Modeling method and system for morphology model of milling surface

Abstract

The invention relates to the field of milling, in particular to a modeling method and system for a morphology model of a milling surface, and the method comprises the steps of data collection, data preprocessing, feature extraction, model construction and model verification and optimization. According to the modeling method and system, various actual factors in milling are fully considered, including tool abrasion, cutting force vibration, workpiece material non-uniformity and the like, the fitting degree of the model and the actual milling surface appearance is improved, and the microscopic geometrical characteristics of the machining surface can be reflected more accurately; a machine learning algorithm is adopted to construct a model, so that a complex milling process parameter combination relationship can be flexibly processed, the applicability and accuracy of the model under different milling process parameters are improved, and accurate evaluation of part performance and optimization of a machining process are facilitated; the system is complete in structure, all the modules are clear in division of labor, and the construction, verification and optimization process of the morphology model of the milling surface can be efficiently achieved.

Description

Technical Field

[0001] The invention relates to the field of milling processing, and in particular to a method and system for modeling a topography model of a milling processing surface. Background Art

[0002] Milling is a widely used processing method in mechanical manufacturing. The morphology of the milled surface has a crucial impact on the performance of parts, such as wear resistance, sealing, contact stiffness, etc. However, there are still many challenges in the prediction and accurate modeling of the machined surface morphology in milling.

[0003] Traditional milling surface topography modeling methods are often based on simplified geometric assumptions and ignore many factors in actual processing. For example, factors such as tool wear, vibration caused by cutting force, and unevenness of workpiece materials are not fully considered in some existing modeling methods. This leads to a large deviation between the established model and the actual milling surface topography, and cannot accurately reflect the micro-geometric characteristics of the machined surface, thus affecting the accurate evaluation of part performance and the optimization of the processing technology.

[0004] In addition, existing modeling methods lack flexibility in dealing with complex combinations of milling process parameters. The effects of different milling parameters, such as cutting speed, feed rate, cutting depth, etc., on surface morphology are interrelated, but traditional methods are difficult to accurately describe this complex relationship, which greatly reduces the applicability and accuracy of the model when facing multiple milling process parameter changes.

[0005] Therefore, those skilled in the art provide a method and system for modeling a topography model of a milling surface to solve the problems raised in the above background technology. Summary of the invention

[0006] In order to solve the above technical problems, the present invention provides a method and system for modeling a topography model of a milling surface.

[0007] Modeling method of topography model of milling surface

[0008] Data collection: collecting parameter data during milling and surface data after milling;

[0009] Data preprocessing: cleaning and removing outliers from various parameter data collected, normalizing data from different sources and converting them to the same numerical range to facilitate subsequent data analysis and model building, interpolating data obtained by the 3D surface topography measuring instrument, supplementing data points that may be missing due to the distance between measurement points, and making the surface topography data more complete and continuous;

[0010] Feature extraction, extracting features related to surface morphology from preprocessed parameter data;

[0011] Model construction, using machine learning algorithms to build a morphological model of the milling surface;

[0012] Model verification and optimization: Use an independent test data set to verify the constructed model. The test data set is divided from the collected data into 80% for training and 20% for testing.

[0013] Preferably, the parameter data include tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters, and the tool parameters include tool radius, tool rake angle, and tool back angle; the process parameters include cutting speed, feed rate, and cutting depth; the machine tool status parameters include machine tool vibration frequency and vibration amplitude; and the process material characteristic parameters include hardness and elastic modulus of the material to be processed.

[0014] Preferably, the tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters are obtained by means of sensors installed on the milling processing equipment and by querying a workpiece material manual.

[0015] Preferably, a three-dimensional surface topography measuring instrument is used to measure the surface after milling to obtain actual data points of the surface microscopic topography. The measurement accuracy reaches the micron level and the measurement range covers the entire machined surface to obtain comprehensive and accurate surface topography information.

[0016] Preferably, unreasonable force values ​​and vibration values ​​generated by short-term interference and collected by the sensor are identified and eliminated, and tool parameters and process parameters are converted into values ​​in the interval [0,1].

[0017] Preferably, the feature extraction is as follows: for tool parameters, extracting the influence characteristics of tool geometry on cutting edge trajectory; for process parameters, extracting the influence characteristics of cutting force and cutting heat factors on material removal; for machine tool state parameters, extracting the influence characteristics of vibration on cutting edge position; for workpiece material characteristic parameters, extracting the influence characteristics of material machinability on surface roughness.

[0018] Preferably, the surface roughness, waviness and texture direction features are extracted from the surface topography measurement data as output target features of the model.

[0019] Preferably, a neural network algorithm is used to construct a network structure of an input layer, a hidden layer and an output layer, and the extracted tool parameters, process parameters, machine tool state parameters and workpiece material characteristic parameters are used as neuron inputs of the input layer, and the extracted machining surface roughness, waviness and texture direction are used as neuron outputs of the output layer.

[0020] Preferably, the neural network is trained by a back-propagation algorithm, and the weights and thresholds in the network are adjusted so that the output of the model is close to the actual measured surface morphological features. During the training process, a reasonable loss function, such as the mean square error (MSE) function, is set to evaluate the difference between the model output and the actual result, and the network parameters are adjusted according to the difference.

[0021] Preferably, the model is optimized according to the verification results. If the prediction error of the model on the test data set exceeds the set threshold, the structure of the model is adjusted, such as increasing the number of neurons in the hidden layer or adjusting the activation function of the neural network (such as from the Sigmoid function to the ReLU function), and then retraining and verification are performed until the prediction error of the model meets the requirements.

[0022] Milling surface topography modeling system

[0023] The data acquisition module includes multiple sensor interfaces for connecting sensors installed on the milling processing equipment to collect tool parameters, process parameters, and machine tool status parameter data. It has an interface connected to the workpiece material database to obtain the workpiece material characteristic parameters and is connected to a three-dimensional surface morphology measuring instrument to obtain actual data points of the surface micromorphology after milling.

[0024] Data preprocessing module

[0025] The data cleaning submodule is used to identify and remove abnormal values ​​in the collected parameter data.

[0026] The data normalization submodule is used to normalize data from different sources.

[0027] The data interpolation submodule is used to perform interpolation processing on the data obtained by the three-dimensional surface topography measuring instrument.

[0028] Feature extraction module

[0029] The input parameter feature extraction submodule is used to extract features related to surface morphology from the preprocessed tool parameters, process parameters, machine tool status parameters and workpiece material characteristic parameters.

[0030] The output target feature extraction submodule is used to extract surface roughness, waviness, and texture direction features from surface topography measurement data.

[0031] Model building module

[0032] A neural network is used to construct the sub-module, and a machine learning algorithm is used to build the morphology model of the milling surface. The neural network algorithm is selected to construct the network structure of the input layer, hidden layer and output layer, and the relevant parameters are used as input and output.

[0033] The model training submodule trains the neural network through the back-propagation algorithm, sets a reasonable loss function and adjusts the network parameters according to the differences.

[0034] Model validation and optimization module

[0035] The model verification submodule uses an independent test data set to verify the constructed model.

[0036] The model optimization submodule optimizes the model according to the verification results, such as adjusting the model structure or network parameters.

[0037] Technical effects and advantages of the present invention:

[0038] 1. The modeling method and system of the present invention fully consider various practical factors in milling processing, including tool wear, cutting force vibration, workpiece material unevenness, etc., improve the fit between the model and the actual milling surface morphology, and can more accurately reflect the micro-geometric characteristics of the machined surface.

[0039] 2. The model is constructed using machine learning algorithms, which can flexibly handle complex combinations of milling process parameters, improve the applicability and accuracy of the model under different milling process parameters, and help to accurately evaluate part performance and optimize the processing technology.

[0040] 3. The system structure of the present invention is complete, and each module has a clear division of labor, which can efficiently realize the construction, verification and optimization process of the milling surface morphology model. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic flow chart of a method for modeling a topography model of a milling machined surface provided in an embodiment of the present application;

[0042] Figure 2 It is a flow chart of a system for modeling a topography model of a milling surface provided in an embodiment of the present application; DETAILED DESCRIPTION

[0043] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments of the present invention are provided for the purpose of illustration and description, and are not intended to be exhaustive or to limit the present invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention, and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

[0044] Example 1

[0045] See also Figures 1-2In this embodiment, a method for modeling a morphology model of a milling surface is provided, including data acquisition, data preprocessing, feature extraction, model construction, and model verification and optimization.

