Three-axis Vertical Machining Center Research Articles

A Novel Data Augmentation Method Based on CoralGAN for Prediction of Part Surface Roughness.

Deep learning networks can be applied to the field of intelligent prediction of part surface roughness. However, the surface roughness samples of parts have the problems of high collection cost, unbalanced categories, and complicated data distribution, which inevitably limit the application of deep learning network models in the field of intelligent prediction of part surface roughness. To solve these problems, this article proposes a novel data augmentation method based on CoralGAN for prediction of part surface roughness, which introduces the domain adaptive method deep coral function to help optimize the network parameters of the generator of generative adversarial network (GAN). Specifically, the vibration signal collected during processing is converted into frequency spectrum data and input into CoralGAN. The training of the generator is guided by coral loss, that is, the distance between the covariances of the real samples and generated samples features, not just the statistical consistency of the traditional GAN. Experiments have been carried out on the three-axis vertical machining center. Research shows that the proposed method can improve the prediction accuracy of part surface roughness to 95.5%.

A novel online tool condition monitoring method for milling titanium alloy with consideration of tool wear law

Due to issues such as limited variability in monitoring data across different tool wear conditions and interference during the machining process, data-driven monitoring models are susceptible to misclassification. Therefore, this paper proposes a pioneering approach that takes into account the tool wear law and the characteristic distribution of tool wear monitoring data. Specifically, the paper proposes an automatic feature extraction and tool condition monitoring method based on Siamese Long Short-term Memory Networks (SLSTMs) to transform the original data distribution, enhance the differentiation of different tool wear condition monitoring data, and achieve accurate prediction of tool wear condition. Additionally, a hybrid data and mechanism-driven tool wear monitoring method is proposed that takes advantage of the fact that tool wear is continuous and irreversible to enable reliable adjustment of the monitoring results of the data-driven model without human intervention. The study conducted milling experiments on a three-axis vertical machining center, using TA2 titanium alloy as the workpiece material. Vibration signals from the spindle in three directions were collected as input to the network, with tool wear state labeled as “0″ or ”1″ as the output. Classical machine learning algorithms such as Decision Trees, K-Nearest Neighbors (KNN) and Support Vector Machines (SVM), as well as classical deep learning algorithms such as Convolutional Neural Networks (CNN), Sparse Stacked Autoencoders (SSAE) and Long Short-Term Memory Neural Networks (LSTM), were used to construct and compare feature extraction and state monitoring models. The proposed model based on Siamese Long Short-Term Memory Neural Networks (SLSTMs) achieved testing average accuracy of 98.2% (σ2 = 0.44), outperforming typical deep learning algorithms such as CNN, LSTM and SSAE as well as traditional machine learning algorithms such as Decision Trees, KNN and SVM in terms of accuracy and robustness. Moreover, the proposed data and mechanism hybrid-driven tool wear monitoring method can effectively improve the monitoring accuracy and robustness of data-driven models such as CNN and SSAE. The monitoring accuracy of CNN and SSAE increased from 95.9% to 97.6% and 91.0% to 94.3%, respectively.

The modeling method of machine tool geometry error based on Bryan principle

The error caused by measurement itself directly affects the accuracy of models. The error caused by the failure to follow the Bryan principle in the process of measuring the volumetric error of the machine tool makes the geometric accuracy of the machine tool after compensation less than the ideal effect. In this paper, it is proposed that the measurement error will be caused by the non coincidence of the measurement point and the tool position point in the measurement of linear axis error. According to Bryan principle, the relationship between the center point of the reflective mirror of the laser interferometer and the tool position point is analyzed. The straightness error produced in the measurement process is introduced into the volumetric error of the machine tool, and the geometric model containing the measurement error is established to improve the modeling accuracy of the machine tool. To prove the effectiveness of compensating measurement errors, an experiment has been conducted on a three-axis vertical machining center. The experimental results show that the geometric accuracy of the machine tool is improved by 23% after the measurement error is compensated.

Validation of the Fabricated Cast Nylon Head Phantom for Stereotactic Radiosurgery End-to-End Test using Alanine Dosimeter.

