CN117826599A - Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning - Google Patents
Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning Download PDFInfo
- Publication number
- CN117826599A CN117826599A CN202410004177.2A CN202410004177A CN117826599A CN 117826599 A CN117826599 A CN 117826599A CN 202410004177 A CN202410004177 A CN 202410004177A CN 117826599 A CN117826599 A CN 117826599A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- real
- machine tool
- numerical control
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010801 machine learning Methods 0.000 title claims abstract description 50
- 238000005457 optimization Methods 0.000 title claims abstract description 24
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 15
- 238000005520 cutting process Methods 0.000 claims abstract description 55
- 238000000034 method Methods 0.000 claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 31
- 238000003754 machining Methods 0.000 claims abstract description 24
- 230000008569 process Effects 0.000 claims abstract description 21
- 238000013136 deep learning model Methods 0.000 claims abstract description 16
- 230000000694 effects Effects 0.000 claims abstract description 10
- 238000012795 verification Methods 0.000 claims abstract description 10
- 238000012544 monitoring process Methods 0.000 claims abstract description 9
- 239000000463 material Substances 0.000 claims abstract description 6
- 238000012706 support-vector machine Methods 0.000 claims description 25
- 238000012549 training Methods 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 238000011217 control strategy Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000001276 controlling effect Effects 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000001131 transforming effect Effects 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003801 milling Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000007514 turning Methods 0.000 description 1
Landscapes
- Numerical Control (AREA)
Abstract
The invention discloses a numerical control machine tool optimization algorithm based on artificial intelligence and machine learning, which comprises the following steps: step 1: collecting data in the machining process of the numerical control machine tool; step 2: normalizing and normalizing the characteristics by using the cutting speed, the feeding speed, the cutting depth and the workpiece material characteristics; step 3: predicting an optimal tool path using a machine learning model; step 4: adjusting parameters of the model by using a cross verification method to improve the generalization performance of the model, and introducing a deep learning model; step 5: real-time monitoring data in the processing process, including a cutter motion trail, vibration frequency and temperature, inputting the real-time data into a trained machine learning model, and predicting an optimal cutter path; step 6: and dynamically adjusting control parameters of the numerical control machine according to the prediction result of the model so as to realize the optimal cutting effect. The invention can realize self-adaptive control, and adjust the cutter path and cutting parameters according to real-time data, so that the processing process is more intelligent and flexible.
Description
Technical Field
The invention relates to the technical field of numerical control machine tools, in particular to a numerical control machine tool optimization algorithm based on artificial intelligence and machine learning.
Background
The numerical control machine optimization algorithm is an algorithm designed for improving the machining efficiency, reducing the energy consumption and improving the product quality of the numerical control machine, and is mainly applied to a numerical control (Computer Numerical Control, CNC) machine, wherein the machine controls and operates the movement of tools through a computer program and is used for completing various machining tasks such as milling, turning, drilling and the like. The problem of the tool path of the numerical control machine tool is a key technical problem faced in the machining process of the numerical control machine tool, and relates to determining the movement track of the tool on the surface of a workpiece so as to realize efficient and accurate machining under given cutting conditions. However, the existing control cannot realize self-adaptive control, and the processing process is inflexible.
Disclosure of Invention
In order to solve the problems, the invention provides a numerical control machine tool optimization algorithm based on artificial intelligence and machine learning.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
the invention discloses a numerical control machine tool optimization algorithm based on artificial intelligence and machine learning, which comprises the following steps:
step 1: collecting data in the machining process of the numerical control machine tool, including cutting parameters, machine tool states and machining results, and preprocessing the data;
step 2: normalizing and normalizing the characteristics by using the cutting speed, the feeding speed, the cutting depth and the workpiece material characteristics;
step 3: predicting an optimal tool path using a machine learning model;
step 4: adjusting parameters of the model by using a cross verification method to improve the generalization performance of the model, and introducing a deep learning model to capture complex nonlinear relations;
step 5: real-time monitoring data in the processing process, including a cutter motion trail, vibration frequency and temperature, inputting the real-time data into a trained machine learning model, and predicting an optimal cutter path;
step 6: and dynamically adjusting control parameters of the numerical control machine according to the prediction result of the model so as to realize the optimal cutting effect.
