Square drill cutting process optimization method

Abstract

The invention discloses a square drill cutting process optimization method, and relates to the technical field of gemstone processing and cutting processes, and the method comprises the following steps: collecting raw gemstone data and cutting process parameters by using a high-resolution 3D scanning and sensor technology, predicting an optimal cutting proportion and finished product quality in combination with a gradient lifting decision tree model, and obtaining a gemstone cutting process. The precision is verified through a cross validation technology, the improvement effect is tested through experiments, an actual result optimization model is fed back, market information is updated regularly, the result is displayed through visual software, and efficient and accurate optimization of the square drill cutting process is achieved. According to the square drill cutting process optimization method, the cutting precision and the production efficiency are remarkably improved, the production cost is reduced, meanwhile, the personalized customization capacity of products is enhanced, intelligent adjustment of the cutting scheme is achieved through real-time data collection and model prediction, the consistency of the quality of finished products and the market adaptability are ensured, and the method is suitable for popularization and application. And revolutionary technical breakthrough is brought to the jewel processing industry.

Description

Technical Field

[0001] The invention relates to the technical field of gemstone processing and cutting technology, and in particular to a method for optimizing the cutting process of square diamonds. Background Art

[0002] Square diamond cutting is a widely used technique in gem processing, specifically the process of cutting raw gemstones into square or other rectangular shapes. This process not only gives gemstones a more unique and attractive appearance, but also significantly increases their value and market appeal. In the square diamond cutting process, the size and shape of the gemstones need to be accurately measured and calculated first to ensure that the final cutting effect meets the design requirements. Then, professional diamond cutters are used to cut the gemstones into square or rectangular shapes, a process that requires extremely high precision and skills. After cutting, the surface of the gemstone may be rough, and at this time it is necessary to use grinders and polishers to cut and polish it to improve its appearance and texture, making it smoother and more transparent. Square diamond cutting is not only suitable for diamonds, but can also be used for other gemstones as well as materials such as glass and ceramics. However, this process requires a high degree of technology and experience, so it is usually completed by professional processing personnel and equipment. In general, the square diamond cutting process is a complex and delicate process that can create square gemstones with unique beauty and market value through professional cutting, grinding and polishing processes. This process has broad application prospects in the fields of gem processing and jewelry making.

[0003] In order to solve the bottleneck problem of square diamond cutting process in improving product quality and production efficiency, the existing technology adopts the method of combining traditional mechanical cutting with manual polishing. However, this method still has problems such as insufficient cutting accuracy and uneven polishing effect in actual operation, which leads to inconsistent brilliance of finished diamonds, affecting the overall appearance and market value. In addition, the traditional process has high skill requirements for operators and a long cutting cycle, which makes it difficult to meet the growing demand for personalized customization and the rapidly changing market rhythm. In view of these problems, a square diamond cutting process optimization method is proposed. Summary of the invention

[0004] The present invention aims to provide a method for optimizing the cutting process of square diamonds to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for optimizing the cutting process of square diamonds, comprising the following steps: S1. Collect the three-dimensional dimensions, crystal structure and internal defect data of the original stone through a high-resolution 3D scanner; S2, collecting temperature data and applied pressure data during the cutting process through a temperature sensor and a pressure sensor; S3, extract the hardness distribution and color grade of the rough stone as the key variables for predicting the cutting effect, use the optimal cutting path angle calculated according to the crystal direction as the derived variable, build a gradient boosting decision tree model, and output the optimal cutting ratio parameters and expected finished product quality; S4. Use cross-validation technology and test sets to verify the prediction accuracy of the model, and conduct comparative experiments to check whether product quality and production efficiency have been significantly improved; S5. Feedback the actual results of cutting to the system, adjust the model parameters, and regularly update the market preference information; S6. Use visual report generation software to display optimized cutting results.

[0006] A further improvement of the technical solution of the present invention is that in S1, the process of collecting the three-dimensional dimensions, crystal structure and internal defect data of the rough stone by means of a high-resolution 3D scanner comprises: A high-resolution 3D scanner is deployed on the workbench, and the rough stone to be cut is placed in the scanning area. The high-resolution 3D scanner is based on optical principles and uses a white light source to measure the surface of the object point by point. The scanning light emitted by the white light source is irradiated on the surface of the rough stone and then reflected back to the receiver. The spatial coordinates of each point on the surface of the rough stone are calculated based on the time difference of the reflected light, and then a three-dimensional model of the rough stone is constructed. The transmission imaging technology is used to allow the light to penetrate the rough stone, capture the image information of the internal features of the rough stone, and then obtain the crystal structure and internal defect data; Preprocessing the collected data on the three-dimensional dimensions, crystal structure and internal defects of the rough stone, wherein the preprocessing operation includes removing abnormal values ​​and outliers in the data, performing smoothing to reduce the interference caused by irregularities on the surface of the rough stone, and splicing and fusing the scanning results at different viewing angles to form a three-dimensional data set; The shape contour, crystal texture and defect location are extracted from the preprocessed three-dimensional data as the key features for optimizing the square diamond cutting process. Among them, the shape contour is used to determine the boundary range and geometric shape of the rough stone, the crystal texture is used to identify the crystallization direction and arrangement inside the rough stone, and the defect location is used to locate the specific position and distribution of cracks and inclusions inside the rough stone.

[0007] A further improvement of the technical solution of the present invention is that in S2, the process of collecting temperature data and applying pressure data during the cutting process by means of a temperature sensor and a pressure sensor includes: A temperature sensor is deployed on the workbench near the cutting area, and a pressure sensor is deployed between the contact surface of the cutting tool and the rough stone. When the temperature changes, the physical properties of the material inside the temperature sensor change, and the physical property change is converted into an electrical signal output. The actual temperature reading is obtained by calibrating the value. When subjected to external force, the pressure sensor generates a voltage change, which is converted into an electrical signal and converted into a specific pressure value after calibration. Preprocessing the collected temperature data and pressure data includes identifying and eliminating data points that are obviously deviated from the normal range, applying a low-pass filter to smooth the electrical signal curve, and performing time synchronization and data integration operations; Temperature fluctuation, pressure distribution and dynamic response are extracted from the pre-processed temperature and pressure data as key features for optimizing the square diamond cutting process. The temperature fluctuation feature is used to analyze the temperature change trend during the cutting process, identify the key moments of temperature rise and fall, and whether there is abnormally high temperature. The pressure distribution feature is used to calculate the maximum, minimum and average values ​​of the applied pressure during the cutting process, and evaluate whether the pressure is evenly distributed and whether there is any mutation. The dynamic response feature is used to explore the relationship between temperature and pressure and the relationship between the two over time, and to assist in understanding the temperature rise caused by frictional heat generation during the cutting process and the influence of applied pressure adjustment on the cutting effect.

[0008] A further improvement of the technical solution of the present invention is that in S3, the process of extracting the hardness distribution and color grade of the rough stone as key variables for predicting the cutting effect includes: A hardness tester is introduced to measure the point hardness at different positions of the rough stone to obtain the hardness value, and the hardness value is mapped to the coordinate position corresponding to the three-dimensional model to form a hardness distribution map covering the entire rough stone surface and part of the internal structure. The hardness distribution map is statistically analyzed to identify whether there is a clear boundary between the hard area and the soft area; The color values ​​of the pixels are extracted from the color images obtained during the 3D scanning process. According to the color grading system of diamond industry standards, the collected color values ​​are classified to determine the color grade of the rough stone. The color grade information is integrated into the 3D model to form a spatial distribution map of the color grade.

[0009] A further improvement of the technical solution of the present invention is that in S3, the process of calculating the optimal cutting path angle according to the crystal direction and using it as a derived variable includes: The crystal texture information extracted from the pre-processed 3D data is used to determine the main arrangement direction of the crystals inside the original stone. The geometric algorithm is used to calculate the effects of light reflection and refraction at different cutting angles, and the angle that maximizes the display of brightness and fire is selected as the optimal cutting path angle. The calculation process is as follows: ; in, represents the optimal cutting angle, Indicates the possible cutting angles, Represents the reflectivity function at a given angle, reflecting the light at that angle Reflection efficiency under angle, Represents the transmittance function at a given angle, reflecting the light at that angle Transmission efficiency under angle.

[0010] A further improvement of the technical solution of the present invention is that: in S3, the process of constructing a gradient boosting decision tree model and outputting the optimal cutting ratio parameters and the expected finished product quality includes: The pre-processed shape contour, crystal texture, defect location, temperature fluctuation, pressure distribution, dynamic response, hardness distribution, color grade and optimal cutting path angle data were used as training sets and validation sets for building the gradient boosting decision tree model, where 80% of the data was used as the training set and 20% of the data was used as the validation set; The average value of the sample is used as the initial prediction value to construct an iterative process, which includes calculating the residual between the initial prediction value and the actual value of the current model, reducing the residual as the target variable to train a new decision tree, multiplying the prediction result of the new decision tree by the learning rate and adding it to the existing model. In each round of iteration, the importance of different features for prediction is evaluated by analyzing the features used when the decision tree node is split and their contribution, and the mean square error is used as the loss function. The calculation process is as follows: ; in, represents the loss value, represents the number of samples, represents the cutting ratio and finished product quality of the i-th sample, Represents the model's predicted value for the i-th sample, and stops iterating when the loss no longer decreases significantly; The gradient boosting decision tree model outputs optimal cutting proportion parameters and expected finished product quality based on the input new rough stone data. The optimal cutting proportion parameters include crown angle and pavilion depth ratio, and the expected finished product quality includes brightness, fire and weight retention rate.

[0011] A further improvement of the technical solution of the present invention is that: in S4, the process of verifying the prediction accuracy of the model by using the cross-validation technology and the test set includes: The training set is divided into 10 subsets, each subset is used as a validation set in turn, and the remaining 9 subsets are combined as new training sets. The gradient boosting decision tree model is trained with each subset as a validation set, and the mean square error performance index of the gradient boosting decision tree model on the validation set is recorded. The average performance index of 10 validations is calculated as the final evaluation result of the gradient boosting decision tree model, and the number and learning rate of decision trees in the gradient boosting decision tree model are adjusted. According to the final evaluation result of cross-validation, the number of decision trees and learning rate parameters that minimize the loss are selected as the final gradient boosting decision tree model configuration; A data set that has not participated in model training and validation is used as a test set. The feature information of the original stone is input into the gradient boosting decision tree model, and the optimal cutting ratio parameters and expected finished product quality are output. Actual cutting is performed according to the optimal cutting ratio parameters and the expected finished product quality, and the quality characteristics of the finished product are measured. The prediction accuracy of the model prediction is evaluated by comparing the prediction results of the gradient boosting decision tree model with the quality of the finished product after actual cutting.

[0012] A further improvement of the technical solution of the present invention is that in S4, the process of checking whether the product quality and production efficiency are significantly improved through comparative experiments includes: A batch of rough stones with similar characteristics in shape, size, crystal structure and hardness distribution were selected as experimental objects. The experimental group in the comparative experiment used the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model for cutting operations, and the control group performed cutting operations according to the traditional empirical method, and recorded relevant data before and after cutting, including three-dimensional dimensions, crystal texture, defect location, hardness distribution and color grade information before cutting, temperature fluctuations, pressure distribution and dynamic response key features during the cutting process, quality characteristics of the finished product after cutting, and production time and degree of wear of the cutting equipment; After cutting, calculate the mean and standard deviation of the brightness, fire and weight retention of the finished products of the experimental group and the control group. Use the t-test to determine whether there is a statistically significant difference between the experimental group and the control group. Compare the average production time and tool wear of the experimental group and the control group, and calculate the proportion of production cost savings to quantify the improvement in production efficiency. According to the data analysis results, determine whether the gradient boosting decision tree model can improve product quality and production efficiency. If the finished product of the experimental group is significantly better than that of the control group in terms of brightness, fire color, weight retention rate, etc., and the production time and tool wear are also reduced, it means that the gradient boosting decision tree model can improve product quality and production efficiency. If there is no obvious difference and there is a decline in indicators, adjust the number and learning rate of decision trees of the gradient boosting decision tree model.

[0013] A further improvement of the technical solution of the present invention is that in S5, the process of feeding back the actual result of cutting to the system, adjusting the model parameters, and regularly updating the market preference information includes: After each cutting operation is completed, the quality characteristics of the actual cut product, production time and tool wear are recorded and sorted out, and the actual cutting results are compared with the predicted values ​​of the gradient boosting decision tree model to form new training samples, and the newly collected data are supplemented to the original training set; Use the updated training set to train the gradient boosting decision tree model again. According to the results of cross-validation, adjust the number of decision trees and learning rate of the gradient boosting decision tree model to minimize the prediction error. Calculate the mean square error performance index of the updated gradient boosting decision tree model on the test set and compare it with the historical gradient boosting decision tree model to confirm the improvement effect. Conduct market research regularly to analyze changes in consumers’ requirements for the shape, proportion, color grade, and optical performance of finished diamonds. Convert the results of market research into quantifiable preference indicators and incorporate them into the feature space of the gradient boosting decision tree model. Combined with new market preference information, iteratively update the gradient boosting decision tree model to ensure that the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model meet the mainstream demand of the current market.

[0014] A further improvement of the technical solution of the present invention is that: in S6, the process of displaying the optimized cutting results by visual report generation software includes: Use bar graphs to compare the differences in brightness, fire and weight retention between the experimental and control groups. Use line graphs to show the temperature fluctuations, pressure distribution and production efficiency trends over time. Use scatter graphs to show the relationship between cutting angles and the optical properties of finished diamonds. Use heat maps to represent the spatial distribution of hardness distribution and color grade. Integrate bar graphs, line graphs, scatter graphs and heat maps into interactive dashboards, allowing users to filter different data dimensions according to their needs, add annotations and descriptions next to the charts, and use visual report generation software to automatically generate a comprehensive optimized cutting results report.

[0015] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art: The present invention provides a square diamond cutting process optimization method, which can significantly improve the cutting accuracy. The raw stone data is accurately collected by a high-resolution 3D scanner, and the optimal cutting ratio is predicted by combining a gradient boosting decision tree model, ensuring that each cutting can maximize the natural beauty of the raw stone, while reducing defects and improving the quality of the finished product.

[0016] The present invention provides a square diamond cutting process optimization method, which can optimize production efficiency, monitor temperature and pressure data in the cutting process in real time, adjust cutting parameters in time, avoid over-cutting or damage, shorten the production cycle, and improve overall production efficiency.

[0017] The present invention provides a square diamond cutting process optimization method, which can reduce production costs, reduce waste generation, and improve raw material utilization by accurately predicting cutting effects, thereby effectively reducing production costs and increasing corporate profits.

[0018] The present invention provides a square diamond cutting process optimization method, which can improve the product customization capability, flexibly adjust the cutting plan according to market preference information and specific customer needs, meet diversified and personalized market needs, and enhance market competitiveness.

[0019] The present invention provides a square diamond cutting process optimization method, which can enhance decision support. Through visual report generation software, the cutting results are intuitively displayed, providing data support for management, and facilitating the formulation of more scientific and reasonable production plans and market strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0021] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] Examples, such as Figure 1 As shown, the present invention provides a method for optimizing the cutting process of square diamonds, comprising the following steps: S1. Use a high-resolution 3D scanner to collect the three-dimensional dimensions, crystal structure and internal defect data of the raw stone. Deploy a high-resolution 3D scanner on the workbench and place the raw stone to be cut in the scanning area. The high-resolution 3D scanner is based on optical principles and uses a white light source to measure the surface of the object point by point. The scanning light emitted by the white light source is irradiated on the surface of the raw stone and then reflected back to the receiver. The spatial coordinates of each point on the surface of the raw stone are calculated based on the time difference of the reflected light, and then a three-dimensional model of the raw stone is constructed. The transmission imaging technology is used to allow the light to penetrate the raw stone, capture the image information of the internal features of the raw stone, and then obtain the crystal structure and internal defect data. The three-dimensional size, crystal structure and internal defect data of the rough stone are preprocessed, and the preprocessing operation includes removing outliers and outliers in the data, performing smoothing to reduce the interference caused by the irregularities of the rough stone surface, splicing and fusing the scanning results under different viewing angles to form a three-dimensional data set, and extracting the shape contour, crystal texture and defect location from the preprocessed three-dimensional data as key features for optimizing the square diamond cutting process, wherein the shape contour is used to determine the boundary range and geometric shape of the rough stone, the crystal texture is used to identify the crystallization direction and arrangement inside the rough stone, and the defect location is used to locate the specific position and distribution of the cracks and inclusions inside the rough stone.

[0024] S2. The temperature data and applied pressure data of the cutting process are collected through temperature sensors and pressure sensors. The temperature sensor is deployed on the workbench near the cutting area, and the pressure sensor is deployed between the contact surface of the cutting tool and the original stone. When the temperature changes, the physical properties of the material inside the temperature sensor change. The physical property change is converted into an electrical signal output, and the actual temperature reading is obtained through the calibration value. When subjected to external force, the pressure sensor generates a voltage change, which is converted into an electrical signal and converted into a specific pressure value after calibration. The collected temperature data and pressure data are preprocessed. The preprocessing operation includes identifying and eliminating data points that are obviously deviated from the normal range, applying a low-pass filter, and filtering the data. The electrical signal curve is smoothed, and time synchronization and data integration operations are performed. Temperature fluctuation, pressure distribution and dynamic response are extracted from the preprocessed temperature and pressure data as key features for optimizing the square diamond cutting process. Among them, the temperature fluctuation feature is used to analyze the temperature change trend during the cutting process, identify the key moments of temperature rise and fall, and whether there is abnormal high temperature. The pressure distribution feature is used to calculate the maximum, minimum and average values ​​of the applied pressure during the cutting process, and evaluate whether the pressure is evenly distributed and whether there is any mutation. The dynamic response feature is used to explore the relationship between temperature and pressure and the relationship between the two over time, and to assist in understanding the temperature rise caused by frictional heat generation during the cutting process and the impact of applied pressure adjustment on the cutting effect.

[0025] S3. Extract the hardness distribution and color grade of the rough stone as key variables for predicting the cutting effect, use the optimal cutting path angle calculated according to the crystal direction as a derived variable, build a gradient boosting decision tree model, output the optimal cutting ratio parameters and the expected finished product quality, introduce a hardness tester to measure the point hardness at different positions of the rough stone, obtain the hardness value, map the hardness value to the coordinate position corresponding to the three-dimensional model, form a hardness distribution map covering the entire rough stone surface and part of the internal structure, statistically analyze the hardness distribution map, identify whether there is a clear boundary between the hard area and the soft area, extract the color value of the pixel point from the color image obtained during the three-dimensional scanning process, classify the collected color values ​​according to the color grading system of the diamond industry standard, determine the color grade of the rough stone, integrate the color grade information into the three-dimensional model, form a spatial distribution map of the color grade, use the crystal texture information extracted from the pre-processed three-dimensional data to determine the main arrangement direction of the crystal inside the rough stone, apply geometric algorithms to calculate the effects of light reflection and refraction at different cutting angles, and select the angle that maximizes the display of brightness and fire as the optimal cutting path angle. The calculation process is as follows: ; in, represents the optimal cutting angle, Indicates the possible cutting angles, Represents the reflectivity function at a given angle, reflecting the light at that angle Reflection efficiency under angle, Represents the transmittance function at a given angle, reflecting the light at that angle The transmission efficiency under the angle, the pre-processed shape contour, crystal texture, defect location, temperature fluctuation, pressure distribution, dynamic response, hardness distribution, color grade and optimal cutting path angle data are used as the training set and verification set for constructing the gradient boosting decision tree model, wherein 80% of the data are used as the training set and 20% of the data are used as the verification set, and the average value of the sample is used as the initial prediction value to construct an iterative process, wherein the iterative process includes calculating the residual between the initial prediction value and the actual value of the current model, and training a new decision tree with the reduced residual as the target variable, and multiplying the prediction result of the new decision tree by the learning rate and adding it to the existing model. In each round of iteration, the importance of different features for prediction is evaluated by analyzing the features used when the decision tree nodes are split and their contribution, and the mean square error is used as the loss function, and the calculation process is as follows: ; in, represents the loss value, represents the number of samples, represents the cutting ratio and finished product quality of the i-th sample, Represents the model's predicted value for the i-th sample. When the loss no longer decreases significantly, the iteration is stopped. The gradient boosting decision tree model outputs the optimal cutting ratio parameters and expected finished product quality based on the input new rough stone data. The optimal cutting ratio parameters include the crown angle and pavilion depth ratio, and the expected finished product quality includes brightness, fire and weight retention rate.

[0026] S4. Use cross-validation technology and test sets to verify the prediction accuracy of the model. Through comparative experiments, check whether the product quality and production efficiency have been significantly improved. Divide the training set into 10 subsets, each subset is used as a validation set in turn, and the remaining 9 subsets are combined as new training sets. Use each subset as a validation set to train the gradient boosting decision tree model, and record the mean square error performance index of the gradient boosting decision tree model on the validation set. Calculate the average performance index of 10 validations as the final evaluation result of the gradient boosting decision tree model, adjust the number and learning rate of the gradient boosting decision tree model, and select the best decision tree model according to the final evaluation result of cross-validation. The number of decision trees and learning rate parameters that minimize the loss are used as the final gradient boosting decision tree model configuration. A data set that has not participated in model training and verification is used as the test set. The feature information of the rough stone is input into the gradient boosting decision tree model, and the optimal cutting ratio parameters and expected finished product quality are output. Actual cutting is carried out according to the optimal cutting ratio parameters and the expected finished product quality, and the quality characteristics of the finished product are measured. The prediction accuracy of the model prediction is evaluated by comparing the prediction results of the gradient boosting decision tree model with the quality of the finished product after actual cutting. A batch of rough stones with similar characteristics in shape, size, crystal structure and hardness distribution are selected as experimental objects, and the actual results in the experiment are compared. The experimental group uses the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model to perform cutting operations, and the control group performs cutting operations according to the traditional empirical method, and records relevant data before and after cutting, including three-dimensional dimensions, crystal texture, defect location, hardness distribution and color grade information before cutting, temperature fluctuations, pressure distribution and dynamic response key features during cutting, quality characteristics of finished products after cutting, and production time and wear of cutting equipment. After cutting, the mean and standard deviation of the finished products of the experimental group and the control group in terms of brightness, fire and weight retention rate are calculated, and the t-test is used to determine whether there is a statistically significant difference between the experimental group and the control group. The average production time and tool wear of the experimental group and the control group are compared, and the proportion of production cost savings is calculated to quantify the improvement in production efficiency. According to the data analysis results, it is determined whether the gradient boosting decision tree model can improve product quality and production efficiency. If the finished product of the experimental group is significantly better than that of the control group in terms of brightness, fire, weight retention rate, etc., and the production time and tool wear are also reduced, it means that the gradient boosting decision tree model can improve product quality and production efficiency. If there is no obvious difference and there is a decline in indicators, the number and learning rate of decision trees of the gradient boosting decision tree model are adjusted.

[0027] S5. Feedback the actual cutting results to the system, adjust the model parameters, and regularly update the market preference information. After each cutting operation, record and organize the quality characteristics, production time and tool wear of the actual cut products, compare the actual cutting results with the predicted values ​​of the gradient boosting decision tree model, form new training samples, add the newly collected data to the original training set, and use the updated training set to train the gradient boosting decision tree model again. According to the results of cross-validation, adjust the number of decision trees and learning rate of the gradient boosting decision tree model to minimize the prediction error, calculate the mean square error performance index of the updated gradient boosting decision tree model on the test set, and compare the historical gradient boosting decision tree model to confirm the improvement effect. Regularly conduct market research to analyze the changes in consumers' requirements for the shape, proportion, color grade and optical performance of diamond products. Convert the results of market research into quantifiable preference indicators and incorporate them into the feature space of the gradient boosting decision tree model. Combined with the new market preference information, iteratively update the gradient boosting decision tree model to ensure that the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model meet the mainstream needs of the current market.

[0028] S6. Visual report generation software is used to display the optimized cutting results. A bar graph is used to compare the differences in brightness, fire and weight retention between the experimental group and the control group. A line graph is used to display the temperature fluctuation, pressure distribution and the trend of production efficiency over time. A scatter plot is used to display the relationship between the cutting angle and the optical performance of the finished diamond. A heat map is used to represent the spatial distribution of hardness distribution and color grade. The bar graph, line graph, scatter plot and heat map are integrated into an interactive dashboard, allowing users to filter different data dimensions according to their needs, add annotations and instructions next to the chart, and use the visual report generation software to automatically generate a comprehensive optimized cutting results report.

[0029] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A square diamond cutting process optimization method, characterized in that: The following steps are involved: S1. Collect the three-dimensional dimensions, crystal structure and internal defect data of the original stone through a high-resolution 3D scanner; S2, collecting temperature data and applied pressure data during the cutting process through a temperature sensor and a pressure sensor; S3, extract the hardness distribution and color grade of the rough stone as the key variables for predicting the cutting effect, use the optimal cutting path angle calculated according to the crystal direction as the derived variable, build a gradient boosting decision tree model, and output the optimal cutting ratio parameters and expected finished product quality; S4. Use cross-validation technology and test sets to verify the prediction accuracy of the model, and conduct comparative experiments to check whether product quality and production efficiency have been significantly improved; S5. Feedback the actual results of cutting to the system, adjust the model parameters, and regularly update the market preference information; S6. Use visual report generation software to display optimized cutting results.

2. The method for optimizing square diamond cutting process according to claim 1, characterized in that: In S1, the process of collecting the three-dimensional dimensions, crystal structure and internal defect data of the rough stone by means of a high-resolution 3D scanner includes: A high-resolution 3D scanner is deployed on the workbench, and the rough stone to be cut is placed in the scanning area. The high-resolution 3D scanner is based on optical principles and uses a white light source to measure the surface of the object point by point. The scanning light emitted by the white light source is irradiated on the surface of the rough stone and then reflected back to the receiver. The spatial coordinates of each point on the surface of the rough stone are calculated based on the time difference of the reflected light, and then a three-dimensional model of the rough stone is constructed. The transmission imaging technology is used to allow the light to penetrate the rough stone, capture the image information of the internal features of the rough stone, and then obtain the crystal structure and internal defect data; Preprocessing the collected data on the three-dimensional dimensions, crystal structure and internal defects of the rough stone, wherein the preprocessing operation includes removing abnormal values ​​and outliers in the data, performing smoothing to reduce the interference caused by irregularities on the surface of the rough stone, and splicing and fusing the scanning results at different viewing angles to form a three-dimensional data set; The shape contour, crystal texture and defect location are extracted from the preprocessed three-dimensional data as the key features for optimizing the square diamond cutting process. Among them, the shape contour is used to determine the boundary range and geometric shape of the rough stone, the crystal texture is used to identify the crystallization direction and arrangement inside the rough stone, and the defect location is used to locate the specific position and distribution of cracks and inclusions inside the rough stone.

3. The method for optimizing square diamond cutting process according to claim 2, characterized in that: In S2, the process of collecting temperature data and applying pressure data during the cutting process by using a temperature sensor and a pressure sensor includes: A temperature sensor is deployed on the workbench near the cutting area, and a pressure sensor is deployed between the contact surface of the cutting tool and the rough stone. When the temperature changes, the physical properties of the material inside the temperature sensor change, and the physical property change is converted into an electrical signal output. The actual temperature reading is obtained by calibrating the value. When subjected to external force, the pressure sensor generates a voltage change, which is converted into an electrical signal and converted into a specific pressure value after calibration. Preprocessing the collected temperature data and pressure data includes identifying and eliminating data points that are obviously deviated from the normal range, applying a low-pass filter to smooth the electrical signal curve, and performing time synchronization and data integration operations; Temperature fluctuation, pressure distribution and dynamic response are extracted from the pre-processed temperature and pressure data as key features for optimizing the square diamond cutting process. The temperature fluctuation feature is used to analyze the temperature change trend during the cutting process, identify the key moments of temperature rise and fall, and whether there is abnormally high temperature. The pressure distribution feature is used to calculate the maximum, minimum and average values ​​of the applied pressure during the cutting process, and evaluate whether the pressure is evenly distributed and whether there is any mutation. The dynamic response feature is used to explore the relationship between temperature and pressure and the relationship between the two over time, and to assist in understanding the temperature rise caused by frictional heat generation during the cutting process and the influence of applied pressure adjustment on the cutting effect.

4. The method for optimizing square diamond cutting process according to claim 3, characterized in that: In S3, the process of extracting the hardness distribution and color grade of the rough stone as key variables for predicting the cutting effect includes: A hardness tester is introduced to measure the point hardness at different positions of the rough stone to obtain the hardness value, and the hardness value is mapped to the coordinate position corresponding to the three-dimensional model to form a hardness distribution map covering the entire rough stone surface and part of the internal structure. The hardness distribution map is statistically analyzed to identify whether there is a clear boundary between the hard area and the soft area; The color values ​​of the pixels are extracted from the color images obtained during the 3D scanning process. According to the color grading system of diamond industry standards, the collected color values ​​are classified to determine the color grade of the rough stone. The color grade information is integrated into the 3D model to form a spatial distribution map of the color grade.

5. The method for optimizing square diamond cutting process according to claim 4, characterized in that: In S3, the process of calculating the optimal cutting path angle according to the crystal direction and using it as a derived variable includes: The crystal texture information extracted from the pre-processed 3D data is used to determine the main arrangement direction of the crystals inside the original stone. The geometric algorithm is used to calculate the effects of light reflection and refraction at different cutting angles, and the angle that maximizes the display of brightness and fire is selected as the optimal cutting path angle. The calculation process is as follows: ； in, represents the optimal cutting angle, Indicates the possible cutting angles, Represents the reflectivity function at a given angle, reflecting the light at that angle Reflection efficiency under angle, Represents the transmittance function at a given angle, reflecting the light at that angle Transmission efficiency under angle.

6. The method for optimizing square diamond cutting process according to claim 5, characterized in that: In S3, the process of constructing a gradient boosting decision tree model and outputting the optimal cutting ratio parameters and the expected finished product quality includes: The pre-processed shape contour, crystal texture, defect location, temperature fluctuation, pressure distribution, dynamic response, hardness distribution, color grade and optimal cutting path angle data were used as training sets and validation sets for building the gradient boosting decision tree model, where 80% of the data was used as the training set and 20% of the data was used as the validation set; The average value of the sample is used as the initial prediction value to construct an iterative process, which includes calculating the residual between the initial prediction value and the actual value of the current model, reducing the residual as the target variable to train a new decision tree, multiplying the prediction result of the new decision tree by the learning rate and adding it to the existing model. In each round of iteration, the importance of different features for prediction is evaluated by analyzing the features used when the decision tree node is split and their contribution, and the mean square error is used as the loss function. The calculation process is as follows: ； in, represents the loss value, represents the number of samples, represents the cutting ratio and finished product quality of the i-th sample, Represents the model's predicted value for the i-th sample, and stops iterating when the loss no longer decreases significantly; The gradient boosting decision tree model outputs optimal cutting proportion parameters and expected finished product quality based on the input new rough stone data. The optimal cutting proportion parameters include crown angle and pavilion depth ratio, and the expected finished product quality includes brightness, fire and weight retention rate.

7. The method for optimizing square diamond cutting process according to claim 6, characterized in that: In S4, the process of verifying the prediction accuracy of the model using the cross-validation technique and the test set includes: The training set is divided into 10 subsets, each subset is used as a validation set in turn, and the remaining 9 subsets are combined as new training sets. The gradient boosting decision tree model is trained with each subset as a validation set, and the mean square error performance index of the gradient boosting decision tree model on the validation set is recorded. The average performance index of 10 validations is calculated as the final evaluation result of the gradient boosting decision tree model, and the number and learning rate of decision trees in the gradient boosting decision tree model are adjusted. According to the final evaluation result of cross-validation, the number of decision trees and learning rate parameters that minimize the loss are selected as the final gradient boosting decision tree model configuration; A data set that has not participated in model training and validation is used as a test set. The feature information of the original stone is input into the gradient boosting decision tree model, and the optimal cutting ratio parameters and expected finished product quality are output. Actual cutting is performed according to the optimal cutting ratio parameters and the expected finished product quality, and the quality characteristics of the finished product are measured. The prediction accuracy of the model prediction is evaluated by comparing the prediction results of the gradient boosting decision tree model with the quality of the finished product after actual cutting.

8. The method for optimizing square diamond cutting process according to claim 7, characterized in that: In S4, the process of testing whether the product quality and production efficiency are significantly improved through comparative experiments includes: A batch of rough stones with similar characteristics in shape, size, crystal structure and hardness distribution were selected as experimental objects. The experimental group in the comparative experiment used the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model for cutting operations, and the control group performed cutting operations according to the traditional empirical method, and recorded relevant data before and after cutting, including three-dimensional dimensions, crystal texture, defect location, hardness distribution and color grade information before cutting, temperature fluctuations, pressure distribution and dynamic response key features during the cutting process, quality characteristics of the finished product after cutting, and production time and degree of wear of the cutting equipment; After cutting, calculate the mean and standard deviation of the brightness, fire and weight retention of the finished products of the experimental group and the control group. Use the t-test to determine whether there is a statistically significant difference between the experimental group and the control group. Compare the average production time and tool wear of the experimental group and the control group, and calculate the proportion of production cost savings to quantify the improvement in production efficiency. According to the data analysis results, determine whether the gradient boosting decision tree model can improve product quality and production efficiency. If the finished product of the experimental group is significantly better than that of the control group in terms of brightness, fire color, weight retention rate, etc., and the production time and tool wear are also reduced, it means that the gradient boosting decision tree model can improve product quality and production efficiency. If there is no obvious difference and there is a decline in indicators, adjust the number and learning rate of decision trees of the gradient boosting decision tree model.

9. The method for optimizing square diamond cutting process according to claim 8, characterized in that: In S5, the process of feeding back the actual results of the cutting to the system, adjusting the model parameters, and regularly updating the market preference information includes: After each cutting operation is completed, the quality characteristics of the actual cut product, production time and tool wear are recorded and sorted out, and the actual cutting results are compared with the predicted values ​​of the gradient boosting decision tree model to form new training samples, and the newly collected data are supplemented to the original training set; Use the updated training set to train the gradient boosting decision tree model again. According to the results of cross-validation, adjust the number of decision trees and learning rate of the gradient boosting decision tree model to minimize the prediction error. Calculate the mean square error performance index of the updated gradient boosting decision tree model on the test set and compare it with the historical gradient boosting decision tree model to confirm the improvement effect. Conduct market research regularly to analyze changes in consumers’ requirements for the shape, proportion, color grade, and optical performance of finished diamonds. Convert the results of market research into quantifiable preference indicators and incorporate them into the feature space of the gradient boosting decision tree model. Combined with new market preference information, iteratively update the gradient boosting decision tree model to ensure that the optimal cutting ratio parameters and expected finished product quality output by the gradient boosting decision tree model meet the mainstream demand of the current market.

10. The method for optimizing square diamond cutting process according to claim 9, characterized in that: In the S6, the process of displaying the optimized cutting results through the visual report generation software includes: Use bar graphs to compare the differences in brightness, fire and weight retention between the experimental and control groups. Use line graphs to show the temperature fluctuations, pressure distribution and production efficiency trends over time. Use scatter graphs to show the relationship between cutting angles and the optical properties of finished diamonds. Use heat maps to represent the spatial distribution of hardness distribution and color grade. Integrate bar graphs, line graphs, scatter graphs and heat maps into interactive dashboards, allowing users to filter different data dimensions according to their needs, add annotations and descriptions next to the charts, and use visual report generation software to automatically generate a comprehensive optimized cutting results report.

Images