Stereotactic radiosurgery (SRS) is an alternative to surgery as it precisely delivers single-large doses to small tumors. Cast nylon is used in phantom due to its computed tomography (CT) number of about 56-95 HU, which is close to that of the soft tissue. Moreover, cast nylon is also more budget-friendly than the commercial phantoms. The aim of this study is to design and validate the fabricated cast nylon head phantom for SRS end-to-end test using an alanine dosimeter. The phantom was designed using cast nylon. It was initially created by a computer numerical control three-axis vertical machining center. Then, the cast nylon phantom was scanned using a CT simulator. Finally, the validation of the fabricated phantom using alanine dosimeter proficiency with four Varian LINAC machines was performed. The fabricated phantom presented a CT number of 85-90 HU. The outcomes of VMAT SRS plans showed percentage dose differences from 0.24 to 1.55, whereas the percentage dose differences in organ at risk (OAR) were 0.09-10.80 due to the low-dose region. The distance between the target (position 2) and the brainstem (position 3) was 0.88 cm. Variation in dose for OAR is higher, which might be due to a high-dose gradient in the area where measurement was being conducted. The fabricated cast nylon end-to-end test head phantom had been suitably designed to image and irradiate during an end-to-end test for SRS using an alanine dosimeter.

A New Multitask Learning Method for Tool Wear Condition and Part Surface Quality Prediction

Deep learning has been gradually used in the field of machining condition monitoring. However, at present only single-task prediction can be performed, which results in increased experimental costs, wasted datasets, and repetitive work. In this article, a new multitask learning method based on a deep belief network (DBN) is proposed, which can be used to predict the tool wear condition and part surface quality. The single-task data transmission of the last few hidden layers of the DBN network is improved to multitask parallel data transmission so that the improved DBN can realize multitask learning. The loss function of the multitask learning model is defined as the weighted sum of all single-task loss functions. According to the loss of different tasks in the iteration process, the weight of corresponding tasks can be adjusted automatically. Furthermore, the multitask deep learning method can realize information sharing, suppress overfitting, improve prediction accuracy, and require less computing time. Combined with the abovementioned improvements, a multitask model for tool wear and part surface quality was developed. Experimental verification was performed on a KVC850M three-axis vertical machining center. The results show that the accuracy of the proposed multitask prediction model is 99% for the tool wear prediction and 92.86% for part surface quality prediction.

Assembly quality evaluation for linear axis of machine tool using data-driven modeling approach

During the batch assembly analysis of linear axis of machine tool, assembly quality evaluation is crucial to reduce assembly quality fluctuations and improve efficiency. This study presented a data-driven modeling approach for evaluating assembly quality of linear axis based on normalized mutual information and random sampling with replacement (NMI-RSWR) variable selection method, synthetic minority over-sampling technique (SMOTE), and genetic algorithm (GA)-optimized multi-class support vector machine (SVM). First, a variable selection method named NMI-RSWR was proposed to select key assembly parameters which affected assembly quality of linear axis. Then, a hybrid method based on SMOTE and GA-optimized multi-class SVM was presented to construct assembly quality evaluation model. In this method, Class imbalance problem was solved by using SMOTE, and parameters optimization problem was solved by using GA. Finally, the assembly-related data from the batch assembly of x-axis of a three-axis vertical machining center were collected to validate the proposed method. The results indicate that the proposed NMI-RSWR approach has capacity for selecting the highly related assembly parameters with assembly quality of linear axis, and the proposed data-driven modeling approach is effective for assembly quality evaluation of linear axis.

基于i5OS数控系统的三轴立式加工中心误差建模及补偿研究

基于i5OS数控系统的三轴立式加工中心误差建模及补偿研究

Assembly consistency improvement of straightness error of the linear axis based on the consistency degree and GA-MSVM-I-KM

Fluctuation on the assembly quality of the linear axis of machine tools (LA-MT) at the same batch is urgent problem need to be solved in assembly of machine tools. In this paper, a new concept of assembly consistency degree was introduced for defining the fluctuation degree of assembly quality. Based on assembly consistency degree, a hybrid machine learning method, genetic algorithm optimized multi-class support vector machine and improved Kuhn–Munkres (GA-MSVM-I-KM) was proposed for improving assembly consistency of LA-MT. The assembly of linear axis of a three-axis vertical machining center was regarded as an example, and the assembly consistency influence factors on straightness error of Y-axis (SE-YA) were analyzed through the Kruskal–Wallis statistical method. The main factors affected on the assembly consistency of SE-YA turned out to be the machining errors of bed and the assembly team technical levels. Based on this, the assembly consistency improvement model was established. Then, the prediction model of SE-YA based on assembly experiment data and genetic algorithm optimized multi-class support vector machine (GA-MSVM) was constructed, and I-KM method was applied for improving assembly consistency of SE-YA. The results show that the GA-MSVM-I-KM method can effectively enhance the assembly consistency of SE-YA, and the assembly consistency degree is reduced from 0.19 to 0.08.