Further: the step 1 comprises the following steps:
the method comprises the steps that a sensor is arranged on a numerical control machine tool and used for monitoring data in the machining process in real time, wherein the data comprise cutting speed V, feeding speed F, cutting depth D, vibration Vib and temperature T, and a numerical control system is used for recording cutter paths, cutting parameters and machine tool state information;
integrating the data collected from different sensors and systems to form a data set, aligning time stamps, and ensuring the time consistency of each data source;
identifying and processing outliers to ensure data integrity;
according to the processing result, the data is labeled as a target variable, and the label is ensured to correspond to the corresponding characteristic;
taking into account the cutting force F c The relation between the cutting speed V, the feed speed F, and the cutting depth D is expressed by using an empirical formula:
F C ＝k 1 ·V n ·F m ·D p
wherein k1 is an empirical coefficient, and n, m and p are empirical indexes.
Further: the step 2 comprises the following steps:
selecting a characteristic which has obvious influence on the optimal cutter path;
the selected features are normalized to ensure similar dimensions under different dimensions, the normalization formula is:
wherein x' is a normalized characteristic value, x is an original characteristic value, and μ and σ are the mean value and standard deviation of the characteristic respectively;
combining the features with strong correlation to form new features so as to enhance the fitting capability of the model to complex relations;
according to the performance feedback of the model, continuously adjusting the selection and processing modes of the features;
the influence of the features on the model output is known by utilizing the explanatory analysis of the model, so that the process of the feature engineering is further optimized;
considering the relation between the cutting power P and the cutting speed V, the feeding speed F and the cutting depth D in the machining process, and using a formula after characteristic engineering to express:
P＝k 2 ·V′+k 3 ·F′+k 4 ·D′
wherein k is 2 、k 3 、k 4 Weights obtained for model training, V ', F ', D ', are normalized and transformed features.
Further: the step 3 comprises the following steps:
using a support vector machine regression model;
dividing the data set processed by the feature engineering into a training set and a testing set;
training a support vector machine regression model by using a training set;
the training process of the support vector machine regression model is represented by the following optimization problem:
the limiting conditions are as follows:
y i -w·Φ(x i )-b≤δ+ξ i
wherein C is a penalty parameter, y i Is the actual output value, w is the weight, Φ (x i ) Is input data x i Is the bias term, ε is the marginal width, ζ i Andis a relaxation variable, i is a sample index, n is the number of samples;
and verifying the trained model by using a test set, evaluating the performance of the model, and measuring the prediction accuracy of the model by using a mean square error index:
where n is the number of samples tested, y i Is a true value of the code,is a model predictive value;
the predictive function of the support vector machine regression model is:
wherein K (x i X) is a kernel function, alpha i Andis the Lagrangian multiplier of the support vector machine, b is the bias term;
and analyzing parameters of the regression model of the support vector machine to adjust the complexity of the model.
Further: the step 4 comprises the following steps:
performing super-parameter tuning on the support vector machine regression model by using a cross verification method so as to improve the generalization performance of the model;
introducing a multi-layer perceptron neural network to capture complex nonlinear relations, and building a neural network model comprising a plurality of hidden layers by using a deep learning model;
the multi-layer perceptron neural network output is expressed as:
wherein X is an input feature, W (1) And W is (2) Is a weight matrix, b (1) And b (2) Is a bias term, σ is an activation function;
performing performance comparison on the optimized support vector machine regression model and the deep learning model, and evaluating generalization capability of the support vector machine regression model by using a verification set;
selecting a model with better performance as a final numerical control machine tool optimization algorithm model;
for the selected final model, carrying out model explanatory analysis to know the influence of the model on the cutter path;
and interpreting the prediction result of the model by using an interpretability tool in the deep learning model.
