance, but the amount of material cut by the tool side per unit time may also decrease. Therefore, it is necessary to adjust the axial traverse speed of the tool simultaneously to ensure the overall accuracy of the part. In this solution, the maximum axial traverse speed is increased based on the average depth of cut, and the increase in the maximum axial traverse speed is inversely proportional to the average depth of cut. This helps to ensure the real-time linkage optimization of the maximum single radial traverse speed and the maximum axial traverse speed, thereby improving machining accuracy and avoiding the impact of tool wear on the accuracy of the part.

[0115] The process of adjusting the maximum single radial traverse speed of the tool, combined with reducing the maximum single radial traverse speed of the tool based on the cumulative usage time of the tool, includes:

[0116] The cumulative usage time T of the tool is obtained, and the cumulative usage time T of the tool is compared with the first preset cumulative usage time T1 and the second preset cumulative usage time T2. The first preset cumulative usage time T1 is half of the tool's service life, and the second preset cumulative usage time T2 is two-thirds of the tool's service life.

[0117] If the cumulative usage time T is less than or equal to the first preset cumulative usage time T1, then the maximum single radial movement speed Vr is corrected by the first duration correction coefficient γ1, wherein the corrected maximum single radial movement speed Vr' = Vr × γ1, and the first duration correction coefficient γ1 is set to 0.99.

[0118] If the cumulative usage time T is greater than the first preset cumulative usage time T1 and less than or equal to the second preset cumulative usage time T2, then the maximum single radial movement speed Vr is corrected by the second duration correction coefficient γ2, wherein the corrected maximum single radial movement speed Vr' = Vr × γ2, and the second duration correction coefficient γ2 is set to 0.98.

[0119] If the cumulative usage time T is greater than the second preset cumulative usage time T2, then the maximum single radial movement speed Vr is corrected by the third duration correction coefficient γ3, wherein the corrected maximum single radial movement speed Vr' = Vr × γ3, and the third duration correction coefficient γ3 is set to 0.96.

[0120] As the cumulative usage time of a cutting tool gradually increases, the tool also gradually becomes dull. Therefore, it is necessary to consider the impact of usage time on the degree of tool dulling. By setting the cumulative usage time of the tool, the maximum single radial movement speed of the tool can be reduced, so that the tool can adjust the maximum single radial movement speed according to time, thereby ensuring the stability of the part's accuracy and avoiding the impact of wear caused by long-term use on the part's accuracy.

[0121] After completing the adjustment of the maximum single radial movement speed of the tool, the cutting edge radius after the tool completes machining the next part is re-determined and the number of machined parts - cutting edge radius curve is updated. Based on the updated curve, the slope under the current node is determined and the tool wear status is determined.

[0122] When the slope is in the first wear state, the maximum single radial movement speed of the tool is redetermined based on the difference between the slope and the first preset slope.

[0123] After adjusting the maximum single radial movement speed of the tool, this embodiment of the invention will remeasure the cutting edge radius and update the curve. This update method helps to ensure that the number of machined parts - cutting edge radius curve can reflect the actual wear of the tool under the current machining conditions in real time, avoiding machining based solely on initial data or fixed parameters. Based on the updated curve, the slope at the current node is determined and the tool wear state is determined. When the slope is determined to be in the first wear state, the maximum single radial movement speed of the tool is re-determined, thus taking into account the instantaneous wear of the tool, making the adjustment more accurate and scientific.

[0124] When the slope is in the first loss state and the maximum single radial traverse speed for the tool needs to be re-determined, if the corrected maximum single radial traverse speed is less than the set critical radial traverse speed, the real-time spindle speed is reduced based on the current tool temperature.

[0125] The real-time spindle speed is inversely proportional to the current tool temperature.