Prediction of cutting force in milling process using vibration signals of machine tool

This paper presents a method based on the theory of Kalman filter (KF) to predict cutting force through vibration signals in milling process. The acceleration signal and displacement signal of the machine tool spindle vibration are acquired and used for the prediction of the cutting force. In actual machining process, different cutting parameters lead to different time and frequency domain characteristics of cutting force. As system output, the vibration of machine tool is influenced by the input signal of force and machine tool dynamic characteristics. The bandwidth of the transfer function between the cutting force and acceleration should be considered when adopting indirect cutting force measurement method by using vibration sensors. Because of the influence of structural modes of the spindle-tool system, the measurement bandwidth is limited and as a result, it leads to the distortion of measurements containing high frequencies. Consequently, it is necessary to compensate the disturbances caused by structural modes and the bandwidth of the system should be increased properly according to machining conditions. In order to implement the compensating strategy for the measurement, a Kalman filter is designed for the test system. Before the construction of the filter, the frequency response functions (FRFs) of the spindle-tool system are tested through experimental modal analysis (EMA). With the implementation of the compensation, the bandwidth of the testing system is increased significantly. Besides, a strong tracking filter (STF) is designed and applied to the system to improve filter performance. Application of the proposed method has been demonstrated experimentally through cutting tests under different experimental conditions on a three-axis vertical machining center.

Analysis of the machine tool dynamic characteristics in manufacturing space based on the generalized dynamic response model

Machine tool dynamic characteristics are seriously affected by the changes of the machining poses and the spindle bearing joints dynamic properties. As these changes contribute much more complexity and uncertainty for predicting the machine tool dynamic characteristics accurately, a new method is developed to research the changing regularity of the whole machine tool dynamic characteristics in generalized manufacturing space. In this method, the dynamic flexibilities at the spindle nose of x, y, and z directions in the focused frequency range are taken to represent the whole machine tool dynamic characteristics. The response surface method (RSM) and the orthogonal experiment design method are combined to establish the generalized dynamic response model, which contains the information of the spatial poses and the spindle bearing dynamic parameters. To establish this model, the simulations arranged by the orthogonal design are conducted by utilizing the dynamic model modified approach based on the validated finite element model (FEM) of the whole machine tool. With evaluating the fitting degree of this generalized dynamic response model, it can be used to predict the dynamic characteristics in the manufacturing space. Furthermore, an algorithm based on the established model is proposed to calculate the effect factors acting on the whole machine tool dynamic characteristics, which are caused by the changes of the spindle bearing dynamic parameters. The proposed analytical method has been applied in a three-axis vertical machining center to establish its generalized dynamic response model. The dynamic flexibilities at the spindle nose in the generalized manufacturing space are predicted, and they are validated by the dynamic experiments. With the calibrated model, effect factors of the spindle bearing joints are also obtained. All the predicted results support a theoretical basis on the optimal process routes planning, and the proposed analytical method can lay a foundation for further study on the dynamic information of the tool point.

Key geometric error extraction of machine tool based on extended Fourier amplitude sensitivity test method

Machining accuracy is a very important index to measure the performance of NC machine tools, and how to improve the machining accuracy grabs the attention of both manufacturers and users. Since the geometric errors are intercoupling and are one of the main factors affecting machining accuracy of machine tools, in this paper, a new method for extracting key geometric errors based on the screw theory and extended Fourier amplitude sensitivity test (EFAST) method was proposed. First, screw theory was introduced, and the machining accuracy modeling of a three-axis vertical machining center (VMC) was established. Then, in order to analyze the coupling effects between geometric errors by high-dimensional nonlinear model, EFAST global sensitivity analysis method was adopted to help in extracting the main geometric errors. An illustration example on a three-axis VMC shows that the proposed method is effective and can be helpful to improve the geometric accuracy of machine tool.

Product of exponential model for geometric error integration of multi-axis machine tools

This paper proposes a product of exponential (POE) model to integrate the geometric errors of multi-axis machine tools. Firstly, three twists are established to represent the six basic error components of each axis in an original way according to the geometric definition of the errors and twists. The three twists represent the basic errors in x, y, and z directions, respectively. One error POE model is established to integrate the three twists. This error POE formula is homogeneous and can express the geometric meaning of the basic errors, which is precise enough to improve the accuracy of the geometric error model. Secondly, squareness errors are taken into account using POE method to make the POE model of geometric errors more systematic. Two methods are proposed to obtain the POE models of squareness errors according to their geometric properties: The first method bases on the geometric definition of errors to obtain the twists directly; the other method uses the adjoint matrix through coordinate system transformation. Moreover, the topological structure of the machine tools is introduced into the POE method to make the POE model more reasonable and accurate. It can organize the obtained 14 twists and eight POE models of the three-axis machine tools. According to the order of these POE models multiplications, the integrated POE model of geometric errors is established. Finally, the experiments have been conducted on an MV-5A three-axis vertical machining center to verify the model. The results show that the integrated POE model is effective and precise enough. The error field of machine tool is obtained according to the error model, which is significant for the error prediction and compensation.

Test Research of Vertical Machine Center's Geometrical Accuracy Stabilization

Domestic CNC machine tool and NC machine tools abroad's main gap is accuracy stabilization rather than accuracy. Accuracy stabilization test is done on some three-axis vertical machining center aiming at the status. Laser interferometer is used to test X, Y, Z axis's positioning accuracy, repeat positioning accuracy and reverse gap. The test is done taking 80 days as the period. The accuracy test results is analyzed. From the positioning accuracy, repeat positioning accuracy and reverse gap curves we can see that, because of the factors of machine's use and temperature's change, X, Y, Z axis's positioning accuracy changes violently, repeat positioning accuracy and reverse gap change a little.

Geometric Error Identification of a Three-Axis Machining Center

Geometric errors of a machining center can cause great influence on machining accuracy, and these geometric errors should be identified and compensated in the actual working conditions. Taking a three-axis vertical machining center as example, 21 geometric errors of the machine tool were solved. By using the 12-line method based on a laser interferometer, identification principle of the positioning errors, straightness errors, pitch errors, yaw errors, roll errors and squareness errors are presented, and all of the 21 geometric errors of the machining center were identified. Geometric errors having great influence effect on machining accuracy can be identified. The research results provide guidance for analyze of geometric errors of machining center.

Trochoidal milling

In this paper, an approach to the modeling of trochoidal milling is proposed. An analytical model for defining the engagement is derived and a force model depending on the engagement is employed for the prediction of the cutting forces. In order to handle more complex part geometries, in addition to the analytical model, a numerical engagement model is developed. With the numerical model, any surface with holes, bosses, etc. can be handled. Moreover, for the maximization of the efficiency in the trochoidal milling operation, a different tool path strategy (double trochoidal milling) is examined. Physical experiments are performed in a three-axis vertical machining center and the predicted forces are compared with the measured forces. It is observed that the predicted forces and the measured forces are in good agreement in all the applications of the study.

Comparative Statistical Analysis of Test Parts Manufactured in Production Environments

Estimating error uncertainties arising in production parts is not a well-understood process. An approach to estimate these uncertainties was developed in this study. Machine tool error components were measured on a three-axis vertical machining center. Multiple parts were produced on the measured machining center then measured on a coordinate measuring machine. Uncertainty models for hole-center to hole-center lengths and orthogonalities were developed using measured machine tool errors. These estimated uncertainties were compared against measured uncertainties. The main conclusion from the study is that the Law of Propagation of Uncertainties can be used to estimate machining uncertainties.

Automated process planning for CNC machining of spherical space-frame nodes

Space-frame nodes are joints connecting linear members in complex space frames. Each spherical node has a particular hole configuration that defines the orientation of the bolt holes used to connect the space-frame members. The hole configurations are determined by CAD software for space-frame design. Nodes with the same hole configuration vary in number depending on the design of the space frame. A computer-automated machining cell has been implemented to batch manufacture the spherical nodes. The machining cell consists of a three-axis vertical machining center interfaced with a two-axis CNC rotary table. To reduce changeover time between the machining of batches, produce consistent and reliable NC programs, and minimize manpower required to run the cell, a PC-based computer-automated process planning system has been developed to directly translate geometric data from the CAD software to NC programs, without intervention by the operator. An important consideration in the process plan for the machining of the space-frame node is the transformation of the CAD-defined bolt hole configuration to a mounting-hole-referenced configuration that minimizes setup and mounting hole requirements. Based on the optimal mounting-hole-referenced configuration, NC toolpaths are derived and corresponding NC programs are generated. This paper presents the approach to determine the optimal mounting-hole-referenced configuration and the operation of the automated process planning system that is suitable for the machining of spherical parts.