[0046] Data collection:

[0047] Collect parameter data during milling and surface data after milling;

[0048] The parameter data includes tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters, and the tool parameters include tool radius, tool rake angle, and tool clearance angle, the process parameters include cutting speed, feed rate, and cutting depth, the machine tool status parameters include machine tool vibration frequency and vibration amplitude, and the process material characteristic parameters include hardness and elastic modulus of the material to be processed;

[0049] The tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters are obtained by means of sensors installed on the milling processing equipment and by querying the workpiece material manual;

[0050] The three-dimensional surface topography measuring instrument is used to measure the surface after milling to obtain the actual data points of the surface microscopic topography. The measurement accuracy reaches the micron level and the measurement range covers the entire processed surface to obtain comprehensive and accurate surface topography information.

[0051] Data preprocessing:

[0052] Clean the collected parameter data, remove outliers, normalize data from different sources, and convert them to the same numerical range to facilitate subsequent data analysis and model building. Interpolate the data obtained by the 3D surface topography measuring instrument to supplement the data points that may be missing due to the distance between the measurement points, so as to make the surface topography data more complete and continuous.

[0053] The unreasonable force and vibration values ​​collected by the sensor due to short-term interference are identified and eliminated, and the tool parameters and process parameters are converted into values ​​in the range of [0,1].

[0054] Feature extraction:

[0055] Extracting features related to surface morphology from the preprocessed parameter data;

[0056] For tool parameters, the influence characteristics of tool geometry on cutting edge trajectory are extracted; for process parameters, the influence characteristics of cutting force and cutting heat factors on material removal are extracted; for machine tool state parameters, the influence characteristics of vibration on cutting edge position are extracted; for workpiece material characteristic parameters, the influence characteristics of material machinability on surface roughness are extracted.

[0057] Model construction:

[0058] Use machine learning algorithms to build a topography model of the milling surface;

[0059] The surface roughness, waviness and texture direction features are extracted from the surface topography measurement data as output target features of the model.

[0060] Model validation and optimization:

[0061] The constructed model is verified using an independent test data set, which is divided from the collected data into 80% for training and 20% for testing.

[0062] A neural network algorithm is used to construct a network structure of input layer, hidden layer and output layer. The extracted tool parameters, process parameters, machine tool status parameters and workpiece material characteristic parameters are used as neuron inputs of the input layer. The extracted machined surface roughness, waviness and texture direction are used as neuron outputs of the output layer. The neural network is trained by the back propagation algorithm, and the weights and thresholds in the network are adjusted to make the output of the model close to the actual measured surface morphology features. During the training process, a reasonable loss function, such as the mean square error (MSE) function, is set to evaluate the difference between the model output and the actual result, and the network parameters are adjusted according to the difference.

[0063] The model is optimized according to the verification results. If the prediction error of the model on the test data set exceeds the set threshold, the structure of the model is adjusted, such as increasing the number of neurons in the hidden layer or adjusting the activation function of the neural network (such as changing from the Sigmoid function to the ReLU function), and then retraining and verification are performed until the prediction error of the model meets the requirements.

[0064] Milling surface topography modeling system

[0065] The data acquisition module includes multiple sensor interfaces for connecting sensors installed on the milling processing equipment to collect tool parameters, process parameters, and machine tool status parameter data. It has an interface connected to the workpiece material database to obtain the workpiece material characteristic parameters and is connected to a three-dimensional surface morphology measuring instrument to obtain actual data points of the surface micromorphology after milling.

[0066] Data preprocessing module

[0067] The data cleaning submodule is used to identify and remove abnormal values ​​in the collected parameter data.

[0068] The data normalization submodule is used to normalize data from different sources.

[0069] The data interpolation submodule is used to perform interpolation processing on the data obtained by the three-dimensional surface topography measuring instrument.

[0070] Feature extraction module

[0071] The input parameter feature extraction submodule is used to extract features related to surface morphology from the preprocessed tool parameters, process parameters, machine tool status parameters and workpiece material characteristic parameters.

[0072] The output target feature extraction submodule is used to extract surface roughness, waviness, and texture direction features from surface topography measurement data.

[0073] Model building module

[0074] A neural network is used to construct the sub-module, and a machine learning algorithm is used to build the morphology model of the milling surface. The neural network algorithm is selected to construct the network structure of the input layer, hidden layer and output layer, and the relevant parameters are used as input and output.

[0075] The model training submodule trains the neural network through the back-propagation algorithm, sets a reasonable loss function and adjusts the network parameters according to the differences.

[0076] Model validation and optimization module

[0077] The model verification submodule uses an independent test data set to verify the constructed model.

[0078] The model optimization submodule optimizes the model according to the verification results, such as adjusting the model structure or network parameters.

[0079] Embodiment 2: Method and system for modeling the topography model of a milling surface

[0080] Including data collection

[0081] Install force sensors, vibration sensors and other sensors on the milling equipment and connect them to the data acquisition module. Import the workpiece material characteristic data from the workpiece material database into the data acquisition module. After the milling process is completed, use a 3D surface topography measuring instrument to measure the machined surface and transmit the measurement data to the data acquisition module.

[0082] Data preprocessing

[0083] In the data preprocessing module, the data cleaning submodule first checks and removes outliers from the collected sensor data and other parameter data. For example, for the cutting force data collected by the force sensor, if the value of a data point is much higher than the normal cutting force range, it will be considered an outlier and deleted.

[0084] The data normalization submodule normalizes the tool parameters, process parameters, machine tool status parameters and workpiece material characteristic parameters after cleaning. For example, for the process parameter of cutting speed, if its value range is [100m / min,500m / min], it is mapped to the interval [0,1].

[0085] The data interpolation submodule interpolates the data obtained by the 3D surface topography measuring instrument. If the measurement point spacing is 1mm, the interpolation algorithm is used to supplement the missing data points between the spacings to make the surface topography data more complete and continuous.

[0086] Feature extraction

[0087] In the feature extraction module, the input parameter feature extraction submodule extracts relevant features from various preprocessed parameter data. For example, for the tool parameter of tool radius, its influence feature on the cutting edge trajectory on the machined surface is calculated.

[0088] The output target feature extraction submodule extracts surface roughness, waviness, texture direction and other features from the surface topography measurement data. For example, the roughness is determined by calculating the statistical parameters of the surface height data, and the waviness and texture direction are determined by analyzing the periodicity of the surface profile.

[0089] Model building

[0090] In the model building module, a three-layer multilayer perceptron neural network is constructed using the neural network building submodule. The tool parameters, process parameters, machine tool status parameters, and workpiece material characteristic parameters are used as input neurons in the input layer, and the input layer has 10 neurons. Two hidden layers are set, each with 20 neurons, and the activation function uses the ReLU function. The surface roughness, waviness, texture direction, etc. are used as output neurons in the output layer, and the output layer has 3 neurons.

[0091] The model training submodule trains the neural network through the back propagation algorithm. The mean square error (MSE) is set as the loss function, and the initial learning rate is 0.01. During the training process, the weights and thresholds in the network are adjusted according to the value of MSE. After 1000 iterations of training, the output of the model is as close as possible to the actual measured surface morphology.

[0092] Model validation and optimization

[0093] In the model validation and optimization module, the model validation submodule uses 20% of the independent test data set to validate the constructed model and calculates the mean square error (MSE) of the model on the test data set.

[0094] If the MSE exceeds 0.01, the model optimization submodule optimizes the model. For example, the number of neurons in the hidden layer is increased to 30, or the activation function is adjusted to the LeakyReLU function, and then retraining and verification are performed until the MSE meets the requirements.

[0095] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field and related fields without creative work should fall within the scope of protection of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention are implemented according to the conventional means in the field unless otherwise specified and limited.

Claims

1. A method for modeling a topography model of a milling surface, characterized in that: The following steps are involved: S1, data collection, collecting parameter data during milling and surface data after milling; S2. Data preprocessing: cleaning the collected parameter data, removing outliers, normalizing the data from different sources, converting them to the same numerical range, so as to facilitate subsequent data analysis and model building, interpolating the data obtained by the 3D surface topography measuring instrument, supplementing the data points that may be missing due to the distance between the measurement points, and making the surface topography data more complete and continuous; S3, feature extraction, extracting features related to surface morphology from the preprocessed parameter data; S4, model construction, using machine learning algorithm to build the morphology model of the milling surface; S5. Model verification and optimization: Use an independent test data set to verify the constructed model. The test data set is divided from the collected data into 80% for training and 20% for testing.

2. The method for modeling the topography model of a milling surface according to claim 1, characterized in that: The parameter data in S1 include tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters, and the tool parameters include tool radius, tool front angle, and tool back angle, the process parameters include cutting speed, feed rate, and cutting depth, the machine tool status parameters include machine tool vibration frequency and vibration amplitude, and the process material characteristic parameters include hardness and elastic modulus of the material to be processed, and the tool parameters, process parameters, machine tool status parameters, and process material characteristic parameters are obtained by sensors installed on the milling processing equipment and by querying the workpiece material manual.

3. The method for modeling the topography model of a milling surface according to claim 1, characterized in that: The three-dimensional surface topography measuring instrument is used to measure the surface after milling to obtain the actual data points of the surface microscopic topography. The measurement accuracy reaches the micron level and the measurement range covers the entire processed surface to obtain comprehensive and accurate surface topography information.

4. The method and system for modeling the topography model of a milling surface according to claim 2, characterized in that: The unreasonable force and vibration values ​​collected by the sensor due to short-term interference are identified and eliminated, and the tool parameters and process parameters are converted into values ​​in the range of [0,1].

5. The method and system for modeling the topography model of a milling surface according to claim 1, characterized in that: The feature extraction in S3 is as follows: for tool parameters, extract the influence characteristics of tool geometry on cutting edge trajectory; for process parameters, extract the influence characteristics of cutting force and cutting heat factors on material removal; for machine tool state parameters, extract the influence characteristics of vibration on cutting edge position; for workpiece material characteristic parameters, extract the influence characteristics of material machinability on surface roughness.

6. The method and system for modeling the topography model of a milling surface according to claim 1, characterized in that: The model in S4 is constructed as follows: the surface roughness, waviness and texture direction features are extracted from the surface topography measurement data as output target features of the model.

7. The method and system for modeling the topography model of a milling surface according to claim 1, characterized in that: A neural network algorithm is used to construct a network structure of input layer, hidden layer and output layer. The extracted tool parameters, process parameters, machine tool state parameters and workpiece material characteristic parameters are used as neuron inputs of the input layer, and the extracted machining surface roughness, waviness and texture direction are used as neuron outputs of the output layer.

8. The method and system for modeling the topography model of a milling surface according to claim 1, characterized in that: The model is optimized based on the verification results. If the prediction error of the model on the test data set exceeds the set threshold, the structure of the model is adjusted.

9. A milling surface topography modeling system, characterized in that: Data acquisition module: including multiple sensor interfaces for connecting sensors installed on milling equipment to collect tool parameters, process parameters, and machine tool status parameter data. It has an interface for connecting to the workpiece material database to obtain workpiece material characteristic parameters and connect to a three-dimensional surface topography measuring instrument to obtain actual data points of the surface micro-topography after milling. Data preprocessing module: including a data cleaning submodule, a data normalization submodule and a data interpolation submodule, wherein the data cleaning submodule is used to identify and remove abnormal values ​​in the collected parameter data, the data normalization submodule is used to normalize data from different sources, and the data interpolation submodule is used to interpolate data acquired by the three-dimensional surface topography measuring instrument; Feature extraction module: including an input parameter feature extraction submodule and an output target feature extraction submodule, wherein the input parameter feature extraction submodule is used to extract features related to surface morphology from preprocessed tool parameters, process parameters, machine tool state parameters and workpiece material characteristic parameters, and the output target feature extraction submodule is used to extract surface roughness, waviness and texture direction features from surface morphology measurement data; Model construction module: including a neural network construction submodule and a model training submodule. The neural network construction submodule uses a machine learning algorithm to construct a morphology model of the milling surface, selects a neural network algorithm to construct a network structure of an input layer, a hidden layer, and an output layer, and uses relevant parameters as input and output; the model training submodule trains the neural network through a back propagation algorithm, sets a reasonable loss function, and adjusts the network parameters according to the difference; Model verification and optimization module: includes a model verification submodule and a model optimization submodule. The model verification submodule verifies the constructed model using an independent test data set, and the model optimization submodule optimizes the model according to the verification result.

Images