[0126] Specifically, if the corrected maximum single radial traverse speed is still greater than the set critical radial traverse speed, then the process of determining the spindle speed based on the current tool temperature includes:

[0127] Obtain the current tool temperature T, and compare the current tool temperature T with the set first preset tool temperature T1 and second preset tool temperature T2, wherein the first preset tool temperature T1 is 0.85 times the coefficient of thermal expansion of the material used to prepare the part, and the second preset tool temperature T2 is 0.95 times the coefficient of thermal expansion of the material used to prepare the part;

[0128] If the current tool temperature T is less than or equal to the first preset tool temperature T1, then the real-time spindle speed RS is obtained, and the real-time spindle speed RS is corrected using a first speed correction coefficient δ1, wherein the corrected real-time spindle speed RS' = RS × δ1, and the first speed correction coefficient δ1 is set to 0.95;

[0129] If the current tool temperature T is greater than the first preset tool temperature T1 and less than or equal to the second preset tool temperature T2, then the real-time spindle speed RS is obtained, and the real-time spindle speed RS is corrected by the second speed correction coefficient δ2, wherein the corrected real-time spindle speed RS' = RS × δ2, and the second speed correction coefficient δ2 is set to 0.91;

[0130] If the current tool temperature T is greater than the second preset tool temperature T2, the real-time spindle speed RS is obtained, and the real-time spindle speed RS is corrected by the third speed correction coefficient δ3, wherein the corrected real-time spindle speed RS' = RS × δ3, and the third speed correction coefficient δ3 is set to 0.86.

[0131] During part machining, an increase in machining temperature may cause the raw material to expand and deform, affecting the accuracy and quality of the final part. In particular, after recalibrating the maximum single radial traverse speed, a decrease in the maximum single radial traverse speed may bring higher frictional heat, leading to material deformation. Therefore, when it is determined that the maximum single radial traverse speed is less than the set critical radial traverse speed, it is necessary to adjust the spindle speed according to the temperature to ensure that the real-time spindle speed is inversely proportional to the tool temperature. This will help avoid the impact of temperature on part machining, while ensuring machining efficiency, reducing tool wear, and guaranteeing part accuracy.

[0132] The process of determining whether the tool wear state belongs to the second wear state based on the slope of the curve node of the number of machined parts - the radius of the cutting edge arc includes:

[0133] Obtain the slope K of the j-th part at the node of the curve representing the number of processed parts minus the radius of the cutting edge. j , where j = 1, 2, ..., n;

[0134] The slope K j Compare with the second preset slope K2, where K2 = tan(k), k ∈ (40°, 80°], and k represents the angle between the tangent at the j-th node and the x-axis;

[0135] If the slope K j If the slope is greater than the second preset slope K2, the tool wear state is determined to be the second wear state. At this time, there may be severe vibration between the tool and the workpiece being machined, or severe tool wear. The historical slope variance refers to the variance of the slope values ​​obtained after calculating the slope of multiple historical data points. It reflects the distribution of these slope values, that is, the degree of fluctuation of the slope values. When the tool is in the second wear state, a smaller variance means more severe wear, while a larger variance may be caused by vibration. Therefore, it is necessary to combine the historical slope variance to judge the current tool state and make targeted adjustments.

[0136] Specifically, the process of determining the current tool status based on the historical slope variance of the quantity includes:

[0137] Obtain the historical slope variance Q of the quantity, and compare it with the preset slope variance Q1. The specific value of the preset slope variance Q1 can be derived from the actual situation and the pattern of each historical data.

[0138] If the historical slope variance Q of the quantity is less than or equal to the preset slope variance Q1, it indicates that the reason why the tool is in the second wear state is due to normal tool wear, and the tool should be replaced at this time.

[0139] If the historical slope variance Q of the quantity is greater than the preset slope variance Q1, it indicates that the slope of the tool is relatively discrete. This may be because the overhang ratio of the part does not meet the processing requirements, resulting in large vibration between the tool and the processed part, which affects the slope of the curve. At this time, the overhang ratio of the part is adjusted by combining the difference between the historical slope variance of the quantity and the preset slope variance.

[0140] The process of adjusting the overhang ratio of the part based on the difference between the historical slope variance and the preset slope variance includes:

[0141] Obtain the variance difference value C, and compare the variance difference value C with the set first preset variance difference value C1 and second preset variance difference value C2, wherein the first preset variance difference value C1∈[4,9] and the second preset variance difference value C2∈(9,15];

[0142] If the variance difference C is less than or equal to the first preset variance difference C1, then the overhang ratio L of the part is corrected by the first overhang ratio correction coefficient ε1, wherein the corrected overhang ratio L' = L × ε1, and the first overhang ratio correction coefficient ε1 is set to 0.98.