Further: the step 5 comprises the following steps:
the sensors are reasonably arranged on the numerical control machine tool so as to monitor various data in the processing process in real time, and proper sampling frequency is set to ensure that enough real-time data are acquired;
real-time data acquired by the sensor are processed, so that the quality and usability of the data are ensured;
inputting real-time data into a model by using a trained machine learning model to obtain a real-time prediction result of the model, and predicting an optimal cutter path by the machine learning model according to the real-time input data:
wherein,is a real-time prediction result, and the ML_model is a trained machine learning Model, X real-time Is data input in real time;
applying the optimal tool path predicted in real time by the machine learning model to a control system of the numerical control machine tool, and dynamically adjusting control parameters of the numerical control machine tool to realize an optimal cutting effect;
the input and parameters of the model are continuously adjusted according to actual processing feedback, so that the adaptability and the robustness of the algorithm are further improved:
new parameter = old parameter + alpha× (target value-actual value)
Wherein alpha is the learning rate and is used for controlling the step length of adjustment;
and updating the model in a certain time interval, and training the model by utilizing the latest real-time data so as to adapt to the change in the processing process.
Further, the step 6 includes:
according to the real-time prediction result of the machine learning model, dynamically adjusting the control parameters of the numerical control machine tool, and comparing the model output serving as a reference signal with the actual control parameters by utilizing the self-adaptive control strategy to adjust the control parameters:
new control parameter = old control parameter + a× (target value-actual value)
Wherein alpha is the learning rate and is used for controlling the step length of adjustment;
and comparing the real-time feedback with a model prediction result through a proportional-integral-derivative controller to form a closed-loop control system:
where u (t) is the controller output, e (t) is the error between the actual value and the target value, K p 、K i 、K d Proportional, integral and differential coefficients, respectively;
implemented by a gradient descent optimization algorithm to minimize the error between the actual value and the model predicted value:
where θ is a model parameter, J (θ) is a loss function, η is a learning rate,is the gradient of the loss function with respect to the parameter.
Compared with the prior art, the invention has the following technical progress:
the traditional numerical control machine tool generally uses a predefined tool path and cutting parameters, and cannot flexibly adapt to different working conditions. In contrast, the algorithm based on artificial intelligence and machine learning can realize self-adaptive control, and adjust the cutter path and cutting parameters according to real-time data, so that the machining process is more intelligent and flexible. According to the invention, by learning a large amount of processing data, the algorithm can predict the optimal cutter path under specific conditions, so that the cutting time is minimized, the processing efficiency is improved, the method far exceeds the rule of thumb set by traditional manual, and the cutting effect is more optimized. The invention analyzes real-time data through a machine learning model, and an algorithm can predict potential vibration and impact and take the potential vibration and impact into consideration in tool path planning so as to minimize the adverse effects, and the complex dynamic changes are difficult to consider in the traditional method.
The invention can adjust the cutter path in real time in the processing process due to the real-time monitoring and learning mechanism so as to adapt to the changes of factors such as workpiece materials, cutter abrasion, machine tool states and the like, improve the real-time performance and be beneficial to coping with complex processing environments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention.
In the drawings:
FIG. 1 is a flow chart of the present invention.
Detailed Description
The following embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. Embodiments of the present invention will be described below with reference to the accompanying drawings.
As shown in fig. 1, the invention discloses a numerical control machine tool optimization algorithm based on artificial intelligence and machine learning, which comprises the following steps:
step 1: data collection and preparation
Data in the machining process of the numerical control machine tool is collected, including cutting parameters (speed, feeding and the like), machine tool states (vibration, temperature and the like) and machining results (surface quality, machining time and the like), and the data are preprocessed, including abnormal value removal, missing data filling and the like.
Step 2: feature engineering
Features related to machining performance, such as cutting speed, feeding speed, cutting depth, workpiece materials and the like, are selected, and the features are subjected to normalization, normalization and the like so as to ensure stability in a machine learning model.
Step 3: establishing machine learning model
Selecting an appropriate machine learning model for predicting an optimal tool path, dividing the data set into a training set and a test set, training the model with the training set, and verifying the performance of the model on the test set.
