Production control method and device for die steel

Abstract

The invention discloses a production control method and device for die steel, and relates to the technical field of intelligent control. Performing surface roughness analysis on the initial die steel, and substituting the roughness processing model to obtain initial working parameters; polishing the initial die steel according to the initial working parameters, and then performing surface roughness analysis to obtain second target roughness; if the second target roughness is larger than a preset roughness threshold value, a correction factor is obtained according to the first target roughness, the preset roughness threshold value and the second target roughness, machining parameter constraints in the roughness processing model are updated, and the updated second target roughness is obtained for polishing treatment. The initial die steel is subjected to roughness analysis, the initial working parameters are obtained through the roughness processing model, manual intervention is reduced, when the roughness is larger than the preset threshold value, the model can adjust the machining parameters according to the correction factors, dynamic working parameter adjustment is achieved, and the production efficiency of the die steel is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent control, and particularly relates to a production control method and device for die steel. Background Art

[0002] In traditional processing, especially in the polishing stage of die steel, the control of roughness often depends on the operator's experience or preliminary parameter settings. Since the physical properties (such as hardness, composition, etc.) of each batch of die steel materials may be different, the control of roughness may not be precise enough, resulting in the surface roughness not meeting the design requirements, thereby affecting the service performance and life of the die.

[0003] With the continuous development of robot technology, robot polishing technology has gradually gained wide application in the field of die steel processing. Automated robot polishing can not only reduce errors and labor costs, but also improve workplace safety and production efficiency. With the wide popularization of robot technology, robot polishing is expected to become an important part of the die steel manufacturing and processing industry. Although robots can perform automated polishing, the adjustment of the current polishing process parameters of robots is usually based on preliminary settings and relies on manual experience. Once the process parameters are set, they often do not dynamically adjust according to the processing results, resulting in difficulty in further optimizing the roughness. And when there are problems with the polishing process parameters, it is still modified manually, making the production efficiency of die steel low. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem that the current polishing process parameters of robots are determined in advance and cannot be modified according to the actual situation. When there are problems with the polishing process parameters, it is still modified manually, resulting in low production efficiency of die steel, and to propose a production control method and device for die steel.

[0005] In the first aspect of the implementation of the present invention, a production control method for die steel is first proposed. The method includes: Obtain the initial die steel, and perform surface roughness analysis on the initial die steel to obtain the first target roughness; Substitute the first target roughness into the roughness processing model to obtain the initial working parameters; Perform polishing treatment on the initial die steel according to the initial working parameters to obtain the target die steel, and perform surface roughness analysis on the target die steel to obtain the second target roughness; If the second target roughness is greater than the preset roughness threshold, then obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; Update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as the target working parameter, and polish the die steel according to the target working parameter.

[0006] Optionally, the obtaining of the first target roughness by performing surface roughness analysis on the initial die steel includes: Measure the surface roughness of the initial die steel with a stylus roughness meter to obtain the first roughness value; Obtain the complete surface 3D topology information of the initial die steel with a white light interferometer to obtain the second roughness value; Scan the surface of the initial die steel with a digital microscope to obtain the third roughness value, and obtain the first target roughness according to the first roughness value, the second roughness value, and the third roughness value.

[0007] Optionally, substituting the first target roughness into the roughness processing model to obtain the initialization working parameters includes: Substitute the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as the initial process parameter population; Perform crossover and mutation on all process parameter combinations through a preset ratio to obtain an updated process parameter population; Calculate the roughness value and the polishing depth value of each individual in the updated process parameter population through the roughness function and the polishing depth function, divide the Pareto front levels according to the roughness values and the polishing depth values of all individuals, and calculate the crowding degree for each front level; Select the optimal individuals from the initial process parameter population and the updated process parameter population according to the Pareto front level and the crowding degree to obtain the Pareto optimal solution set, and obtain the optimal solution in the Pareto optimal solution set to obtain the initialization working parameters.

