A Material Removal Depth Prediction Method Based on Improved XGBoost

Abstract

This invention relates to the field of machine learning technology, and in particular to a method for predicting material removal depth based on an improved XGBoost algorithm. The method first collects a grinding and polishing process dataset and divides it into training and testing sets. After preprocessing the dataset, an improved Egret Optimization Algorithm (IESOA) is used to optimize the hyperparameters of the Extreme Gradient Boosting Algorithm (XGBoost), thereby establishing a predictive model for material removal depth. Finally, the contribution of each grinding and polishing process parameter to the material removal depth is quantified based on the Shapley Additive Interpretation (SHAP) method, enabling visual analysis of the model's decision-making process. This invention can significantly improve the prediction accuracy of material removal depth in grinding and polishing processes, and clarifies the influence mechanism of each process parameter through interpretability analysis, possessing the dual advantages of high prediction accuracy and interpretable decision-making.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a material removal depth prediction method based on an improved XGBoost. Background Technology

[0002] With the continuous development of industrial technology, the requirements for product surface quality are increasing. In key components such as aircraft engine blades and automobile bodies, surface quality not only directly affects the product's appearance but also has a significant impact on its performance and reliability. Grinding and polishing, as the final process for improving workpiece surface quality, is particularly crucial in its processing effect. Currently, workpiece surface treatment mainly relies on three methods: manual grinding, CNC machine tool grinding, and robotic grinding. Manual grinding is highly dependent on the experience and skill level of the operators, and has limitations such as poor processing stability and difficulty in ensuring surface consistency. Although CNC machine tool grinding has high precision, it suffers from high equipment costs, limited workspace, and insufficient adaptability and flexibility. In contrast, robotic grinding has advantages such as high repeatability and good processing consistency, and is gradually becoming an important development direction in the field of industrial grinding.

[0003] In robotic automated grinding and polishing processes, stable and uniform material removal from different types of workpieces can be achieved by adjusting process parameters such as spindle speed, feed rate, and grinding contact force. However, how to comprehensively coordinate these factors to maximize grinding and polishing efficiency while ensuring surface quality remains a challenge. The key lies in establishing an accurate material removal depth prediction model as the basis for comprehensive control.

[0004] The material removal depth in robotic polishing is influenced by the coupling effects of multiple factors, including process parameters (spindle speed, feed rate, polishing contact force), polishing geometry (polishing angle, polishing tool radius), and workpiece physical properties (workpiece surface curvature). These influencing factors exhibit a complex and highly nonlinear relationship with the material removal depth. Traditional regression models, limited by their reliance on fitting methods, struggle to capture the complex interdependencies among multiple variables, leading to insufficient prediction accuracy. While neural networks or deep learning methods perform well, they typically require large amounts of data and are highly dependent on sample size. Furthermore, machine learning models often treat the complex material removal mechanism as a "black box," making it difficult to understand the model's decision-making process and resulting in poor model interpretability. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] This invention provides a material removal depth prediction method based on an improved XGBoost, which overcomes the problems of strong sample dependence, low prediction accuracy of traditional regression models, and poor model interpretability in existing technologies.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, this invention provides a method for predicting material removal depth based on an improved XGBoost, comprising the following steps:

[0009] Step S1: Collect the original dataset containing grinding and polishing process parameters and material removal depth, divide the original dataset into training set and test set, and preprocess the data in the original dataset.

[0010] Step S2: Construct a prediction model based on the XGBoost algorithm, with grinding and polishing process parameters as input and material removal depth as output;

[0011] Step S3: The improved egret flock optimization algorithm IESOA is used to optimize the hyperparameters in the XGBoost prediction model in step S2.

[0012] The improved egret flock optimization algorithm IESOA uses Tent chaotic mapping to initialize the population and introduces sine and cosine operators for position updates.

[0013] Step S4: Train the optimized XGBoost prediction model using the training set divided in Step S1, test the trained XGBoost prediction model using the test set divided in Step S1, and calculate the evaluation index of the XGBoost prediction model.

[0014] Step S5: Based on the Shapley additive interpretation (SHAP) method, quantify the contribution of each grinding and polishing process parameter to the material removal depth, and analyze the influence mechanism of each grinding and polishing process parameter on the material removal depth.

[0015] Preferably, the expression for the Tent chaotic mapping in step S3 is:

[0016] in, , The value is 2.

[0017] Preferably, the position update formula for the sine and cosine operators introduced in step S3 is as follows:

[0018] in, This represents the spatial position of the current individual in the i-th dimension during the t-th iteration. To adjust the parameters, , and The parameter is random.

[0019] The The The .

[0020] Preferably, the formula for calculating the SHAP value in step S5 is:

[0021] in, This represents the Shapley value of the i-th feature. Represents the total number of features. Indicating in the feature subset Add features based on The model prediction value at that time, Indicates using only a subset of features The model's predicted value at that time.

[0022] (III) Beneficial Effects

[0023] This invention provides a method for predicting material removal depth based on an improved XGBoost algorithm. By employing the improved Egret Optimization Algorithm (IESOA) to optimize the hyperparameters of the XGBoost model, the prediction accuracy of material removal depth is significantly improved. The IESOA algorithm uses the Tent chaotic mapping to initialize the population, enhancing the diversity and ergodicity of the population in the search space; and it introduces sine and cosine operators to participate in individual position updates, effectively balancing the algorithm's global exploration and local exploitation capabilities, thereby avoiding getting trapped in local optima. This allows the optimized XGBoost model to more accurately capture the complex nonlinear mapping relationship between grinding and polishing process parameters and material removal depth.

[0024] Based on this, the present invention also innovatively introduces the Shapley additive interpretation method. By calculating the Shapley value of each grinding and polishing process parameter, the contribution of each parameter to the material removal depth is quantitatively evaluated, which significantly improves the interpretability and credibility of the model. This method not only realizes the visualization and transparency of the model decision-making process, but also provides a reliable theoretical basis for the optimization and control of process parameters, thereby supporting the further development of grinding and polishing processes towards precision. Attached Figure Description

[0025] Figure 1 This diagram illustrates a process flow of a material removal depth prediction method based on an improved XGBoost according to the present invention.

[0026] Figure 2 This demonstrates a material removal depth prediction model based on an improved XGBoost model.

[0027] Figure 3 The importance of the grinding and polishing process parameters is ranked.

[0028] Figure 4 A scatter plot showing the importance of grinding and polishing process parameters;

[0029] Figure 5 The SHAP dependency graph of grinding and polishing contact force is shown.

[0030] Figure 6 This shows the SHAP dependency graph of the grinding and polishing spindle speed;

[0031] Figure 7 The SHAP dependency graph of the robot feed rate is shown. Detailed Implementation

[0032] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. The technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] In the description of this invention, it is necessary to understand that the orientations or positional relationships indicated by terms such as "upper," "lower," "left," "right," "inner," "outer," "top," and "bottom" are based on the orientations or positional relationships shown in the accompanying drawings. They are intended only to facilitate the description of this invention and to simplify the description, and are not intended to indicate or imply that the components referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0034] like Figure 1-7 As shown, this invention provides a method for predicting material removal depth based on an improved XGBoost, comprising the following steps:

[0035] This invention provides a processing depth prediction method based on IESOA-XGBoost, comprising the following steps:

[0036] Step S1: Collect the original dataset containing grinding and polishing process parameters and material removal depth, divide the original dataset into training set and test set, and preprocess the data in the original dataset.

[0037] In step S1, a stratified sampling method is used to divide the collected original dataset into a training set and a test set, where the training set is 80% of the original dataset and the test set is 20% of the original dataset.

[0038] The data preprocessing in step S1 of the original dataset includes:

[0039] Step S11: Perform max-min normalization on the data in the original dataset to map it to the [0,1] interval;

[0040] In step S1, the original dataset contains grinding and polishing process parameters and material removal depth. These grinding and polishing process parameters include: spindle speed. Robot feed speed and grinding and polishing contact force ;

[0041] In step S11, the maximum-minimum normalization process maps the data to the interval [0,1]. This maximum-minimum normalization aims to eliminate outliers and inconsistencies in the units of measurement. The formula for calculating the maximum-minimum normalization is as follows:

[0042]

[0043] in, It is an eigenvalue. and These are the minimum and maximum values ​​of the feature, respectively. This is the result after eigenvalue normalization.

[0044] Step S2: Construct a prediction model based on the XGBoost algorithm, taking the grinding and polishing process parameters as input and the material removal depth as output;

[0045] The objective function of the XGBoost algorithm is:

[0046]

[0047] in, To train the loss function, For regularization terms;

[0048] The model predicts the value. The calculation formula is:

[0049]

[0050] in, Indicates the number of samples. For the number of regression trees, Indicates the first Decision tree, These are the eigenvectors.

[0051] Step S3: The improved egret flock optimization algorithm IESOA is used to optimize the hyperparameters in the XGBoost prediction model in step S2.

[0052] The IESOA algorithm uses Tent chaotic mapping to initialize the population and introduces sine and cosine operators for position updates;

[0053] The hyperparameters, hyperparameter value ranges, and optimal hyperparameters of the XGBoost model are shown in Table 1:

[0054] Table 1 Hyperparameter Values

[0055]

[0056] In step S3, the IESOA algorithm uses the Tent chaotic map to initialize the population. Compared with traditional random initialization or simple logical mapping, the Tent chaotic map has a more uniform distribution and a faster traversal speed. It can effectively avoid the problem of uneven distribution of the initial position of the population in the search space and clustering in local areas, and can significantly enhance the diversity and traversability of the population.