Step 4: model optimization
And the model is subjected to parameter adjustment by using methods such as cross verification and the like so as to improve the generalization performance of the model, and a deep learning model such as a neural network is introduced so as to better capture complex nonlinear relations.
Step 5: real-time data monitoring and feedback
The method comprises the steps of deploying sensors on a numerical control machine tool, monitoring data in the machining process in real time, including tool motion trail, vibration frequency, temperature and the like, inputting the real-time data into a trained machine learning model, and predicting an optimal tool path.
Step 6: adaptive control
According to the prediction result of the model, the control parameters of the numerical control machine tool, such as cutter speed, feeding speed and the like, are dynamically adjusted to achieve the optimal cutting effect, closed-loop control is implemented, the model is continuously updated according to real-time feedback, and the adaptability and the robustness of the algorithm are improved.
Specifically, step 1 includes:
and (3) data collection: sensors are installed on the numerically controlled machine tool for real-time monitoring of various data during machining, including but not limited to cutting speed V, feed speed F, cutting depth D, vibration Vib, temperature T, etc., and important information such as tool path, cutting parameters, machine tool state, etc. is recorded by the numerically controlled system.
Data integration: the data collected from the different sensors and systems are integrated to form a comprehensive data set. The time stamps are aligned to ensure time consistency of the individual data sources.
Feature extraction: depending on the important variables in the process, suitable features, such as V, F, D, vib, T, etc., are selected to improve the performance of the machine learning model.
Outlier handling and missing value filling: abnormal values are identified and processed by using a statistical method or a machine learning model, so that the quality of data is ensured. For missing data, interpolation methods or other suitable filling strategies may be used to ensure the integrity of the data.
Data tagging: according to the processing result, the data is labeled as a target variable, such as an optimal tool path Y, so that the label corresponds to the corresponding feature.
Taking into account the cutting force F c The relation between the cutting speed V, the feed speed F, and the cutting depth D is expressed by using an empirical formula:
F c ＝k 1 ·V n ·F m ·D p
where k1 is an empirical coefficient, n, m, p are empirical indices, and this formula can be used as one of the features for training of the machine learning model.
Specifically, step 2 includes:
feature selection: according to the key influencing factors of the field knowledge and the machining performance, selecting the characteristics related to the machining of the numerical control machine tool, wherein the characteristics comprise cutting speed V, feeding speed F, cutting depth D, workpiece materials and the like, and selecting the characteristics which have obvious influence on the optimal tool path, so that the characteristics can comprehensively reflect the complexity of the machining process.
And (3) feature processing: selected features are normalized to ensure that they have similar dimensions under different dimensions. This helps to avoid problems in model training due to feature scale differences.
Using a standardized formula:
where x' is the normalized eigenvalue, x is the original eigenvalue, and μ and σ are the mean and standard deviation of the feature, respectively.
Feature transformation: for nonlinear relationships, consider transforming features, such as taking the logarithm, square root, etc., to better accommodate the needs of the machine learning model, e.g., logarithmically transforming the depth of cut:
D′＝log(D)
feature combination: combining some highly correlated features to form new features may enhance the fitting ability of the model to complex relationships, e.g., the ratio of cutting speed to feed speed as new features:
new feature = V/F
Feedback and iteration of feature engineering: and continuously adjusting the selection and processing modes of the features according to the performance feedback of the model. And the influence of the features on the model output is known by using the explanatory analysis of the model, so that the process of the feature engineering is further optimized.
Considering the relation between the cutting power P and the cutting speed V, the feeding speed F and the cutting depth D in the machining process, and using a formula after characteristic engineering to express:
P＝k 2 ·V′+k 3 ·F′+k 4 ·D′
wherein k is 2 、k 3 、k 4 Weights obtained for model training, V ', F ', D ', are normalized and transformed features.
Specifically, step 3 includes:
selecting a machine learning model: in view of the problem of optimizing the tool path, a machine learning model is selected that is suitable for the regression task. A support vector machine regression (Support Vector Regression, SVR) model is used in this embodiment.