[0008] Optionally, substituting the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as the initial process parameter population includes: Obtain historical polishing data, substitute the historical polishing data into the neural network model for forward model training to obtain a model of the relationship between surface roughness and processing parameters denoted as the roughness prediction model; Substitute the first target roughness into the roughness prediction model to obtain the first processing parameter solution; Define an optimization function and set processing parameter constraints, repeatedly update the first processing parameter solution through an optimization algorithm, and add each updated first processing parameter solution to the initial process parameter population.

[0009] Optionally, the obtaining of the correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness includes: Calculate the difference between the first target roughness and the second target roughness to obtain the actual correction difference; Calculate the difference between the first target roughness and the preset roughness threshold to obtain the actual processing difference; Divide the actual correction difference by the actual processing difference to obtain the correction factor.

[0010] In the second aspect of the implementation of the present invention, a production control device for die steel is proposed, including: A roughness analysis module, configured to obtain initial die steel and perform surface roughness analysis on the initial die steel to obtain a first target roughness; An initialization working parameter determination module, configured to substitute the first target roughness into a roughness processing model to obtain initialization working parameters; A polishing processing module, configured to polish the initial die steel according to the initialization working parameters to obtain target die steel, and perform surface roughness analysis on the target die steel to obtain a second target roughness; A correction factor determination module, configured to, if the second target roughness is greater than a preset roughness threshold, obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; A parameter constraint update module, configured to update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as target working parameters, and polish the die steel according to the target working parameters.

[0011] Optionally, the roughness analysis module includes: A first roughness value determination module, configured to measure the surface roughness of the initial die steel by a stylus roughness meter to obtain a first roughness value; A second roughness value determination module, configured to obtain a second roughness value by obtaining the complete surface 3D topology information of the initial die steel by a white light interferometer; A third roughness value determination module, configured to scan the surface of the initial die steel by a digital microscope to obtain a third roughness value, and obtain the first target roughness according to the first roughness value, the second roughness value, and the third roughness value.

[0012] Optionally, the initialization working parameter determination module includes: An initial process parameter population determination module, configured to substitute the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as the initial process parameter population; A process parameter combination update module, configured to perform crossover and mutation on all process parameter combinations by a preset ratio to obtain an updated process parameter population; A crowding degree calculation module, which is used to calculate the roughness value and polishing depth value of each individual in the updated process parameter population through a roughness function and a polishing depth function, divide the Pareto front levels according to the roughness values and polishing depth values of all individuals, and calculate the crowding degree for each front level; An initialization working parameter determination module, which is used to select the optimal individuals from the initial process parameter population and the updated process parameter population according to the Pareto front levels and the crowding degree to obtain a Pareto optimal solution set, and obtain the initialization working parameters by acquiring the optimal solutions in the Pareto optimal solution set.

[0013] Optionally, the process parameter combination update module includes: A roughness prediction model generation module, which is used to obtain historical polishing data, substitute the historical polishing data into a neural network model for forward model training to obtain a model of the relationship between surface roughness and processing parameters, denoted as a roughness prediction model; A first processing parameter solution determination module, which is used to substitute the first target roughness into the roughness prediction model to obtain a first processing parameter solution; A first processing parameter solution update module, which is used to define an optimization function and set processing parameter constraints, repeatedly update the first processing parameter solution through an optimization algorithm, and add each updated first processing parameter solution to the initial process parameter population.

[0014] Optionally, the correction factor determination module includes: An actual correction difference determination module, which is used to calculate the difference between the first target roughness and the second target roughness to obtain an actual correction difference; An actual processing difference determination module, which is used to calculate the difference between the first target roughness and the preset roughness threshold to obtain an actual processing difference; A correction factor generation module, which is used to divide the actual correction difference by the actual processing difference to obtain a correction factor.

[0015] The beneficial effects of the present invention: The present invention proposes a production control method for die steel. Obtain the initial die steel, perform surface roughness analysis on the initial die steel to obtain the first target roughness; substitute the first target roughness into the roughness processing model to obtain the initial working parameters; polish the initial die steel according to the initial working parameters to obtain the target die steel, and perform surface roughness analysis on the target die steel to obtain the second target roughness; if the second target roughness is greater than the preset roughness threshold, then obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as the target working parameters, and polish the die steel according to the target working parameters. By performing roughness analysis on the initial die steel, it is possible to clearly know the initial surface roughness, and obtain the initial working parameters through the roughness processing model, reducing manual intervention. Then, perform secondary roughness analysis on the die steel polished with the initial working parameters. When the roughness is greater than the preset threshold, the model will adjust the processing parameters according to the correction factor. The automatic adjustment avoids the error of manual experience and can maintain a consistent high surface quality in different die steel materials, realizing dynamic adjustment of working parameters and improving the production efficiency of die steel. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 is a flowchart of a production control method for die steel provided by an embodiment of the present invention; Figure 2 is a schematic structural diagram of a production control device for die steel provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments.

[0019] All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0020] The embodiments of the present invention provide a production control method for die steel. Refer to Figure 1 , Figure 1 is a flowchart of a production control method for die steel provided by an embodiment of the present invention. The method includes the following steps: S101, obtain the initial die steel, and perform surface roughness analysis on the initial die steel to obtain the first target roughness; S102, Substitute the first target roughness into the roughness processing model to obtain the initial working parameters; S103, Polish the initial die steel according to the initial working parameters to obtain the target die steel, and perform surface roughness analysis on the target die steel to obtain the second target roughness; S104, If the second target roughness is greater than the preset roughness threshold, obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; S105, Update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as the target working parameters, and polish the die steel according to the target working parameters.

[0021] Based on a production control method for die steel provided by an embodiment of the present invention, by performing roughness analysis on the initial die steel, it is possible to clearly know the initial roughness of the surface, and obtain the initial working parameters through the roughness processing model, reducing manual intervention. Then, perform secondary roughness analysis on the die steel polished with the initial working parameters. When the roughness is greater than the preset threshold, the model will adjust the processing parameters according to the correction factor. The automated adjustment avoids the error of manual experience and can maintain a consistent high surface quality in different die steel materials, realizing dynamic adjustment of working parameters and improving the production efficiency of die steel.

[0022] In one implementation, by performing roughness analysis on the initial die steel, it is possible to clearly know the initial roughness of the surface, thus providing an accurate starting point for subsequent polishing treatment; after obtaining the first target roughness, use the roughness processing model for polishing treatment, and obtain the second target roughness of the target die steel. If the second target roughness does not meet the expectation, the model will correct the processing parameters according to the roughness deviation to ensure that the final roughness meets the requirements.

[0023] In one implementation, if the roughness is greater than the preset threshold, the model will adjust the processing parameters according to the correction factor. The automated adjustment avoids the error of manual experience and can maintain a consistent high surface quality in different die steel materials.

[0024] In one implementation, the initial working parameters include polishing pressure, tool rotation speed, feed speed, etc.; if the second target roughness is less than or equal to the preset roughness threshold, it means that the roughness meets the requirements at this time and no adjustment is needed.

[0025] In one implementation, if the second target roughness is greater than the preset roughness threshold for a continuous preset number of times, an alarm will be issued to notify the technician. At this time, there may be a malfunction of the polishing tool, or the types of die steel processed in batches are not unified, etc.

[0026] In one implementation, there are parameter constraints in the roughness processing model, and the types of parameter constraints are the same as those of the initialized working parameters. For example, when the initialized working parameters are polishing pressure, tool rotation speed, and feed rate, the parameter constraints are polishing pressure, tool rotation speed, and feed rate. The upper and lower limit values of polishing pressure, tool rotation speed, and feed rate will be set by technicians; specifically, updating the processing parameter constraints in the roughness processing model according to the correction factor means that the correction factor is divided by two and then added by 1 to obtain the upper limit factor and the lower limit factor. The original upper limit value is multiplied by the upper limit factor to obtain the new upper limit value, and the original lower limit value is multiplied by the lower limit factor to obtain the new lower limit value. For example, if the original upper limit of polishing pressure is 50 and the lower limit is 80, and the correction factor is 0.2, then the upper limit factor is 1.1 and the lower limit factor is 1.1. The new upper limit value is 55 and the new lower limit value is 88.