[0057] By introducing sine and cosine operators to participate in individual position updates, the algorithm's global exploration and local development capabilities are effectively balanced, thereby avoiding getting trapped in local optima. This allows the optimized XGBoost model to more accurately capture the complex nonlinear mapping relationship between grinding and polishing process parameters and material removal depth.

[0058] In step S3, the expression for the Tent chaotic mapping is:

[0059]

[0060] in, , The value is 2;

[0061] In step S3, the sine and cosine position update formulas are as follows:

[0062]

[0063] in, This represents the spatial position of the current individual in the i-th dimension during the t-th iteration. To adjust the parameters, , and The parameter is random.

[0064] Should ,Should ,Should .

[0065] Step S4: Train the optimized XGBoost prediction model using the training set divided in step S1, test the trained XGBoost prediction model using the test set divided in step S1, and calculate the evaluation index of the XGBoost prediction model.

[0066] In step S4, the training set is trained using the five-fold cross-validation method. This method maximizes the use of limited training data and effectively improves the model's generalization ability. The model evaluation metric is calculated using the following formula:

[0067]

[0068]

[0069]

[0070] in, The actual value of the target variable. These are the model's predicted values. The target variable is the average value, and n represents the sample size. The evaluation metrics for this model are shown in Table 2.

[0071] Table 2 Model Evaluation Index Table

[0072]

[0073] Step S5: Based on the Shapley additive interpretation SHAP method, quantify the contribution of each grinding and polishing process parameter to the material removal depth, and analyze the influence mechanism of each grinding and polishing process parameter on the material removal depth.

[0074] In step S5, the formula for calculating the SHAP value is as follows:

[0075]

[0076] in, This represents the Shapley value of the i-th feature. Represents the total number of features. Indicating in the feature subset Add features based on The model prediction value at that time, Indicates using only a subset of features Model predictions at that time;

[0077] It is understood that the various embodiments mentioned above in this invention can be combined with each other to form combined embodiments without violating the principle and logic. Due to space limitations, this invention will not elaborate further.

[0078] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0079] This invention provides a method for predicting material removal depth based on an improved XGBoost algorithm. By employing the improved Egret Optimization Algorithm (IESOA) to optimize the hyperparameters of the XGBoost model, the prediction accuracy of material removal depth is significantly improved. The IESOA algorithm utilizes the Tent chaotic mapping to initialize the population, enhancing the diversity and ergodicity of the population in the search space. Furthermore, it introduces sine and cosine operators to participate in individual position updates, effectively balancing the algorithm's global exploration and local exploitation capabilities, thus avoiding getting trapped in local optima. This allows the optimized XGBoost model to more accurately capture the complex nonlinear mapping relationship between grinding and polishing process parameters and material removal depth. Building upon this, an innovative Shapley additive interpretation method is introduced. By calculating the Shapley values ​​of each grinding and polishing process parameter, the contribution of that parameter to the material removal depth is quantitatively evaluated, significantly improving the model's interpretability and reliability. This method not only visualizes and makes the model's decision-making process transparent but also provides a reliable theoretical basis for the optimization and control of process parameters, thereby supporting the further development of grinding and polishing processes towards precision.

[0080] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for predicting material removal depth based on an improved XGBoost, characterized in that, Includes the following steps: Step S1: Collect the original dataset containing grinding and polishing process parameters and material removal depth, divide the original dataset into training set and test set, and preprocess the data in the original dataset. Step S2: Construct a prediction model based on the XGBoost algorithm, with grinding and polishing process parameters as input and material removal depth as output; Step S3: The improved egret flock optimization algorithm IESOA is used to optimize the hyperparameters in the XGBoost prediction model in step S2. The improved egret flock optimization algorithm IESOA uses Tent chaotic mapping to initialize the population and introduces sine and cosine operators for position updates. Step S4: Train the optimized XGBoost prediction model using the training set divided in Step S1, test the trained XGBoost prediction model using the test set divided in Step S1, and calculate the evaluation index of the XGBoost prediction model. Step S5: Based on the Shapley additive interpretation (SHAP) method, quantify the contribution of each grinding and polishing process parameter to the material removal depth, and analyze the influence mechanism of each grinding and polishing process parameter on the material removal depth.

2. The material removal depth prediction method based on improved XGBoost according to claim 1, characterized in that, The expression for the Tent chaotic mapping in step S3 is: in, , The value is 2.

3. The material removal depth prediction method based on improved XGBoost according to claim 2, characterized in that, The position update formula for the sine and cosine operators introduced in step S3 is as follows: in, This represents the spatial position of the current individual in the i-th dimension during the t-th iteration. To adjust the parameters, , and The parameter is random. The The The .

4. The material removal depth prediction method based on improved XGBoost according to claim 1, characterized in that, The formula for calculating the SHAP value in step S5 is as follows: in, This represents the Shapley value of the i-th feature. Represents the total number of features. Indicating in the feature subset Add features based on The model prediction value at that time, Indicates using only a subset of features The model's predicted value at that time.

Images