Data set partitioning: the feature engineering processed data set is divided into training and testing sets, typically 80% of the data is used for training and 20% of the data is used for testing.
Model training: the support vector machine regression model is trained using the training set, and the goal of the SVR is to find a function that minimizes the difference between the predicted value and the true value.
The training process of SVR is represented by the following optimization problem:
the limiting conditions are as follows:
y i -w·Φ(x i )-b≤ε+ξ i
wherein C is a penalty parameter, y i Is the actual output value, w is the weight, Φ (x i ) Is input data x i Is the bias term, ε is the marginal width, ζ i Andis a relaxation variable, i is the sample index, and n is the number of samples.
Model verification: the trained model is validated using a test set, the performance of the model is assessed, and a mean square error (Mean Squared Error, MSE) indicator can be used to measure the prediction accuracy of the model.
Where n is the number of samples tested, y i Is a true value of the code,is a model predictive value.
The predictive function of the support vector machine regression model is:
wherein K (x i X) is a kernel function, alpha i Andis the lagrangian multiplier of the support vector machine and b is the bias term.
Model interpretation and tuning: parameters of the support vector machine regression model, such as penalty parameter C, kernel function selection, etc., are analyzed to adjust the complexity of the model, and the influence of the model on the tool path is known using an explanatory tool of the model, such as feature importance analysis.
Specifically, step 4 includes:
super parameter tuning: and performing super-parameter tuning on the support vector machine regression model by using methods such as Cross-Validation and the like so as to improve the generalization performance of the model. For SVR models, common hyper-parameters include penalty parameter C, selection of kernel functions, kernel function parameters, and the like. A Grid Search (Grid Search) or Random Search (Random Search) method may be used to find the optimal super-parameter combination.
Deep learning model introduction: consider the introduction of a deep learning model, such as a multi-layer perceptron (Multilayer Perceptron, MLP) neural network, to better capture complex nonlinear relationships, through which the deep learning model can learn abstract features in the data. Using a deep learning model, such as TensorFlow or PyTorch, a neural network model is built that contains multiple hidden layers, the MLP model outputs:
wherein X is an input feature, W (1) And W is (2) Is a weight matrix, b (1) And b (2) Is a bias term, σ is an activation function (e.g., reLU or Sigmoid).
Model performance comparison and selection: and performing performance comparison on the optimized support vector machine regression model and the deep learning model, evaluating generalization capability of the model by using a verification set, and selecting a model with better performance as a final numerical control machine tool optimization algorithm model.
Model interpretation and interpretation: for the selected final model, analysis of model interpretation is performed, the influence of the model on the tool path is known, and the prediction result of the model is interpreted by using an interpretation tool in a deep learning model, such as LIME (Local Interpretable Model-agnostic Explanations) and the like.
Specifically, step 5 includes:
sensor deployment and data acquisition: sensors, such as an acceleration sensor, a temperature sensor and the like, are reasonably arranged on the numerical control machine tool so as to monitor various data in the processing process in real time, and an appropriate sampling frequency is set to ensure that enough real-time data are acquired.
Real-time data processing: and processing the real-time data acquired by the sensor to ensure the quality and usability of the data. This may involve steps such as noise removal, outlier processing, etc., pre-processing the data to conform to the input requirements of the machine learning model.
Data input machine learning model: inputting real-time data into a model by using a trained machine learning model to obtain a real-time prediction result of the model, wherein the machine learning model predicts an optimal cutter path according to the real-time input data:
wherein,is a real-time prediction result, and the ML_model is a trained machine learning Model, X redl-time Is data input in real time.
The prediction result is applied to the control of the numerical control machine tool: the optimal cutter path predicted in real time by the machine learning model is applied to a control system of the numerical control machine tool, and control parameters of the numerical control machine tool, such as cutter speed, feeding speed and the like, are dynamically adjusted according to the output of the model so as to realize the optimal cutting effect.