[0027] In one implementation, by continuously optimizing parameters and adjusting roughness, it is possible to ensure that the surface quality of the die steel for each machining is close to the target value, thereby improving the consistency of the product and reducing manual intervention and errors; by precisely controlling the parameters during the polishing process, it is possible to reduce the waste of unqualified die steel and reduce the waste of materials and time, thereby improving the overall production efficiency and cost-effectiveness.

[0028] In one embodiment, obtaining the first target roughness by analyzing the surface roughness of the initial die steel includes: Measuring the surface roughness of the initial die steel with a stylus roughness instrument to obtain the first roughness value; Obtaining the complete surface 3D topology information of the initial die steel with a white light interferometer to obtain the second roughness value; Scanning the surface of the initial die steel with a digital microscope to obtain the third roughness value, and obtaining the first target roughness based on the first roughness value, the second roughness value, and the third roughness value.

[0029] In one implementation, each measuring tool provides different types of roughness data. The stylus roughness instrument mainly provides the classical parameters of surface roughness and is suitable for measuring the surface microstructure. The white light interferometer can provide the surface 3D topology information, which helps to capture the subtle changes in the surface topography, especially under high-precision requirements. This multi-dimensional measurement method can more comprehensively and accurately describe the surface characteristics of the die steel.

[0030] In one implementation, the measurement method of stylus roughness measurement is to select 5 regions (upper, middle, lower, edge, center), measure 3 times in each region and take the average value, calculate the Ra value, and determine the overall surface roughness; the measurement method of the white light interferometer (obtaining the 3D topology information of the complete surface) is to select 3 different regions (rough region, smooth region, boundary region), collect 3D height data, analyze the surface topography (peak-valley structure), calculate the roughness parameters (Sa, Sz, Sq) and establish a roughness map; the measurement method of the digital microscope is to measure the polishing depth at 3 different positions perpendicular to the second scanning path, calculate the thickness of the RZ (remelted zone) + HAZ (heat affected zone) layer, define the final polishing depth, and take the average value through 3 measurements to reduce errors.

[0031] In one implementation, the first roughness value, the second roughness value, and the third roughness value are adjusted to a unified range through standardization, and then averaged to obtain the first target roughness value.

[0032] In one embodiment, substituting the first target roughness into the roughness processing model to obtain the initialized working parameters includes: Substituting the first target roughness into the roughness processing model to obtain a target number of process parameter combinations, denoted as the initial process parameter population; Performing crossover and mutation on all process parameter combinations through a preset ratio to obtain an updated process parameter population; Calculating the roughness value and the polishing depth value of each individual in the updated process parameter population through the roughness function and the polishing depth function, dividing the Pareto front rank according to the roughness values and the polishing depth values of all individuals, and calculating the crowding degree for each front rank; Selecting the optimal individuals from the initial process parameter population and the updated process parameter population according to the Pareto front rank and the crowding degree to obtain the Pareto optimal solution set, and obtaining the optimal solution in the Pareto optimal solution set to obtain the initialized working parameters.

[0033] In one implementation, by simultaneously optimizing the two objectives of the roughness value and the polishing depth value, an ideal balance can be found between the two. Different polishing processes may result in different roughnesses and depths. Through the division of the Pareto front, a reasonable compromise can be made between different objectives to ensure the best processing effect.

[0034] In one implementation, through the crossover and mutation processes of the genetic algorithm, various potential combinations of process parameters can be explored, rather than being limited to the initially set parameter range. This makes the optimization process less likely to fall into local optimal solutions and has strong global search capabilities; the preset ratio is used to adjust crossover and mutation, and the exploration range and depth can be flexibly adjusted according to the actual situation to adapt to die steels of different materials and different processing requirements. The preset ratio is determined by technicians.

[0035] In one implementation, the types of process parameter combinations are the same as those of the initialized working parameters; when the process parameter combinations are polishing pressure, tool rotation speed, and feed rate, the roughness function and the polishing depth function are obtained by selecting experimental points in the parameter space (polishing pressure, tool rotation speed, and feed rate) through Box - Behnken, measuring the surface roughness under each parameter combination through laser polishing experiments, and then fitting with a second - order response surface model (RSM); the number of objectives is determined by technicians.

[0036] In one implementation, for the individuals in each front rank, they are sorted in ascending order according to the roughness value and the polishing depth value respectively. The crowding degrees of the solutions at both ends after sorting (the minimum and maximum values) are set to infinity (retaining the extreme solutions), and the crowding degree of the middle solutions is calculated as: for the j - th objective of the i - th solution, through the formula is obtained, and the total crowding degree of each solution is calculated as: through the sum of the distances of each objective is obtained, where when j = 1, Y1 is the roughness value, when j = 2, Y2 is the polishing depth value, and i is the number of solutions; if roughness is given priority, select the solution with the smallest Y1 in the first front rank; if polishing depth is given priority, select the solution with the smallest Y2 in the first front rank.

[0037] In one implementation, the calculation of Pareto front ranks and crowding degrees can help select those process parameter combinations that perform excellently in multiple objectives. Individuals with lower crowding degrees are usually the "core" of optimization, meaning that these solutions are relatively "unique" in the objective space, more representative and effective. Through this method, process combinations that perform well in each objective can be systematically selected, rather than just optimizing a single objective, ensuring the comprehensive optimization of process parameters.

[0038] In one implementation, traditional process parameter optimization often relies on experience and manual adjustment, with great uncertainty and subjectivity. Genetic algorithms and Pareto optimization can reduce human intervention and rely more on data - driven and algorithm analysis, thereby improving the objectivity and scientific nature of the optimization process. Through automated algorithm optimization, work efficiency can be improved, and the time for manual adjustment and trial - and - error can be reduced.

[0039] In one implementation, by precisely calculating the roughness values and polishing depth values of each process parameter combination and relying on the selection mechanism of the Pareto front, it is possible to ensure that the obtained process parameter combinations have high accuracy and consistency in practical applications.

[0040] In one embodiment, substituting the first target roughness into the roughness processing model to obtain a target number of process parameter combinations, denoted as the initial process parameter population, including: Obtain historical polishing data, substitute the historical polishing data into the neural network model for forward model training to obtain a model of the relationship between surface roughness and processing parameters, denoted as the roughness prediction model; Substitute the first target roughness into the roughness prediction model to obtain the first processing parameter solution; Define an optimization function and set processing parameter constraints, repeatedly update the first processing parameter solution through an optimization algorithm, and add each updated first processing parameter solution to the initial process parameter population.

[0041] In one implementation, by using historical data to train the neural network model, it is possible to capture the complex relationship between surface roughness and processing parameters. This data-driven method can provide more accurate prediction results than traditional empirical rules; the neural network can effectively model complex non-linear relationships and optimize through continuous learning, so as to more accurately predict the surface roughness that may occur under given processing conditions. This helps to reduce the trial-and-error process and improve production efficiency.

[0042] In one implementation, the historical polishing data records different surface roughnesses and their corresponding optimal working process parameters in historical data. The input layer of the neural network model is determined by the types of initialized working parameters. Assuming the processing parameters are polishing pressure, tool rotation speed, and feed rate, the neural network model has an input layer with 3 neurons, corresponding to the 3 processing parameters respectively; the hidden layer has 9 layers, with 30 units in each layer; the ReLU activation function is used to avoid the problem of gradient disappearance; the output layer has only 1 neuron, representing the predicted surface roughness; the mean squared error (MSE) is used as the loss function, and the goal is to minimize the gap between the predicted surface roughness and the actual surface roughness; the Adam optimizer is used as the optimizer because it has good performance and convergence speed in most cases. When training the neural network, we use the training dataset to update the network parameters (weights and biases). The following are the training steps: Data input: Use the training data as the input and the roughness as the target output; the neural network calculates the predicted surface roughness based on the input parameters; calculate the loss value according to the mean squared error (MSE) to measure the gap between the predicted value and the true value; calculate the gradient of each neuron through the backpropagation algorithm and adjust the weights and biases to reduce the loss; repeat this process until the loss function converges, that is, until the set maximum number of iterations is reached or the loss value is small enough.

[0043] In one implementation, the goal of the optimization function is to minimize the error according to the set target roughness, that is, to minimize the error between the predicted and the target roughness. The ultimate goal of the optimization is to find a combination of parameters that makes the surface roughness as close as possible to the target value; in addition to minimizing the error, other optimization goals such as minimum processing time and minimum energy consumption can also be added.

[0044] In one implementation, the processing parameter constraints are determined by technicians, and the types of processing parameters are the same as the types of initialized working parameters; the purpose of selecting the optimization algorithm is to find the optimal combination of working parameters in the processing parameter space, and methods such as particle swarm optimization, differential evolution, and Bayesian optimization can be used.

[0045] In one implementation, substituting the target roughness value into the trained prediction model can obtain a preliminary solution of the processing parameters as the starting point of the optimization, which can ensure that the optimization goal conforms to the actual production requirements and avoid the possible errors in the traditional empirical methods; continuously updating the first solution of the processing parameters through the optimization algorithm can make the processing parameters gradually approach the optimal value. This not only improves the stability of the processing process but also ensures that the final obtained process parameters can meet the expected surface quality requirements.

[0046] In one implementation, an optimization algorithm can be used to automatically update the machining parameters and adjust the process according to the optimization results, reducing the need for manual intervention. Compared with traditional manual debugging, machine learning and optimization algorithms can find the best parameter combination more efficiently, significantly reducing the trial-and-error time and cost. The set optimization function and machining parameter constraints can ensure that the obtained machining parameters not only meet the requirements of surface roughness but also can satisfy other machining conditions or production environment requirements.

[0047] In one implementation, by using a neural network prediction model, the relationship between machining parameters and surface roughness can be accurately predicted before each machining, so as to precisely control the final surface quality. This helps to maintain consistent quality standards in mass production and reduce the occurrence of defective products.

[0048] In one embodiment, obtaining the correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness includes: Calculating the difference between the first target roughness and the second target roughness to obtain the actual correction difference; Calculating the difference between the first target roughness and the preset roughness threshold to obtain the actual processing difference; Dividing the actual correction difference by the actual processing difference to obtain the correction factor.

[0049] In one implementation, the correction factor can quantify the deviation between the actual roughness and the target roughness and calculate its relative importance by comparing with the ideal roughness threshold. This can help manufacturers precisely adjust the polishing process to ensure that the roughness of the final product is close to the target value.

[0050] In one implementation, by calculating the actual correction difference and the actual processing difference, the correction factor can find the optimization direction in the polishing process more quickly, avoid unnecessary repeated tests, and can greatly improve production efficiency and shorten the production cycle.

[0051] Based on the same inventive concept, the embodiment of the present invention also provides a production control device for die steel. Refer to Figure 2 , Figure 2 which is a schematic structural diagram of a production control device for die steel provided by the embodiment of the present invention, including: A roughness analysis module, configured to obtain the initial die steel and perform surface roughness analysis on the initial die steel to obtain the first target roughness; An initialization working parameter determination module, configured to substitute the first target roughness into the roughness processing model to obtain the initialization working parameters; A polishing processing module, configured to polish the initial die steel according to the initialization working parameters to obtain the target die steel, and perform surface roughness analysis on the target die steel to obtain the second target roughness; A correction factor determination module, configured to, if the second target roughness is greater than a preset roughness threshold, obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; A parameter constraint update module, configured to update the machining parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness and record it as the target working parameter, and polish the die steel according to the target working parameter.

[0052] Based on a production control device for die steel provided by an embodiment of the present invention, by performing roughness analysis on the initial die steel, it is possible to clearly know the initial surface roughness, and obtain the initial working parameters through the roughness processing model, reducing manual intervention. Then, perform secondary roughness analysis on the die steel polished with the initial working parameters. When the roughness is greater than the preset threshold, the model will adjust the machining parameters according to the correction factor. The automatic adjustment avoids the error of manual experience and can maintain a consistent high surface quality in different die steel materials, realizing dynamic adjustment of working parameters and improving the production efficiency of die steel.

[0053] In one embodiment, the roughness analysis module includes: A first roughness value determination module, configured to measure the surface roughness of the initial die steel with a stylus roughness meter to obtain a first roughness value; A second roughness value determination module, configured to obtain a second roughness value by acquiring the complete surface 3D topology information of the initial die steel with a white light interferometer; A third roughness value determination module, configured to scan the surface of the initial die steel with a digital microscope to obtain a third roughness value, and obtain the first target roughness according to the first roughness value, the second roughness value, and the third roughness value.

[0054] In one embodiment, the initial working parameter determination module includes: An initial process parameter population determination module, configured to substitute the first target roughness into the roughness processing model to obtain a target number of process parameter combinations, which is recorded as the initial process parameter population; A process parameter combination update module, configured to perform crossover and mutation on all process parameter combinations by a preset ratio to obtain an updated process parameter population; A crowding degree calculation module, configured to calculate the roughness value and the polishing depth value of each individual in the updated process parameter population through a roughness function and a polishing depth function, divide the Pareto front levels according to the roughness values and polishing depth values of all individuals, and calculate the crowding degree for each front level; An initialization working parameter determination module, configured to select optimal individuals from an initial process parameter population and an updated process parameter population according to the Pareto front rank and crowding degree to obtain a Pareto optimal solution set, and obtain an initialization working parameter by acquiring the optimal solution in the Pareto optimal solution set.

[0055] In one embodiment, the process parameter combination update module includes: A roughness prediction model generation module, configured to acquire historical polishing data, substitute the historical polishing data into a neural network model for forward model training to obtain a model of the relationship between surface roughness and machining parameters, denoted as a roughness prediction model; A first machining parameter solution determination module, configured to substitute a first target roughness into the roughness prediction model to obtain a first machining parameter solution; A first machining parameter solution update module, configured to define an optimization function and set machining parameter constraints, repeatedly update the first machining parameter solution through an optimization algorithm, and add each updated first machining parameter solution to the initial process parameter population.

[0056] In one embodiment, the correction factor determination module includes: An actual correction difference determination module, configured to calculate the difference between the first target roughness and the second target roughness to obtain an actual correction difference; An actual processing difference determination module, configured to calculate the difference between the first target roughness and a preset roughness threshold to obtain an actual processing difference; A correction factor generation module, configured to obtain a correction factor by dividing the actual correction difference by the actual processing difference.

[0057] The above has described in detail one embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered as limiting the implementation scope of the present invention. All equivalent changes and improvements made according to the scope of the present invention application should still fall within the patent coverage scope of the present invention.

Claims

1. A production control method for die steel, characterized in that, The method includes: Obtain the initial die steel, and perform surface roughness analysis on the initial die steel to obtain the first target roughness; Substitute the first target roughness into the roughness processing model to obtain the initialized working parameters; Perform polishing treatment on the initial die steel according to the initialized working parameters to obtain the target die steel, and perform surface roughness analysis on the target die steel to obtain the second target roughness; If the second target roughness is greater than the preset roughness threshold, obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; Update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as the target working parameters, and perform polishing treatment on the die steel according to the target working parameters.

2. The production control method for die steel according to claim 1, wherein Performing surface roughness analysis on the initial die steel to obtain the first target roughness includes: Measure the surface roughness of the initial die steel with a stylus roughness instrument to obtain the first roughness value; Obtain the complete surface 3D topology information of the initial die steel with a white light interferometer to obtain the second roughness value; Scan the surface of the initial die steel with a digital microscope to obtain the third roughness value, and obtain the first target roughness according to the first roughness value, the second roughness value, and the third roughness value.

3. A production control method for die steel according to claim 1, characterized in that, Substituting the first target roughness into the roughness processing model to obtain the initialized working parameters includes: Substitute the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as the initial process parameter population; Perform crossover and mutation on all process parameter combinations through a preset ratio to obtain an updated process parameter population; Calculate the roughness value and the polishing depth value of each individual in the updated process parameter population through the roughness function and the polishing depth function, divide the Pareto front levels according to the roughness values and the polishing depth values of all individuals, and calculate the crowding degree for each front level; Select the optimal individuals from the initial process parameter population and the updated process parameter population according to the Pareto front level and the crowding degree to obtain the Pareto optimal solution set, and obtain the optimal solution in the Pareto optimal solution set to obtain the initialized working parameters.

4. A production control method for die steel according to claim 3, characterized in that, Substituting the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as the initial process parameter population includes: Obtain historical polishing data, and substitute the historical polishing data into the neural network model for forward model training to obtain a model of the relationship between surface roughness and processing parameters denoted as the roughness prediction model; Substitute the first target roughness into the roughness prediction model to obtain the first processing parameter solution; Define an optimization function and set processing parameter constraints, repeatedly update the first processing parameter solution through an optimization algorithm, and add each updated first processing parameter solution to the initial process parameter population.

5. A production control method for die steel according to claim 1, characterized in that, Obtaining a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness includes: Calculate the difference between the first target roughness and the second target roughness to obtain the actual correction difference; Calculate the difference between the first target roughness and the preset roughness threshold to obtain the actual processing difference; Obtain a correction factor by dividing the actual correction difference by the actual processing difference.

6. A production control device for die steel, characterized in that, The device includes: A roughness analysis module, configured to obtain an initial die steel, and perform surface roughness analysis on the initial die steel to obtain a first target roughness; An initialization working parameter determination module, configured to substitute the first target roughness into a roughness processing model to obtain initialization working parameters; A polishing processing module, configured to polish the initial die steel according to the initialization working parameters to obtain a target die steel, and perform surface roughness analysis on the target die steel to obtain a second target roughness; A correction factor determination module, configured to, if the second target roughness is greater than a preset roughness threshold, obtain a correction factor according to the first target roughness, the preset roughness threshold, and the second target roughness; A parameter constraint update module, configured to update the processing parameter constraints in the roughness processing model according to the correction factor, obtain the updated second target roughness denoted as target working parameters, and polish the die steel according to the target working parameters.

7. The production control device for die steel according to claim 6, characterized in that The roughness analysis module includes: A first roughness value determination module, configured to measure the surface roughness of the initial die steel by a stylus roughness instrument to obtain a first roughness value; A second roughness value determination module, configured to obtain a complete surface 3D topology information of the initial die steel by a white light interferometer to obtain a second roughness value; A third roughness value determination module, configured to scan the surface of the initial die steel by a digital microscope to obtain a third roughness value, and obtain a first target roughness according to the first roughness value, the second roughness value, and the third roughness value.

8. The production control device for die steel according to claim 6, characterized in that, The initialization working parameter determination module includes: An initial process parameter population determination module, configured to substitute the first target roughness into the roughness processing model to obtain a target number of process parameter combinations denoted as an initial process parameter population; A process parameter combination update module, configured to perform crossover and mutation on all process parameter combinations by a preset ratio to obtain an updated process parameter population; A crowding degree calculation module, configured to calculate the roughness value and the polishing depth value of each individual in the updated process parameter population through a roughness function and a polishing depth function, divide the Pareto front rank according to the roughness values and the polishing depth values of all individuals, and calculate the crowding degree for each front rank; An initialization working parameter determination module, configured to select the optimal individuals from the initial process parameter population and the updated process parameter population according to the Pareto front rank and the crowding degree to obtain a Pareto optimal solution set, and obtain the optimal solution in the Pareto optimal solution set to obtain initialization working parameters.

9. The production control device for die steel according to claim 8, characterized in that The process parameter combination update module includes: A roughness prediction model generation module, configured to obtain historical polishing data, and substitute the historical polishing data into a neural network model for forward model training to obtain a model of the relationship between surface roughness and processing parameters denoted as a roughness prediction model; A first processing parameter solution determination module, configured to substitute the first target roughness into the roughness prediction model to obtain a first processing parameter solution; The first machining parameter solution update module is used to define an optimization function and set machining parameter constraints, repeatedly update the first machining parameter solution through an optimization algorithm, and add the first machining parameter solution updated each time to the initial process parameter population.

10. The production control device for die steel according to claim 6, wherein, The correction factor determination module includes: The actual correction difference determination module is used to calculate the difference between the first target roughness and the second target roughness to obtain the actual correction difference; The actual processing difference determination module is used to calculate the difference between the first target roughness and the preset roughness threshold to obtain the actual processing difference; The correction factor generation module is used to obtain the correction factor by dividing the actual correction difference by the actual processing difference.

Images