Cutter structure data prediction method and device, equipment and storage medium

Abstract

The invention relates to the technical field of tool structure data prediction, in particular to a tool structure data prediction method, device and equipment and a storage medium, and the method comprises the steps: firstly constructing a data sample library for training, and executing simulation operation based on the data sample library to construct a training sample set; performing multi-round iterative training on the initial index prediction model by using the training sample set, and calling the obtained target index prediction model to perform machining key index prediction on each to-be-predicted cutter structure data group to obtain an index data group corresponding to each to-be-predicted cutter structure data group, obtaining an initial tool structure data set based on the plurality of to-be-predicted tool structure data sets, performing multi-round iterative updating on the initial tool structure data set based on the plurality of to-be-predicted tool structure data sets and the plurality of index data sets by adopting a genetic algorithm, and obtaining an optimized tool structure data set when a preset iteration stop condition is met, the method aims at accurately predicting the key physical quantity of the tool structure.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tool structure data prediction, and particularly relates to a tool structure data prediction method, device, equipment and storage medium. BACKGROUND

[0002] Fused quartz occupies an important position in the fields of aerospace, laser electronics and high-generation inkjet printing OLED (organic light-emitting diode) manufacturing due to its excellent performance. Key components such as print head micro-hole plates and mask plate substrates made of fused quartz have strict requirements on the surface shape precision, surface integrity and hole precision of ultra-precision machining. In ultra-precision grinding machining, subsurface damage directly affects the performance and reliability of components, and is related to the display screen pixel precision and production yield, and becomes the core bottleneck restricting the processing quality.

[0003] In order to meet the requirements of high precision and low defects, in-situ laser-assisted ultra-precision machining technology and negative rake angle tools are applied. The former can improve the ductility domain machining performance of fused quartz and reduce cracks, and the latter can suppress brittle fracture and promote plastic removal by enhancing hydrostatic pressure. Tool rake angle, tool relief angle and machining temperature are key parameters affecting machining effect, and reasonable matching of the three can optimize material removal mode, but excessive temperature can cause thermal damage, and the design of the relief angle needs to balance between cutting stability and tool strength.

[0004] Existing research still has many limitations. The removal mechanism in fused quartz machining is not fully clear, and micro changes are difficult to directly observe. The traditional finite element method is prone to grid distortion when simulating large deformation and crack propagation, and traditional prediction methods such as response surface model are difficult to capture the complex relationship of highly nonlinear systems. Existing tool optimization patents are mostly for conventional machining, and do not fully adapt to the micro-nano scale characteristics and brittle-plastic mechanism of ultra-precision machining, and have insufficient universality and precision. SUMMARY

[0005] In order to overcome the deficiencies of the prior art, the purpose of the present application is to provide a tool structure data prediction method, device, equipment and storage medium, which aims to accurately predict the key physical quantities of the tool structure, and to realize efficient optimization of the tool structure and machining parameters, so as to improve the machining quality and efficiency.

[0006] The first aspect of the present application provides a tool structure data prediction method, comprising: constructing a data sample library for training, performing simulation operations based on the data sample library to construct a training sample set; constructing an initial index prediction model, and performing multi-round iterative training on the initial index prediction model by using the training sample set to obtain a target index prediction model; obtaining a plurality of groups of tool structure data to be predicted, calling the target index prediction model to respectively predict machining key indicators of each group of the tool structure data to be predicted, and obtaining an index data group corresponding to each group of the tool structure data to be predicted; obtaining an initial tool structure data group based on the plurality of groups of tool structure data to be predicted; and performing multi-round iterative updating on the initial tool structure data group based on the plurality of groups of tool structure data to be predicted and the plurality of groups of index data by using a genetic algorithm, and when a preset iteration stop condition is met, taking the updated tool structure data group as an optimized tool structure data group.

[0007] Optionally, in the first implementation manner of the first aspect of the present application, the training sample set includes a training set and a test set, and the data sample library includes a plurality of tool machining parameters; the simulation operations are performed based on the data sample library to construct the training sample set, including: performing full permutation and combination on the plurality of tool machining parameters by using a full permutation and combination algorithm to obtain a plurality of input working condition combinations; performing simulation operations based on each working condition combination respectively to obtain a simulation index data group corresponding to each working condition combination; and integrating the plurality of tool machining parameters and the simulation index data group corresponding to each working condition combination to obtain the training sample set; and the training sample set is divided into the training set and the test set based on a preset division ratio.

[0008] Optionally, in the second implementation manner of the first aspect of the present application, the initial index prediction model includes an initial Actor network and an initial Critic network; the multi-round iterative training on the initial index prediction model by using the training sample set to obtain the target index prediction model includes: constructing an initial experience replay pool based on the training sample set, and obtaining a current state sample and a next state sample from the initial experience replay pool; performing action prediction on the current state sample by using the initial Actor network to obtain a current action; obtaining a predefined reward function, calculating a reward value of the current action based on the reward function to obtain a current reward value; storing the current state sample, the current action, the current reward value and the next state sample into the initial experience replay pool to obtain an expanded experience replay pool; and performing multi-round iterative updating on the initial Actor network and the initial Critic network based on the expanded experience replay pool to obtain the target index prediction model.

[0009] Optionally, in a third implementation form of the first aspect of the present application, the method further includes: obtaining the current state sample, the current action, the current reward value and the next state sample based on the expanded experience replay pool; performing action prediction on the next state sample by using the initial Actor network to obtain a next action; performing value evaluation processing based on the current state sample, the current action, the next state sample and the next action by using the initial Critic network to obtain a current state value and a next state value; performing error calculation based on the current reward value, the current state value and the next state value to obtain a TD error; and performing multi-round iterative update processing on the initial Actor network and the initial Critic network based on the TD error to obtain the target index prediction model.

[0010] Optionally, in a fourth implementation form of the first aspect of the present application, the method further includes: performing multi-round iterative update processing on the initial Critic network based on the TD error by using a back propagation algorithm to obtain an optimized Critic network; performing multi-round iterative update processing on the initial Actor network based on the TD error by using a policy gradient algorithm to obtain an optimized Actor network; and combining the optimized Critic network and the optimized Actor network to form the target index prediction model.

[0011] Optionally, in a fifth implementation form of the first aspect of the present application, the genetic algorithm includes a roulette algorithm, a crossover operation algorithm and a mutation operation algorithm; and the method further includes: performing roulette selection processing on the plurality of groups of the to-be-predicted tool structure data groups based on the plurality of groups of the index data groups by using the roulette algorithm to obtain a plurality of groups of parent tool structure data groups; performing gene exchange processing on the plurality of groups of the parent tool structure data groups by using the crossover operation algorithm to obtain a plurality of groups of crossover tool structure data groups; performing gene fine-tuning processing on the plurality of groups of the crossover tool structure data groups by using the mutation operation algorithm to obtain a plurality of groups of mutation tool structure data groups; performing multi-round iterative update processing on the initial tool structure data group based on the plurality of groups of the mutation tool structure data groups; and when a preset iteration stop condition is met, taking the updated tool structure data group as an optimized tool structure data group.

[0012] Optionally, in a sixth implementation form of the first aspect of the present application, the adopting the roulette algorithm, based on multiple groups of the index data groups, performing roulette selection processing on multiple groups of the to-be-predicted tool structure data groups to obtain multiple groups of parent tool structure data groups, comprises: obtaining a pre-defined fitness function; based on the fitness function, performing fitness calculation on multiple groups of the index data groups to obtain fitness values corresponding to multiple groups of the to-be-predicted tool structure data groups; adopting the roulette algorithm, based on multiple fitness values, calculating roulette selection probabilities to obtain roulette selection probabilities corresponding to multiple groups of the to-be-predicted tool structure data groups; obtaining a pre-set selection rule, based on the selection rule and multiple roulette selection probabilities, performing roulette selection processing on multiple groups of the to-be-predicted tool structure data groups to obtain multiple groups of the parent tool structure data groups.

[0013] The second aspect of the present application provides a tool structure data prediction device, comprising: a simulation module, configured to construct a data sample library for training, perform simulation operation based on the data sample library to construct a training sample set; a training model, configured to construct an initial index prediction model, perform multi-round iterative training on the initial index prediction model by using the training sample set to obtain a target index prediction model; a prediction module, configured to obtain multiple groups of to-be-predicted tool structure data groups, call the target index prediction model to respectively predict machining key indexes of each group of the to-be-predicted tool structure data groups to obtain an index data group corresponding to each group of the to-be-predicted tool structure data groups; a data acquisition module, configured to obtain an initial tool structure data group based on multiple groups of the to-be-predicted tool structure data groups; an optimization module, configured to adopt a genetic algorithm, based on multiple groups of the to-be-predicted tool structure data groups and multiple groups of the index data groups, perform multi-round iterative update on the initial tool structure data group, and when a pre-set iteration stop condition is met, take the updated tool structure data group as an optimized tool structure data group.

[0014] The third aspect of the present application provides a tool structure data prediction device, comprising: a memory and at least one processor, the memory storing instructions; at least one processor calling the instructions in the memory to make the tool structure data prediction device perform each step of the tool structure data prediction method described in any of the above.

[0015] The fourth aspect of the present application provides a computer readable storage medium, the computer readable storage medium storing instructions, the instructions being executed by a processor to implement each step of the tool structure data prediction method described in any of the above.

[0016] In the technical solution of the present application, first, a data sample library for training is constructed, a simulation operation is performed based on the data sample library to construct a training sample set, then an initial index prediction model is constructed, the initial index prediction model is trained by multiple rounds of iteration using the training sample set to obtain a target index prediction model, then a plurality of groups of to-be-predicted tool structure data groups are acquired, the target index prediction model is called to predict the machining key indicators of each group of to-be-predicted tool structure data groups, and index data groups corresponding to each group of to-be-predicted tool structure data groups are obtained, an initial tool structure data group is acquired based on the plurality of groups of to-be-predicted tool structure data groups, finally, a genetic algorithm is used to perform multiple rounds of iterative updating on the initial tool structure data group based on the plurality of groups of to-be-predicted tool structure data groups and the plurality of groups of index data groups, and when a preset iteration stopping condition is met, the updated tool structure data group is taken as an optimized tool structure data group, aiming to accurately predict the key physical quantities of the tool structure and realize efficient optimization of the tool structure and the machining parameters, so as to improve the machining quality and efficiency, improve the machining consistency, surface shape precision and surface quality of the fused quartz element, effectively prolong the tool life and reduce the energy consumption, can fully meet the extreme requirements of the inkjet printing equipment on the micro-hole precision, positioning stability and wear resistance of the key quartz parts, and provide a new way for large-scale, low-loss precision manufacturing of high-value-added optical glass elements. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A logic flow chart of a tool structure data prediction method provided for an embodiment of the present application is shown in the figure. Figure 2 A structural schematic diagram of a tool structure data prediction device provided for an embodiment of the present application is shown in the figure. Figure 3 A structural schematic diagram of a tool structure data prediction device provided for an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0018] The present application provides a tool structure data prediction method, device, equipment and storage medium, in the present application, the terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "includes" or "has" and any variation thereof is intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0019] For ease of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 One embodiment of the tool structure data prediction method in the embodiments of the present application includes: 101, constructing a data sample library for training, performing simulation operation based on the data sample library to construct a training sample set; In this embodiment, the data sample library for training needs to take the structural parameters of quartz glass and diamond tool and cutting temperature as the core variables, and determine multiple groups of values of tool rake angle, tool relief angle and cutting temperature as the data sample library. When performing simulation operation based on the sample library, the corresponding cutting simulation model is built in Abaqus software relying on the smoothed particle dynamics method, wherein the workpiece adopts the SPH (Smoothed Particle Hydrodynamics) particle model, the model parameters can be set as , the total particle number is set to 600000, the tool is configured as a diamond tool in the form of Lagrange grid, and the model parameters strictly match the physical properties such as density, Poisson's ratio and specific heat capacity of quartz glass and diamond. The simulation operation based on the SPH method can construct the training sample set, which can accurately capture the nonlinear behaviors such as large deformation and crack propagation in the cutting process of hard and brittle materials, effectively avoid the mesh mismatch problem compared with the traditional finite element method, and make the output physical quantities in the sample set more consistent with the actual machining mechanical and material removal characteristics.

[0020] 102, constructing an initial index prediction model, and performing multi-round iterative training on the initial index prediction model by using the training sample set to obtain a target index prediction model; In the embodiment, the initial index prediction model is constructed using an Actor-Critic deep reinforcement learning framework. The Actor network uses a neural network as the core structure, with the input being machining parameters such as tool rake angle, relief angle, and cutting temperature, and the output being key machining physical quantities such as shear angle, shear plane length, and maximum hydrostatic pressure, and undertakes the function of generating continuous prediction actions. The Critic network is also built based on a neural network, with the input being the current machining parameters and the predicted physical quantities output by the Actor, and the output being the value evaluation value corresponding to the state-action combination, which realizes quantitative judgment of the quality of the prediction results. The two networks cooperatively constitute the core architecture of the initial index prediction model. In the process of multiple rounds of iterative training of the initial index prediction model, the input parameters (tool rake angle, relief angle, and cutting temperature) and the corresponding machining physical quantity data (maximum hydrostatic pressure, high pressure particle number, shear angle in the cutting area, and shear plane length) in the training sample set are used to continuously adjust and optimize the model parameters, so that the model gradually learns the correlation between the input and the output. When the prediction accuracy of the model meets the preset threshold or the number of training reaches a certain number, the target index prediction model is obtained. The target index prediction model can accurately adapt to the complex nonlinear relationship between the tool and process parameters and the key machining physical quantities in the ultra-precision machining of fused quartz, and quickly output the machining physical quantity prediction results corresponding to different parameter combinations, providing direct and reliable quantitative basis for the intelligent optimization of the tool structure, and effectively improving the surface shape precision and surface integrity of the ultra-precision machining of fused quartz components.

[0021] 103. Obtain a plurality of sets of predicted tool structure data, and call the target index prediction model to predict the machining key indicators for each set of predicted tool structure data, to obtain an index data set corresponding to each set of predicted tool structure data; In the embodiment, the obtained multiple sets of tool structure data to be predicted, specifically referring to the tool rake angle, relief angle and cutting temperature parameter set covering different value combinations, the parameter value range fits the actual process requirements of fused quartz ultra-precision machining. By reasonably covering the potential optimization interval of the key parameters, the data set to be predicted is formed. When calling the trained target index prediction model, the rake angle, relief angle and cutting temperature parameters in each set of tool structure data to be predicted are used as model inputs. Based on the correlation between the learned parameters and the machining indexes of the model, the corresponding machining key index results are quickly output by the model, that is, each set of input parameter combination corresponds to an index data set. Each set of index data set includes maximum hydrostatic pressure, high pressure particle number, cutting region shear angle and shear plane length. The target index prediction model can complete the machining index prediction of multiple parameter combinations in an efficient and accurate manner, greatly reducing the time cost and calculation cost of parameter evaluation. At the same time, the predicted index data set can directly quantify the machining effect corresponding to different tool structure parameters, providing comprehensive and intuitive decision basis for subsequent multi-objective optimization of tool structure, helping to quickly select the optimal tool structure parameter combination that meets the requirements of ultra-precision machining, and then ensuring the machining surface shape precision and surface integrity of fused quartz parts, and improving the optimization efficiency and reliability of the machining process.

[0022] 104. Obtain an initial tool structure data set based on the multiple sets of tool structure data to be predicted. In the embodiment, the initial tool structure data set is obtained based on the multiple sets of tool structure data to be predicted. Essentially, within the reasonable value range of the tool rake angle, relief angle and cutting temperature of the data set to be predicted, a certain number of parameter combinations are selected by random sampling to form the initial tool structure data set, providing diversified search starting points for subsequent tool structure optimization. This helps the optimization algorithm to fully explore the potential optimal solution in the parameter space, improves the efficiency and global optimization ability of the optimization process, and lays a balanced and reliable foundation for finally selecting the optimal tool structure parameter combination that meets the requirements of ultra-precision machining.

[0023] 105. Use a genetic algorithm to perform multiple rounds of iterative updates on the initial tool structure data set based on the multiple sets of tool structure data to be predicted and the multiple sets of index data sets. When the preset iteration stopping condition is met, the updated tool structure data set is used as the optimized tool structure data set.

[0024] In the embodiment, when the genetic algorithm is used for tool structure parameter optimization, a plurality of groups of tool structure data groups to be predicted are used as parameter search bases, corresponding index data groups are used as optimization evaluation bases, and the initial tool structure data group randomly extracted is subjected to multiple rounds of iteration and update. In the iteration process, the genetic algorithm evaluates and selects the adaptability of each tool structure parameter combination based on the machining effect reflected by the index data group, until the preset iteration stopping condition is met, at this time, the tool structure data group updated finally is determined as the optimized tool structure data group. The genetic algorithm has strong global optimization capability, can efficiently explore the optimal solution in a wide parameter space based on the quantitative feedback of the key machining indicators, and effectively avoid the local optimal problem. At the same time, the optimization process is closely combined with the process feasible interval of the data group to be predicted and the quantitative results of the machining effect of the index data group, so that the optimized tool structure data group obtained finally can accurately adapt to the quality demand of fused quartz ultra-precision machining, ensure that the material removal effect in the machining process is better, and then improve the surface shape precision and surface integrity of the part, and provide reliable support for efficient optimization of the machining process.

[0025] In the embodiment of the present application, the training sample set includes a training set and a test set, and the data sample library includes a plurality of tool machining parameters. The simulation operation based on the data sample library is performed to construct the training sample set, which includes: using a full permutation combination algorithm to perform full permutation combination on the plurality of tool machining parameters to obtain a plurality of input working condition combinations; performing simulation operation based on each working condition combination respectively to obtain a simulation index data group corresponding to each working condition combination; integrating the plurality of tool machining parameters and the simulation index data group corresponding to each working condition combination to obtain the training sample set; and dividing the training sample set into the training set and the test set based on a preset division ratio.

[0026] In the embodiment, the data sample library covers a plurality of tool machining parameters, specifically tool rake angle, relief angle and cutting temperature. A full permutation and combination algorithm is used to perform full permutation and combination on the plurality of machining parameters, so as to obtain a plurality of input condition combinations. For example, when the tool rake angle includes four values of 0°, -15°, -35° and -65°, the relief angle includes four values of 2°, 5°, 10° and 15°, and the cutting temperature includes seven gradient values between 25℃ and 600℃, the full permutation and combination algorithm is used to perform combination operation on the plurality of tool machining parameters, so as to obtain 112 different input condition combinations. Each combination completely covers the value dimensions of the tool rake angle, the relief angle and the cutting temperature, and fits the process feasible interval of the fused quartz ultra-precision machining. Then, simulation operations are respectively performed based on each input condition combination, that is, corresponding cutting simulation models are constructed in Abaqus software based on the smoothed particle hydrodynamics method. The workpiece adopts an SPH particle model of 240×40×4μm³, the total particle number is 600000, the tool is configured as a diamond tool in the form of a Lagrange grid, and the model parameters strictly match the physical properties such as the density and Poisson's ratio of quartz glass and diamond. Through simulation calculation, machining characteristic data under each condition can be obtained. For example, for the condition combination of a rake angle of -15° and a temperature of 25℃, the simulation shows that the material is removed in a brittle mode and discontinuous fragments are generated, and the corresponding shear angle and shear plane length are obtained. For the condition combination of a rake angle of 15° and a temperature of 600℃, the simulation shows that the material exhibits plastic flow characteristics and the fragments are continuous, and the corresponding shear angle data are also obtained. For the condition combination of a rake angle gradually changing from 0° to -65°, the simulation can obtain the results that the shear plane length continuously increases and the maximum hydrostatic pressure gradually increases. These data jointly constitute a simulation index data set corresponding to each condition combination, that is, the shear angle, the shear plane length, the maximum hydrostatic pressure and the number of particles with a hydrostatic tensile stress less than -200MPa under each condition are obtained through simulation calculation, to form a simulation index data set corresponding to the condition combination. Then, a plurality of tool machining parameters and corresponding simulation index data sets are integrated to form a structured training sample set. Based on a preset division ratio (such as 9:1), the training sample set is divided into a training set for model parameter learning and a test set for verifying the generalization ability of the model.The application of the full permutation combination algorithm ensures that the input working condition combination can fully cover the potential value interval of the tool machining parameter, and avoids the lack of local parameter dimension leading to insufficient representativeness of the sample set; the simulation operation based on the smoothed particle dynamics method can accurately capture the nonlinear mechanical behavior in the cutting process of the hard and brittle material, so that the simulation index data set is more in line with the actual material removal and stress distribution characteristics in the machining, and high-quality label data is provided for the training sample set; and the reasonable division of the training set and the test set ensures that the model can fully learn the correlation between the parameters and the indexes through the training set, and the generalization performance of the model can be effectively verified through the test set, so that the model is prevented from being over-fitted, and a reliable data foundation is laid for the training of the subsequent index prediction model.

[0027] In the embodiment of the application, the initial index prediction model comprises an initial Actor network and an initial Critic network; the multi-round iterative training of the initial index prediction model based on the training sample set to obtain a target index prediction model comprises: constructing an initial experience replay pool based on the training sample set, and obtaining a current state sample and a next state sample from the initial experience replay pool; predicting an action for the current state sample by using the initial Actor network to obtain a current action; obtaining a predefined reward function, calculating a reward value for the current action based on the reward function to obtain a current reward value; storing the current state sample, the current action, the current reward value and the next state sample into the initial experience replay pool to obtain an expanded experience replay pool; and performing multi-round iterative update processing on the initial Actor network and the initial Critic network based on the expanded experience replay pool to obtain the target index prediction model.

[0028] In the embodiment, the initial index prediction model is composed of an initial Actor network and an initial Critic network, both of which are core structures of neural networks. When the training sample set is used to perform multi-round iterative training, an initial experience replay pool is first constructed based on the training sample set. The initial experience replay pool is used to store experience samples containing states, actions, rewards and next states. The tool rake angle, tool relief angle and cutting temperature in the training sample set are state samples, from which the current state sample and the next state sample can be obtained from the initial experience replay pool. Then, the initial Actor network is used to predict an action for the current state sample. The action is a continuous prediction of the machining key indexes such as the maximum hydrostatic pressure, the number of high-pressure particles, the shear angle of the cutting area and the length of the shear plane, and the current action is obtained. Then, a predefined reward function is obtained. The function takes the negative value of the mean square error of the predicted value and the true value as the calculation basis, and calculates the true machining index corresponding to the current state sample and the current action based on this to obtain the current reward value. The calculation expression of the reward function is: , in, This indicates the current action, specifically the predicted values ​​for key machining parameters such as maximum hydrostatic pressure, number of high-pressure particles, shear angle in the cutting zone, and length of the shear plane. This represents the actual value corresponding to the current action. This indicates the dimension of the output variable, corresponding to the number of key processing indicators that need to be predicted (such as maximum hydrostatic pressure, number of high-pressure particles, etc.). Indicates representative t The reward value at any given moment.

[0029] Subsequently, the current state sample, current action, current reward value, and next state sample are stored in the initial experience replay pool to form an expanded experience replay pool, enabling continuous accumulation and updating of experience samples. Then, based on the expanded experience replay pool, multiple rounds of iterative updates are performed on the initial Actor network and the initial Critic network until the network converges, ultimately obtaining the target index prediction model. This model achieves continuous and accurate prediction of key machining indicators through the Actor network and continuously corrects the prediction strategy through the value evaluation of the Critic network. This allows the target index prediction model to fully capture the complex nonlinear relationship between tool parameters and machining indicators, enhancing the model's generalization ability and prediction accuracy, and providing reliable quantitative support for subsequent intelligent optimization of tool structures.

[0030] In this embodiment of the invention, the step of performing multiple rounds of iterative updates on the initial Actor network and the initial Critic network based on the expanded experience replay pool to obtain the target indicator prediction model includes: obtaining the current state sample, the current action, the current reward value, and the next state sample based on the expanded experience replay pool; using the initial Actor network to predict the action of the next state sample to obtain the next action; using the initial Critic network to perform value evaluation processing based on the current state sample, the current action, the next state sample, and the next action to obtain the current state value and the next state value; calculating the error based on the current reward value, the current state value, and the next state value to obtain the TD (Temporal Difference) error; and performing multiple rounds of iterative updates on the initial Actor network and the initial Critic network based on the TD error to obtain the target indicator prediction model.

[0031] In this embodiment, when iterating the initial Actor network and the initial Critic network multiple times based on the expanded experience replay pool, the current state sample, current action, current reward value, and next state sample are first obtained from the expanded experience replay pool. The current state sample is a combination of the tool rake angle, clearance angle, and cutting temperature. The current action is the prediction result of the machining index for this state by the initial Actor network, and the current reward value is calculated using a predefined reward function. Subsequently, the initial Actor network is used to predict the action of the next state sample to obtain the next action, i.e., continuous prediction of the machining index corresponding to the next state sample. Simultaneously, the initial Critic network is used to perform value evaluation processing, using the current state sample... With the current action As input, we obtain the current state value. The following is a sample of the state. With the next action As input, we obtain the value of the next state. Next, based on the current reward value and the current state value... and the value of the next state Error calculation is performed to obtain the TD error. The specific calculation expression is as follows: , in, Indicates TD error, This represents the current reward value. This represents the discount factor, typically set to 0.99, used to ensure algorithm convergence. Indicates the value of the next state. Indicates the value of the current state. express The current state sample at time t. express The current action at any given moment. express The next state sample at time t. express The next action at any given moment.

[0032] Finally, based on the TD error, the initial Actor network and the initial Critic network are iteratively updated multiple times until the network converges, ultimately obtaining the target index prediction model. The introduction of the TD error enables precise quantification of state value deviation, allowing the evaluation capability of the initial Critic network to continuously iterate and providing reliable feedback for the policy optimization of the initial Actor network.

[0033] In the embodiment of the present application, the multi-round iterative updating of the initial Actor network and the initial Critic network based on the TD error to obtain the target index prediction model comprises: using a back propagation algorithm to perform multi-round iterative updating of the initial Critic network based on the TD error to obtain an optimized Critic network; using a policy gradient algorithm to perform multi-round iterative updating of the initial Actor network based on the TD error to obtain an optimized Actor network; and combining the optimized Critic network and the optimized Actor network to form the target index prediction model.

[0034] In the embodiment, when the initial Actor network and the initial Critic network are updated based on the TD error, first, the initial Critic network is updated using a back propagation algorithm. The TD error is transmitted from the output layer to the input layer of the network layer by layer through the back propagation algorithm, and the weights and thresholds of the initial Critic network are adjusted according to the error size to minimize the deviation of the value evaluation. After multiple iterations, the network converges, and an optimized Critic network with accurate value evaluation capability is obtained. For the initial Actor network, the function is to generate a predicted action of the processing index based on the state sample, and the policy gradient algorithm is used for updating. The specific calculation expression is: wherein, represents the Actor network parameters, i.e., the weights and thresholds of the neural network, and represents the assignment operator, which is used to represent the update operation of the parameters, represents the learning rate, represents the gradient operator, which clearly defines the solution object of the gradient as the Actor network parameters represents the prediction strategy of the Actor network, represents the TD error. The algorithm takes the TD error output by the optimized Critic network as the feedback signal, optimizes the prediction strategy by adjusting the network parameters, so that the prediction result of the initial Actor network is closer to the actual value of the processing index. After multiple iterations, an optimized Actor network with accurate prediction capability is obtained. Finally, the optimized Critic network and the optimized Actor network are combined to form a target index prediction model. The optimized Actor network is responsible for receiving state samples such as tool rake angle, tool relief angle and cutting temperature, and outputting prediction results of processing indexes such as maximum hydrostatic pressure and shear angle. The optimized Critic network continuously evaluates the value of the action output by the optimized Actor network, which provides a basis for subsequent fine-tuning of the prediction strategy. The two work together to achieve accurate prediction of processing indexes.

[0035] ​​In this embodiment of the invention, the genetic algorithm includes a roulette wheel algorithm, a crossover algorithm, and a mutation algorithm. The step of using the genetic algorithm to iteratively update an initial tool structure data set based on multiple sets of data for the tool structure to be predicted and multiple sets of data for the index, and using the updated tool structure data set as the optimized tool structure data set when a preset iteration stopping condition is met, includes: using the roulette wheel algorithm to perform roulette wheel selection on multiple sets of data for the tool structure to be predicted based on multiple sets of data for the index, obtaining multiple sets of parent tool structure data sets; using the crossover algorithm to perform gene exchange on multiple sets of parent tool structure data sets, obtaining multiple sets of crossover tool structure data sets; using the mutation algorithm to perform gene fine-tuning on multiple sets of crossover tool structure data sets, obtaining multiple sets of mutated tool structure data sets; and using the multiple sets of mutated tool structure data sets to iteratively update the initial tool structure data set through multiple rounds of iteration, and using the updated tool structure data set as the optimized tool structure data set when a preset iteration stopping condition is met.

[0036] In this embodiment, when using a genetic algorithm to iteratively update the initial tool structure data set in multiple rounds, the first step is to process multiple sets of tool structure data sets to be predicted and their corresponding index data sets using a roulette wheel algorithm. Specifically, the parent generation is selected based on the fitness value corresponding to each set of tool structure data sets to be predicted. Individuals with higher fitness values ​​have a greater probability of being selected. A random number between 0 and 1 is generated. If this random number is greater than the probability of an individual being selected, that individual is retained for the next iteration. Then, a crossover operation algorithm is used to perform gene exchange processing on the parent tool structure data sets. Here, the genes correspond to parameters such as tool rake angle, clearance angle, and cutting temperature. By selecting and exchanging the corresponding parameter dimensions of different parent data sets, the parameter combinations are recombined, resulting in multiple sets of cross-cutting tool structure data sets, thereby expanding the diversity of parameter search. Next, a mutation operation algorithm is used to perform gene fine-tuning processing on the cross-cutting tool structure data sets, specifically fine-tuning parameters (such as tool rake angle). By introducing small-amplitude random parameter changes, the algorithm avoids getting trapped in local optima. After this processing, multiple sets of mutated tool structure data sets are obtained. The specific calculation expression is as follows: , in, Indicates the relationship with the first The iteration, the... The rake angle value corresponding to the group of tool structure data. Indicates the relationship with the first The iteration, the... The rake angle value corresponding to the group of tool structure data. This represents a random parameter, usually a normal distribution, ranging from -1 to 1.

[0037] After that, the initial tool structure data set is updated by multiple rounds of iteration based on multiple sets of variant tool structure data sets, each round of iteration repeatedly performing the processes of roulette selection, crossover and mutation, until a preset iteration stop condition is met, that is, a maximum number of iterations is reached, or the difference between the fitness values of two consecutive iterations is less than a threshold value, at which time the updated tool structure data set is determined as the optimized tool structure data set. The roulette algorithm realizes the directional selection of high-quality parameter combinations relying on the fitness function, ensuring that the iteration process converges to a better machining effect; the crossover operation algorithm expands the diversity of parameter combinations and improves the coverage of the parameter search space; and the mutation operation algorithm introduces new parameter variations through random fine-tuning, effectively avoiding local optimal problems. The three work together to enable the genetic algorithm to efficiently explore the global optimal solution in the tool parameter space, and the optimized tool structure data set obtained finally can accurately adapt to the process requirements of fused quartz ultra-precision machining, significantly improving the machining surface quality and process stability.

[0038] In the embodiment of the application, the roulette algorithm is adopted to perform roulette selection processing on multiple sets of the to-be-predicted tool structure data sets based on multiple sets of the index data sets, to obtain multiple sets of parent tool structure data sets, which comprises: obtaining a predefined fitness function; calculating the fitness of multiple sets of the index data sets based on the fitness function, to obtain fitness values corresponding to multiple sets of the to-be-predicted tool structure data sets; adopting the roulette algorithm to calculate roulette selection probabilities based on multiple fitness values, to obtain roulette selection probabilities corresponding to multiple sets of the to-be-predicted tool structure data sets; obtaining a preset selection rule, and performing roulette selection processing on multiple sets of the to-be-predicted tool structure data sets based on the selection rule and multiple roulette selection probabilities, to obtain multiple sets of the parent tool structure data sets.

[0039] In the embodiment, in the process of roulette selection processing, a predefined fitness function is first obtained, which is used to quantify the machining adaptability of the to-be-predicted tool structure data set. Assuming that the target is to maximize the maximum hydrostatic pressure and maximize the shear plane length , minimize the shear angle and minimize the number of high hydrostatic tensile stress particles , the fitness function can be defined as: , wherein, , represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, represents the rake angle value corresponding to the i-th iteration and the j-th tool structure data set, indicates the rake angle value corresponding to the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the fitness value corresponding to the tool structure data set of the i th iteration of the j th group, indicates the maximum hydrostatic pressure, indicates the shear plane length, indicates the shear angle, indicates the number of high hydrostatic tensile stress particles.

[0040] Subsequently, based on the fitness function, the index data set parameters corresponding to each group of tool structure data sets to be predicted are substituted into the calculation to obtain the fitness value corresponding to each group of tool structure data sets to be predicted. The size of the value directly reflects the processing effect of the corresponding parameter combination.

[0041] Then, the roulette algorithm is used to calculate the selection probability, and the specific calculation expression is: , wherein, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the tool structure data set of the i th iteration of the j th group, indicates the roulette selection probability corresponding to the tool structure data set of the i th iteration of the j th group, indicates the fitness value corresponding to the tool structure data set of the i th iteration of the j th group, indicates the population size, i.e. the number of tool structure data sets to be predicted, which is generally set to 50. The formula determines the probability of each group of tool structure data sets to be predicted being selected based on the proportion of the fitness value of each group. The higher the fitness value of a group, the greater the corresponding selection probability. After that, a preset selection rule is obtained, which is usually a random number between 0 and 1. The selection probabilities of all tool structure data sets to be predicted are accumulated in size to form a continuous interval. The tool structure data set to be predicted corresponding to the interval into which the random number falls is selected. This operation is repeated until the required number of parent tool structure data sets is obtained. The fitness function realizes the quantitative evaluation of the processing effect, so that the advantages and disadvantages of the tool structure data sets to be predicted have a comparable standard. The calculation of the roulette selection probability directs the selection of parameter combinations with higher fitness through a probability mechanism, ensuring that the subsequent iteration process converges to a better processing effect. At the same time, the characteristics of probability selection retain a certain population diversity, avoiding premature convergence to a local optimal solution, and providing a high-quality and diverse parent basis for subsequent crossover and mutation operations.

[0042]

[0043] ​​The tool structure data prediction method in the embodiments of the application is described above, and the tool structure data prediction device in the embodiments of the application is described below. Please refer to Figure 2 One embodiment of the tool structure data prediction device in the embodiments of the application includes: The simulation module 201 is configured to construct a data sample library for training, perform a simulation operation based on the data sample library, and construct a training sample set. The training model 202 is configured to construct an initial index prediction model, perform multi-round iterative training on the initial index prediction model by using the training sample set, and obtain a target index prediction model. The prediction module 203 is configured to obtain a plurality of groups of tool structure data to be predicted, call the target index prediction model to respectively predict machining key indexes of each group of the tool structure data to be predicted, and obtain index data corresponding to each group of the tool structure data to be predicted. The data acquisition module 204 is configured to acquire an initial tool structure data group based on the plurality of groups of tool structure data to be predicted. The optimization module 205 is configured to perform multi-round iterative updating on the initial tool structure data group based on the plurality of groups of tool structure data to be predicted and the plurality of groups of index data by using a genetic algorithm, and when a preset iteration stop condition is met, update the tool structure data group as an optimized tool structure data group.

[0044] Based on the same idea as the method in the above embodiment, the device provided by the present application can implement the method of the above embodiment.

[0045] The tool structure data prediction method in the embodiments of the application is described above, and the tool structure data prediction device in the embodiments of the application is described below. Please refer to Figure 2 The tool structure data prediction device in the embodiments of the application is described in detail from the perspective of a modular functional entity, and the tool structure data prediction device in the embodiments of the application is described in detail from the perspective of hardware processing.

[0046] Figure 3is a structural schematic diagram of a tool structure data prediction device provided by an embodiment of the present application. The tool structure data prediction device 300 can have great differences due to different configurations or performances, and can include one or more central processing units (CPUs) 310 (for example, one or more processors) and a memory 320, one or more storage media 330 (for example, one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and the storage media 330 can be temporary storage or persistent storage. The programs stored in the storage media 330 can include one or more modules (not shown in the figure), and each module can include a series of instruction operations in the tool structure data prediction device 300. Further, the processor 310 can be configured to communicate with the storage media 330, and execute the series of instruction operations in the storage media 330 on the tool structure data prediction device 300, so as to implement the steps of the tool structure data prediction method provided by the above-mentioned method embodiments.

[0047] The tool structure data prediction device 300 can further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and the like. Those skilled in the art can understand that the tool structure data prediction device 300 can further include other components that are not shown in the figure, and the components shown in the figure do not constitute a limitation on the tool structure data prediction device. Figure 3 The tool structure data prediction device structure shown does not constitute a limitation on the tool structure data prediction device, and can include more or fewer components than shown, or combine certain components, or different arrangement of components.

[0048] The present application also provides a computer readable storage medium, which can be a non-volatile computer readable storage medium, or a volatile computer readable storage medium. The computer readable storage medium stores instructions, and when the instructions are run on a computer, the computer executes the steps of the tool structure data prediction method.

[0049] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system or device, unit can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0050] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0051] Finally, it should be noted that: the above only for the preferred examples of the present application, and not for limiting the present application, although the present application is described in detail with reference to the foregoing embodiments, for those skilled in the art, it still can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting tool structure data, characterized in that, include: Construct a data sample library for training, and perform simulation operations based on the data sample library to construct a training sample set; An initial indicator prediction model is constructed, and the initial indicator prediction model is iteratively trained multiple times using the training sample set to obtain the target indicator prediction model. Multiple sets of tool structure data to be predicted are obtained, and the target index prediction model is called to predict the key machining indexes for each set of tool structure data to be predicted, so as to obtain the index data set corresponding to each set of tool structure data to be predicted. An initial tool structure data set is obtained based on multiple sets of the tool structure data sets to be predicted. A genetic algorithm is used to iteratively update the initial tool structure data set based on multiple sets of the tool structure data set to be predicted and multiple sets of the index data set. When the preset iteration stopping condition is met, the updated tool structure data set is used as the optimized tool structure data set.

2. The tool structure data prediction method according to claim 1, characterized in that, The training sample set includes a training set and a test set, and the data sample library includes various tool machining parameters; the step of performing simulation operations based on the data sample library to construct the training sample set includes: A full permutation combination algorithm is used to perform full permutation combinations on various tool machining parameters to obtain various input working condition combinations; Simulation operations are performed based on each of the aforementioned working condition combinations to obtain a set of simulation index data corresponding to each working condition combination. By integrating various tool processing parameters and simulation index data sets corresponding to each working condition combination, a training sample set is obtained; Based on a preset partitioning ratio, the training sample set is divided into a training set and a test set.

3. The tool structure data prediction method according to claim 1, characterized in that, The initial indicator prediction model includes an initial Actor network and an initial Critic network; the step of performing multiple rounds of iterative training on the initial indicator prediction model using the training sample set to obtain the target indicator prediction model includes: An initial experience replay pool is constructed based on the training sample set, and the current state sample and the next state sample are obtained from the initial experience replay pool. The initial Actor network is used to predict the action of the current state sample to obtain the current action; Obtain a predefined reward function, and calculate the reward value for the current action based on the reward function to obtain the current reward value; The current state sample, the current action, the current reward value, and the next state sample are stored in the initial experience replay pool to obtain the expanded experience replay pool; Based on the expanded experience replay pool, the initial Actor network and the initial Critic network are iteratively updated multiple times to obtain the target index prediction model.

4. The tool structure data prediction method according to claim 3, characterized in that, The process of performing multiple rounds of iterative updates on the initial Actor network and the initial Critic network based on the expanded experience replay pool to obtain the target indicator prediction model includes: The current state sample, the current action, the current reward value, and the next state sample are obtained based on the expanded experience replay pool. The next action is obtained by using an initial Actor network to predict the action of the next state sample. Using an initial Critic network, a value evaluation process is performed based on the current state sample, the current action, the next state sample, and the next action to obtain the current state value and the next state value; The TD error is obtained by calculating the error based on the current reward value, the current state value, and the next state value. Based on the TD error, the initial Actor network and the initial Critic network are iteratively updated multiple times to obtain the target index prediction model.

5. The tool structure data prediction method according to claim 4, characterized in that, The process of performing multiple rounds of iterative updates on the initial Actor network and the initial Critic network based on the TD error to obtain the target index prediction model includes: The backpropagation algorithm is used to perform multiple rounds of iterative updates on the initial Critic network based on the TD error in order to obtain an optimized Critic network. The policy gradient algorithm is used to perform multiple rounds of iterative updates on the initial Actor network based on the TD error to obtain an optimized Actor network. The optimized Critic network and the optimized Actor network are combined to form the target indicator prediction model.

6. The tool structure data prediction method according to claim 1, characterized in that, The genetic algorithm includes a roulette wheel algorithm, a crossover algorithm, and a mutation algorithm. The genetic algorithm iteratively updates the initial tool structure data set through multiple rounds based on multiple sets of the tool structure data set to be predicted and multiple sets of the index data set. When a preset iteration stopping condition is met, the updated tool structure data set is used as the optimized tool structure data set, including: Using the aforementioned roulette wheel selection algorithm, multiple sets of the tool structure data to be predicted are selected based on multiple sets of the aforementioned index data sets, thereby obtaining multiple sets of parent tool structure data sets. The crossover operation algorithm is used to perform gene exchange processing on multiple sets of parent tool structure data sets to obtain multiple sets of cross tool structure data sets; The mutation operation algorithm is used to perform gene fine-tuning on multiple sets of the cross tool structure data sets to obtain multiple sets of mutated tool structure data sets; The initial tool structure data set is iteratively updated multiple times based on multiple sets of the aforementioned variant tool structure data sets. When the preset iteration stop condition is met, the updated tool structure data set is used as the optimized tool structure data set.

7. The tool structure data prediction method according to claim 6, characterized in that, The method employs the roulette wheel selection algorithm, performing roulette wheel selection on multiple sets of the tool structure data sets to be predicted based on multiple sets of the index data sets, to obtain multiple sets of parent tool structure data sets, including: Obtain the predefined fitness function; Based on the fitness function, the fitness of multiple sets of index data groups is calculated to obtain fitness values ​​corresponding to multiple sets of tool structure data groups to be predicted. Using the aforementioned roulette wheel selection algorithm, the roulette wheel selection probability is calculated based on multiple fitness values ​​to obtain the roulette wheel selection probability corresponding to multiple sets of the tool structure data sets to be predicted; A preset selection rule is obtained, and a roulette wheel selection process is performed on multiple sets of the tool structure data sets to be predicted based on the selection rule and multiple roulette wheel selection probabilities to obtain multiple sets of the parent tool structure data sets.

8. A tool structure data prediction device, characterized in that, include: Simulation module: used to build a data sample library for training, and to perform simulation operations based on the data sample library to build a training sample set; Training model: used to build an initial indicator prediction model, and to perform multiple rounds of iterative training on the initial indicator prediction model using the training sample set to obtain the target indicator prediction model; Prediction module: used to acquire multiple sets of tool structure data to be predicted, call the target index prediction model to predict the key machining indicators for each set of tool structure data to be predicted, and obtain the index data set corresponding to each set of tool structure data to be predicted. Data acquisition module: used to acquire an initial tool structure data set based on multiple sets of the tool structure data sets to be predicted; Optimization module: Used to perform multiple rounds of iterative updates on the initial tool structure data set based on multiple sets of the tool structure data set to be predicted and multiple sets of the index data set using a genetic algorithm. When the preset iteration stopping condition is met, the updated tool structure data set is used as the optimized tool structure data set.

9. A tool structure data prediction device, characterized in that, The tool structure data prediction device includes: a memory and at least one processor, wherein the memory stores instructions; At least one of the processors invokes the instructions in the memory to cause the tool structure data prediction device to perform the steps of the tool structure data prediction method as claimed in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the tool structure data prediction method as described in any one of claims 1-7.

Images