Closed loop control and model update: the closed-loop control is implemented, that is, the input and the parameters of the model are continuously adjusted according to the actual processing feedback so as to further improve the self-adaptability and the robustness of the algorithm, a feedback control method can be adopted to compare the actual processing result with the model prediction result, the model is updated in a certain time interval according to the difference, and the model is trained by utilizing the latest real-time data so as to adapt to the change in the processing process.
Specifically, step 6 includes:
and (3) adjusting control parameters: according to the real-time prediction result of the machine learning model, the control parameters of the numerical control machine tool, such as the cutter speed V, the feeding speed F and the like, are dynamically adjusted, the model output is used as a reference signal by utilizing the self-adaptive control strategy, and the model output is compared with the actual control parameters to adjust the control parameters, and the control parameters can be realized through feedback control.
New control parameter = old control parameter + a× (target value-actual value)
Where α is the learning rate, used to control the step size of the adjustment.
Closed loop control is implemented: and comparing the real-time feedback with a model prediction result through a proportional-integral-derivative controller to form a closed-loop control system:
where u (t) is the controller output, e (t) is the error between the actual value and the target value, K p 、K i 、K d Respectively a proportional coefficient, an integral coefficient and a differential coefficient.
Model update and adaptive promotion: the machine learning model is continuously updated according to real-time feedback to improve the adaptability and the robustness of the algorithm, the model can be periodically trained on line by using new real-time data, and the model update can be realized by optimizing the algorithm such as gradient descent, so as to minimize the error between the actual value and the model predicted value:
where θ is a model parameter, J (θ) is a loss function, η is a learning rate,is the gradient of the loss function with respect to the parameter.
Through the self-adaptive control process, the numerical control machine tool can be flexibly adjusted according to real-time working condition changes, so that the optimal tool path and cutting effect are realized.
Finally, it should be noted that: the foregoing description is only a preferred embodiment of the present invention, and the present invention is not limited thereto, but may be modified or substituted for some of the technical features described in the foregoing embodiments by those skilled in the art, even though the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (7)
1. The numerical control machine tool optimization algorithm based on artificial intelligence and machine learning is characterized by comprising the following steps:
step 1: collecting data in the machining process of the numerical control machine tool, including cutting parameters, machine tool states and machining results, and preprocessing the data;
step 2: normalizing and normalizing the characteristics by using the cutting speed, the feeding speed, the cutting depth and the workpiece material characteristics;
step 3: predicting an optimal tool path using a machine learning model;
step 4: adjusting parameters of the model by using a cross verification method to improve the generalization performance of the model, and introducing a deep learning model to capture complex nonlinear relations;
step 5: real-time monitoring data in the processing process, including a cutter motion trail, vibration frequency and temperature, inputting the real-time data into a trained machine learning model, and predicting an optimal cutter path;
step 6: and dynamically adjusting control parameters of the numerical control machine according to the prediction result of the model so as to realize the optimal cutting effect.
2. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 1, wherein step 1 comprises:
the method comprises the steps that a sensor is arranged on a numerical control machine tool and used for monitoring data in the machining process in real time, wherein the data comprise cutting speed V, feeding speed F, cutting depth D, vibration Vib and temperature T, and a numerical control system is used for recording cutter paths, cutting parameters and machine tool state information;
integrating the data collected from different sensors and systems to form a data set, aligning time stamps, and ensuring the time consistency of each data source;
identifying and processing outliers to ensure data integrity;
according to the processing result, the data is labeled as a target variable, and the label is ensured to correspond to the corresponding characteristic;
taking into account the cutting force F c The relation between the cutting speed V, the feed speed F, and the cutting depth D is expressed by using an empirical formula:
F C ＝k 1 ·V n ·F m ·D p
wherein k1 is an empirical coefficient, and n, m and p are empirical indexes.
3. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 2, wherein step 2 comprises:
selecting a characteristic which has obvious influence on the optimal cutter path;
the selected features are normalized to ensure similar dimensions under different dimensions, the normalization formula is:
wherein x' is a normalized characteristic value, x is an original characteristic value, and μ and σ are the mean value and standard deviation of the characteristic respectively;
combining the features with strong correlation to form new features so as to enhance the fitting capability of the model to complex relations;
according to the performance feedback of the model, continuously adjusting the selection and processing modes of the features;
the influence of the features on the model output is known by utilizing the explanatory analysis of the model, so that the process of the feature engineering is further optimized;
considering the relation between the cutting power P and the cutting speed V, the feeding speed F and the cutting depth D in the machining process, and using a formula after characteristic engineering to express:
P＝k 2 ·V′+k 3 ·F′+k 4 ·D′
wherein k is 2 、k 3 、k 4 Weights obtained for model training, V ', F ', D ', are normalized and transformed features.
4. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 3, wherein step 3 comprises:
using a support vector machine regression model;
dividing the data set processed by the feature engineering into a training set and a testing set;
training a support vector machine regression model by using a training set;
the training process of the support vector machine regression model is represented by the following optimization problem:
the limiting conditions are as follows:
y i -w·Φ(x i )-b≤ε+ξ i
wherein C is a penalty parameter, y i Is the actual output value, w is the weight, Φ (x i ) Is input data x i Is the bias term, ε is the marginal width, ζ i Andis a relaxation variable, i is a sample index, n is the number of samples;
and verifying the trained model by using a test set, evaluating the performance of the model, and measuring the prediction accuracy of the model by using a mean square error index:
where n is the number of samples tested, y i Is a true value of the code,is a model predictive value;
the predictive function of the support vector machine regression model is:
wherein K (x i X) is a kernel function, alpha i Andis the Lagrangian multiplier of the support vector machine, b is the bias term;
and analyzing parameters of the regression model of the support vector machine to adjust the complexity of the model.
5. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 4, wherein step 4 comprises:
performing super-parameter tuning on the support vector machine regression model by using a cross verification method so as to improve the generalization performance of the model;
introducing a multi-layer perceptron neural network to capture complex nonlinear relations, and building a neural network model comprising a plurality of hidden layers by using a deep learning model;
the multi-layer perceptron neural network output is expressed as:
wherein X is an input feature, W (1) And W is (2) Is a weight matrix, b (1) And b (3) Is the bias term, σ is the activationA function;
performing performance comparison on the optimized support vector machine regression model and the deep learning model, and evaluating generalization capability of the support vector machine regression model by using a verification set;
selecting a model with better performance as a final numerical control machine tool optimization algorithm model;
for the selected final model, carrying out model explanatory analysis to know the influence of the model on the cutter path;
and interpreting the prediction result of the model by using an interpretability tool in the deep learning model.
6. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 5, wherein step 5 comprises:
the sensors are reasonably arranged on the numerical control machine tool so as to monitor various data in the processing process in real time, and proper sampling frequency is set to ensure that enough real-time data are acquired;
real-time data acquired by the sensor are processed, so that the quality and usability of the data are ensured;
inputting real-time data into a model by using a trained machine learning model to obtain a real-time prediction result of the model, and predicting an optimal cutter path by the machine learning model according to the real-time input data:
wherein,is a real-time prediction result, and the ML_model is a trained machine learning Model, X real-time Is data input in real time;
applying the optimal tool path predicted in real time by the machine learning model to a control system of the numerical control machine tool, and dynamically adjusting control parameters of the numerical control machine tool to realize an optimal cutting effect;
the input and parameters of the model are continuously adjusted according to actual processing feedback, so that the adaptability and the robustness of the algorithm are further improved:
new parameter = old parameter + alpha× (target value-actual value)
Wherein alpha is the learning rate and is used for controlling the step length of adjustment;
and updating the model in a certain time interval, and training the model by utilizing the latest real-time data so as to adapt to the change in the processing process.