[0143] If the variance difference C is greater than the first preset variance difference C1 and less than or equal to the second preset variance difference C2, then the part overhang ratio L is corrected by the second overhang ratio correction coefficient ε2, wherein the corrected part overhang ratio L' = L × ε2, and the second overhang ratio correction coefficient ε2 is set to 0.95.

[0144] If the variance difference C is greater than the second preset variance difference C2, then the overhang ratio L of the part is corrected by the third overhang ratio correction coefficient ε3, wherein the corrected overhang ratio L' = L × ε3, and the third overhang ratio correction coefficient ε3 is set to 0.91.

[0145] The variance difference refers to the difference between the historical slope variance and the preset slope variance. The larger the variance difference, the more severe the vibration phenomenon. Therefore, the adjustment of the overhang ratio of the part is also greater. Since the reduction of the overhang ratio of the part is proportional to the variance difference, this solution can adjust the overhang ratio of the part according to the degree of vibration, thereby alleviating the vibration and avoiding the problems of vibration-induced part chatter marks or tool breakage. While improving the machining accuracy of the part, it also extends the tool life and avoids tool wear.

[0146] After the overhang ratio of the part is adjusted, the process of correcting the overhang ratio of the part based on the average dimensional deviation of historically machined parts includes:

[0147] Obtain the average dimensional deviation F of the historical parts that have been processed, and compare the average dimensional deviation F with the first preset dimensional deviation value F1 and the second preset dimensional deviation value F2. The first preset dimensional deviation value F1 is 0.7 times the dimensional tolerance corresponding to the part model, and the second preset dimensional deviation value F2 is the dimensional tolerance corresponding to the part model. The dimensional tolerance refers to the absolute value of the difference between the maximum limit size and the minimum limit size specified by the national or industry standards for different models of parts.

[0148] If the average value of the dimensional deviation F is less than or equal to the first preset dimensional deviation value F1, then the overhang ratio L of the part is corrected using the first dimensional correction coefficient σ1, wherein the corrected overhang ratio L' = L × σ1, and the first dimensional correction coefficient σ1 = 0.99;

[0149] If the average value of the dimensional deviation F is greater than the first preset dimensional deviation value F1 and less than or equal to the second preset dimensional deviation value F2, then the overhang ratio L of the part is corrected by the second dimensional correction coefficient σ2, wherein the overhang ratio L' of the part after correction is L×σ2 and the second dimensional correction coefficient σ2 is 0.96.

[0150] If the average value of the dimensional deviation F is greater than the second preset dimensional deviation value F2, then the overhang ratio L of the part is corrected by the third dimensional correction coefficient σ3, wherein the overhang ratio L' of the corrected part is L×σ3, and the third dimensional correction coefficient σ3 is 0.94.

[0151] Adjusting the overhang ratio of a part based on variance difference primarily focuses on the dynamic changes in tool wear. Furthermore, this invention also considers the insights gained from the average dimensional deviation of historically machined parts. The average dimensional deviation refers to the average deviation between the actual size and the design size of historically machined parts. Correcting the overhang ratio based on the average dimensional deviation of historically machined parts, in addition to adjusting the overhang ratio based on variance difference, helps ensure that the machined parts are closer to the design size, thereby improving machining quality, enhancing part machining accuracy, and mitigating tool wear.

[0152] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for setting CNC machining parameters for multi-functional cutting tools based on deep learning, characterized in that, include: A machining threshold model and a CNC machining parameter setting model are established based on deep learning algorithms; Obtain a training dataset, obtain the machining threshold corresponding to the basic information of the cutting tool from the training dataset, train the machining threshold model based on several machining thresholds, and obtain the cutting tool machining parameters and part overhang ratio corresponding to the part model and machining threshold from the training dataset, and train the CNC machining parameter setting model based on several cutting tool machining parameters and part overhang ratio. The basic information of the cutting tool is obtained and input into the trained machining threshold model to obtain several machining thresholds. The several machining thresholds and the part model are then input into the CNC machining parameter setting model to obtain model output parameters, including several cutting tool machining parameters and part overhang ratio. Based on the model output parameters, the parts are machined, the cutting edge radius of the tool is obtained, and a curve of the number of machined parts - cutting edge radius is constructed. The wear state of the tool is determined based on the slope at the corresponding node, and it is determined whether to correct the machining threshold for the tool. In response to the processing threshold being adjusted to a critical value, it is determined whether to correct the part overhang ratio based on the wear state of the tool after completing the processing of a single part. The processing thresholds include the maximum single radial movement speed and the maximum axial movement speed.

2. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 1, characterized in that, The process of determining the tool wear state based on the slope at the node of the curve representing the number of machined parts versus the radius of the cutting edge includes: Obtain the slope of a single part at a node in the curve of number of processed parts - radius of the cutting edge arc; When the slope is greater than the first preset slope, the tool wear state is determined to be the first wear state, and the maximum single radial movement speed of the tool is adjusted based on the difference between the slope and the first preset slope. Wherein, the maximum single radial movement speed of the tool is the maximum radial movement speed of the tool in one feed action.

3. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 2, characterized in that, The process of adjusting the maximum single radial traverse speed of the tool includes: The slope difference is compared with the preset slope difference, and the maximum single radial movement speed of the tool is reduced based on the comparison result. The reduction in the maximum single radial movement speed of the tool is proportional to the slope difference. The slope difference refers to the difference between the slope and the first preset slope.

4. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 3, characterized in that, After adjusting the maximum single radial traverse speed for the tool, the maximum axial traverse speed is increased based on the average depth of cut. The increase in the maximum axial movement speed is inversely proportional to the average depth of cut.

5. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 3, characterized in that, During the adjustment of the maximum single radial traverse speed of the tool, the maximum single radial traverse speed of the tool is reduced in combination with the cumulative usage time of the tool, and the maximum single radial traverse speed of the tool is inversely proportional to the cumulative usage time.

6. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 2, characterized in that, After adjusting the maximum single radial movement speed of the tool, the radius of the cutting edge arc after the tool finishes machining the next part is re-determined and the number of machined parts - radius of the cutting edge arc is updated. Based on the updated curve, the slope at the current node is determined and the tool wear status is determined. When the slope is in the first wear state, the maximum single radial movement speed of the tool is redetermined based on the difference between the slope and the first preset slope.

7. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 6, characterized in that, When the slope is in the first loss state and the maximum single radial traverse speed for the tool needs to be re-determined, if the corrected maximum single radial traverse speed is less than the set critical radial traverse speed, the real-time spindle speed is reduced based on the current tool temperature. The real-time spindle speed is inversely proportional to the current tool temperature.

8. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 1, characterized in that, The process of determining whether the tool wear state belongs to the second wear state based on the slope at the node of the curve of number of processed parts - radius of cutting edge includes: Obtain the slope of a single part at a node in the curve of number of processed parts - radius of the cutting edge arc; When the slope is greater than the set second preset slope, the tool wear state is determined to be the second wear state, and the tool state is adjusted based on the historical slope variance of the quantity. If the historical slope variance of the quantity is greater than the preset slope variance, the overhang ratio of the part is adjusted based on the difference between the historical slope variance of the quantity and the preset slope variance. If the historical slope variance of the stated quantity is less than or equal to the preset slope variance, then the tool should be replaced.

9. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 8, characterized in that, The process of adjusting the overhang ratio of the part includes: The variance difference value is compared with the preset variance difference value, and the overhang ratio of the part is reduced based on the comparison result; The reduction in the overhang ratio of the part is proportional to the variance difference. The variance difference refers to the difference between the historical slope variance and the preset slope variance.

10. The method for setting CNC machining parameters for multi-functional cutting tools based on deep learning according to claim 9, characterized in that, After the overhang ratio of the part is adjusted, the reduction in the overhang ratio of the part is corrected based on the average dimensional deviation of the historical parts that have been processed. The reduction in the overhang ratio of the part is proportional to the average dimensional deviation of the historical parts.

Images