7. The numerically controlled machine tool optimization algorithm based on artificial intelligence and machine learning of claim 6, wherein step 6 comprises:
according to the real-time prediction result of the machine learning model, dynamically adjusting the control parameters of the numerical control machine tool, and comparing the model output serving as a reference signal with the actual control parameters by utilizing the self-adaptive control strategy to adjust the control parameters:
new control parameter = old control parameter + a× (target value-actual value)
Wherein alpha is the learning rate and is used for controlling the step length of adjustment;
and comparing the real-time feedback with a model prediction result through a proportional-integral-derivative controller to form a closed-loop control system:
where u (t) is the controller output, e (t) is the error between the actual value and the target value, K p 、K i 、K d Proportional, integral and differential coefficients, respectively;
implemented by a gradient descent optimization algorithm to minimize the error between the actual value and the model predicted value:
where θ is a model parameterJ (θ) is a loss function, η is a learning rate,is the gradient of the loss function with respect to the parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410004177.2A CN117826599A (en) | 2024-01-02 | 2024-01-02 | Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410004177.2A CN117826599A (en) | 2024-01-02 | 2024-01-02 | Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117826599A true CN117826599A (en) | 2024-04-05 |
Family
ID=90507650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410004177.2A Pending CN117826599A (en) | 2024-01-02 | 2024-01-02 | Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117826599A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118153237A (en) * | 2024-05-09 | 2024-06-07 | 广州三拓智能装备有限公司 | All-dimensional numerical control six-sided drill precision machining system |
-
2024
- 2024-01-02 CN CN202410004177.2A patent/CN117826599A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118153237A (en) * | 2024-05-09 | 2024-06-07 | 广州三拓智能装备有限公司 | All-dimensional numerical control six-sided drill precision machining system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Javed et al. | Tool wear monitoring and prognostics challenges: a comparison of connectionist methods toward an adaptive ensemble model | |
| Wong et al. | Technical data-driven tool condition monitoring challenges for CNC milling: a review | |
| Zheng et al. | Long short-term memory network for remaining useful life estimation | |
| Liao et al. | Tool wear state recognition based on GWO–SVM with feature selection of genetic algorithm | |
| Wang et al. | Enhanced particle filter for tool wear prediction | |
| Zhou et al. | A model for real-time failure prognosis based on hidden Markov model and belief rule base | |
| Lee et al. | On-line surface roughness recognition system using artificial neural networks system in turning operations | |
| CN117826599A (en) | Numerical control machine tool optimization algorithm based on artificial intelligence and machine learning | |
| Liu et al. | An approach for tool wear prediction using customized DenseNet and GRU integrated model based on multi-sensor feature fusion | |
| Pratama et al. | Metacognitive learning approach for online tool condition monitoring | |
| Ntemi et al. | Infrastructure monitoring and quality diagnosis in CNC machining: A review | |
| Wang et al. | Meta domain generalization for smart manufacturing: Tool wear prediction with small data | |
| Lermer et al. | Creation of digital twins by combining fuzzy rules with artificial neural networks | |
| CN114218872A (en) | Method for predicting remaining service life based on DBN-LSTM semi-supervised joint model | |
| Zhu et al. | An unsupervised dual-regression domain adversarial adaption network for tool wear prediction in multi-working conditions | |
| Masalimov et al. | Diagnostics of the tool condition in metal-cutting machines by means of recurrent neural networks | |
| Shi et al. | Tool life prediction of dicing saw based on PSO-BP neural network | |
| Jeang | Reliable tool replacement policy for quality and cost | |
| Nguyen et al. | Advanced model-based controller for cyber-physical shot peening process | |
| Abid et al. | Adaptive machine learning approach for fault prognostics based on normal conditions-application to shaft bearings of wind turbine | |
| Liman et al. | Modeling and prediction of surface roughness in ultra-high precision diamond turning of contact lens polymer using RSM and ANN methods | |
| Ferguson et al. | A generalized method for featurization of manufacturing signals, with application to tool condition monitoring | |
| CN116880201A (en) | Water network channel state control system based on fuzzy neural network | |
| CN116401545A (en) | Multimode fusion type turbine runout analysis method | |
| Hoang et al. | Self-adjusting on-line cutting condition for high-speed milling process |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |