Force based tool wear monitoring system for milling process based on relevance vector machine

The monitoring of tool wear status is paramount for guaranteeing the workpiece quality and improving the manufacturing efficiency. In some cases, classifier based on small training samples is preferred because of the complex tool wear process and time consuming samples collection process. In this paper, a tool wear monitoring system based on relevance vector machine (RVM) classifier is constructed to realize multi categories classification of tool wear status during milling process. As a Bayesian algorithm alternative to the support vector machine (SVM), RVM has stronger generalization ability under small training samples. Moreover, RVM classifier results in fewer relevance vectors (RVs) compared with SVM classifier. Hence, it can be carried out much faster compared to the SVM. To show the advantages of the RVM classifier, milling experiment of Titanium alloy was carried out and the multi categories classification of tool wear status under different numbers of training samples and test samples are realized by using SVM and RVM classifier respectively. The comparison of SVM with RVM shows that the RVM can get more accurate results under different number of small training samples. Moreover, the speed of classification is faster than SVM. This method casts some new lights on the industrial environment of the tool condition monitoring.     

Multilevel classification of milling tool wear with confidence estimation

An important problem during industrial machining operations is the detection and classification of tool wear. Past research in this area has demonstrated the effectiveness of various feature sets and binary classifiers. Here, the goal is to develop a classifier which makes use of the dynamic characteristics of tool wear in a metal milling application and which replaces the standard binary classification result with two outputs: a prediction of the wear level (quantized) and a gradient measure that is the posterior probability (or confidence) that the tool is worn given the observed feature sequence. The classifier tracks the dynamics of sensor data within a single cutting pass as well as the evolution of wear from sharp to dull. Different alternatives to parameter estimation with sparsely-labeled training data are proposed and evaluated. We achieve high accuracy across changing cutting conditions, even with a limited feature set drawn from a single sensor.     

Force-torque based on-line tool wear estimation system for CNC milling of Inconel 718 using neural networks

In a modern machining system, tool condition monitoring systems are needed to get higher quality production and to prevent the downtime of machine tools due to catastrophic tool failures. Also, in precision machining processes surface quality of the manufactured part can be related to the conditions of the cutting tools. This increases industrial interest for in-process tool condition monitoring (TCM) systems. TCM supported modern unmanned manufacturing process is an integrated system composed of sensors, signal processing interface and intelligent decision making strategies. This study includes key considerations for development of an online TCM system for milling of Inconel 718 superalloy. An effective and efficient strategy based on artificial neural networks (ANN) is presented to estimate tool flank wear. ANN based decision making model was trained by using real time acquired three axis (F_x, F_y, F_z) cutting force and torque (M_z) signals and also with cutting conditions and time. The presented ANN model demonstrated a very good statistical performance with a high correlation and extremely low error ratio between the actual and predicted values of flank wear.     

Physics guided neural network for machining tool wear prediction

Tool wear prediction is of significance to improve the safety and reliability of machining tools, given their widespread applications in nearly every branch of manufacturing. Mathematical modelling, including data driven modelling and physics-based modelling, is an important tool to predict the degree of tool wear. Howerver, the performance of conventional data driven models is restricted by the absent representation of physical inconsistency. The physics-based models usually fail to consider the complex tool cutting conditions and dynamic changes of physical parameters in practice. To address these issues, a novel physics guided neural network model is presented for tool wear prediction. Firstly, a cross physics-data fusion (CPDF) scheme is proposed as the modelling strategy to fuse the hidden information explored by a physics-based model and a data driven model. Secondly, the information hidden in the unlabelled sample is explored by the physics-based model of tool cutting, inspired by semi-supervised learning. Thirdly, a novel loss function which takes the physical discipline into account is proposed to evaluate the physical inconsistency quantitatively. The advantage of the developed method is that it explores sufficient information from both physics and data domains to eliminate the physical inconsistency existing in conventional data driven models.     

基于贝叶斯和蒙特卡罗的多信号模型扩展研究

传统多信号模型基于确定性测试假设条件,忽略了系统存在不确定性的真实情况,在传统多信号模型基础上引入贝叶斯条件概率来表示不确定性问题,并通过蒙特卡罗方法进行仿真模拟,将不确定性问题转化为单次试验确定性问题,进而使用相关矩阵进行测试性分析,通过程序实现和算例验证了该方法的有效性,并可以根据反馈数据进行参数学习,修正初始条件概率。     

基于多信号模型和盲源分离的复合故障诊断方法

利用多信号模型可简明表征系统因果关系以及盲源分离算法可提取系统本源信息的特点, 提出一种新颖有效的复合故障诊断方法. 首先, 针对复合故障下多信号模型出现冗余测试和故障模糊组的情况, 应用盲源分离算法实现测点信息的盲分离, 基于盲信号重建多信号模型的因果结构; 其次, 理论分析了该方法对复合故障具有良好的可诊断性. 轧制过程AGC系统的实验结果表明, 所提出方法对双复合故障和部分多复合故障的隔离和定位准确率可达100%.

     

基于GEP的高速公路通行费预测方法研究

高速公路通行费未来收入状况的预测对于高速公路运营管理、建设规划有着重要的指导意义。然而,通行费收入水平的变化受到多方面因素的影响,具有较强的非线性和复杂性,传统预测模型无法准确表达通行费收入的发展规律。本文针对复杂的高速公路通行费预测问题,建立了基于基因表达式编程算法(GEP)的高速公路通行费预测模型。该模型利用GEP算法建立通行费当前收入与历史数据之间复杂的函数关系,准确地刻画通行费收入随时间的发展规律。此外,针对节假日期间通行费减免政策的影响,提出了有效的修正模型。最后,本文采集了浙江沪杭甬高速公路股份有限公司等12家公司通行费收入的历史数据进行仿真实验,对比传统的ARIMA以及神经网络预测模型,结果充分验证了本文算法的有效性和准确性。     

A corner-looping based tool path for pocket milling

In milling around corners, cutting resistance rises momentarily due to an increase of cutter contact length. NC tool path generation in dealing with sharp corners thus requires special consideration. This paper describes an improved NC tool path pattern for pocket milling. The basic pattern of the improved tool path is a conventional contour-parallel tool path. Bow-like tool path segments are appended to the basic tool path at the corner positions. When reaching a corner, the cutter loops around the appended tool path segments so that corner material is removed progressively in several passes. By using the corner-looping based tool path, cutter contact length can be controlled by adjusting the number of appended tool path loops. The procedures of creating the improved tool path for different corner shapes are explained. The proposed tool path generation was implemented as an add-on user function in a CAD/CAM system. Cutting tests were conducted to demonstrate and verify the significance of the proposed method.     

基于深度卷积神经网络的刀具寿命动态预测研究

刀具寿命预测对提高工件加工精度和生产加工效率具有重要意义.同工况下同型号刀具监测信号数据分布不一致,导致历史寿命预测模型对刀具寿命预测效果有限.鉴于此,提出一种基于深度卷积神经网络(DCNN)的刀具寿命动态预测方法.首先,利用DCNN挖掘历史刀具监测信号的退化趋势特征,构建刀具寿命预测模型,并加入注意力机制对DCNN输出进行加权,加强对刀具寿命特征的学习,提高寿命预测准确度;然后,通过基于KL散度对刀具监测信号数据分布不一致进行检测,从而在已有刀具寿命预测模型的基础上进行更新迭代;最后,利用迭代后的模型再次进行刀具寿命预测.所提出方法很好地体现了刀具实际加工过程对刀具寿命的影响,以铣削数据集为例验证了所提出方法的有效性.     

基于T_NAG模型的攻击路径预测方法研究

为了更好地刻画攻击者的攻击轨迹,设计出一种基于T_NAG(time attribute network attack graph)模型的路径预测方法。首先,提出新的攻击图模型T_NAG,根据实时行为轨迹对攻击者能力加以区分;其次,依据攻击者具有不同能力的特性,提出攻击意向的概念,统筹考虑操作风险与攻击收益,将时间衰减参数引入到攻击意向计算中,并设计出一种基于攻击者能力的漏洞利用率量化方法;最后,将攻击意向与漏洞利用率进一步融入到对路径可达概率的考量中,给出预测攻击路径的IntenAbi-PathPre算法。实验结果表明,该方法可以有效去除攻击图中的冗余,并且使攻击路径预测的准确性得到明显提高。     

基于BBO-ELM方法的网络视频流量预测仿真研究

针对在互联网络服务中,进一步提升网络视频流量预测的精度以优化网络资源配置和满足用户需求的问题进行了研究,并对如何自适应选取网络视频流量时间序列中有效且必需的历史信息进行了探索,提出一种基于生物地理学优化算法优化极限学习机的BBO-ELM预测模型。在ELM预测模型的基础上,将BBO优化算法用于ELM的网络输入变量、隐含层节点的配置及参数、Tikhonov正则化参数的优化选取。为验证所提出方法的有效性,将BBO-ELM方法应用于真实网络视频流量预测实例中,在同等条件下,与现有方法进行了比较。仿真实验结果表明,该方法能有效地改善预测精度,显示出其有效性以及应用潜力。     

Rapid prediction of posture-dependent FRF of the tool tip in robotic milling

During the robotic milling process, vibration is one of the main factors that affect the machining accuracy and surface quality due to the low stiffness of the robot structure. The robotic milling stability is a function of the frequency response function (FRF) at the tool tip, which is posture-dependent within the workspace. This paper introduces an approach for rapidly predicting the tool tip FRF for industrial robotic milling at any posture. In this method, the models of the one degree-of-freedom (DOF) robot and two DOF robot are extended to a six DOF industrial robot to calculate the FRF at the holder tip based on the FRF acquisition tests at the arranged postures and a standardization process. Considering the coupling effects between the holder and the tool, the tool tip FRF at any posture of the milling robot is calculated using the receptance coupling substructure analysis (RCSA) method. Accordingly, the proposed method is applied to an industrial robot, and the feasibility of this method for predicting the posture-dependent FRF at high frequency in the workspace is validated though the impact tests. Moreover, the stability lobe diagram is calculated and the chatter tests are performed to validate its accuracy. At last, the robot structural modes are observed at the low-frequency dominant modes, whose frequencies are around 10 to 20 Hz.     

一种基于多信号流图的故障隔离方法

多信号流图模型中,检测同一属性或具有相同激励的测试可以同时被执行,并按此对测试进行分组。在故障隔离过程中,按各组能检测出的故障出现的概率大小来决定测试顺序,从而构成了基于测试划分的故障隔离方法。该方法最大程度地减小了测试时间,并兼顾考虑了故障出现的概率和最少测试硬件增加。     

Simulation of three-dimension cutting force and tool deflection in the end milling operation based on finite element method

This paper presents a 3D simulation system which is employed in order to predict cutting forces and tool deflection during end-milling operation. In order to verify the accuracy of 3D simulation, results (cutting forces and tool deflection) were compared with those based on the theoretical relationships, in terms of agreement with experiments. The results obtained indicate that the simulation is capable of predicting the cutting forces and tool deflection.     

基于改进预测方法的DTC控制系统研究

异步电机在低速运行时,定子电阻受温度、定子频率、定子电流等的影响较大,直接转矩控制系统的性能受到直接的影响。为此,将预测控制算法与全阶状态观测器相结合,对定子电阻进行在线识别,由此构成改进的直接转矩控制系统。在MATLAB/Simulink软件中对上述设计进行仿真研究,仿真结果表明,改进的异步电动机直接转矩控制系统能够提高系统的鲁棒性,使得低速性能得到了较好的改善。     

On line tool wear monitoring based on auto associative neural network

This paper presents a new tool wear monitoring method based on auto associative neural network. The main advantage of the model lies that it can be built only by the data under normal cutting condition. Therefore, the training samples of the tool wear status are no longer needed during the training process that makes it easier to be applied in real industrial environment than other neural network models. An averaged distance indicator is proposed to denote not only the occurrence of the tool wear but also its severity. Moreover, the Levenberg–Marquardt (LM) training algorithm is introduced to improve the convergence accuracy of the auto associative neural network. Based on the proposed method, a framework for online tool condition monitoring is illustrated and the cutting force data under different tool wear status are collected to simulate the online modeling and monitoring process for the rough and finish milling respectively. The results show that the proposed indicator can reflect the evolution process of tool wear correctly and the LM algorithm is more accurate in comparison with the gradient descent methods. Therefore, it casts new light on practical application of neural network in the field of on line tool condition monitoring.     

A numerical method for prediction of curved rail wear

Important published papers on rail wear in the past were reviewed. A numerical method was put forward to predict curved rail wear during a railway vehicle curving. The numerical method was discussed in detail. It considered a combination of Kalker’s non-Hertzian rolling contact theory, rail material wear model, the coupling dynamics of the vehicle and track, and the three-dimensional contact geometry analysis of wheel-rail. In its numerical implementation, the dynamical parameters of all the parts of the vehicle and track, such as normal loads and creepages of the wheels and rails, were firstly obtained through the curving dynamics analysis. The wheel-rail contact geometry calculation gave the wheel-rail contact geometry parameters, which were used in the wheel-rail rolling contact calculation with Kalker’s non-Hertzian rolling contact theory modified. The friction work densities on the contact areas of the wheels and rails were obtained in the rolling contact calculation, and were used to predict the rail running surface wears caused by the multiple wheels of the vehicle simultaneously with the rail material wear model. In the rail material wear model, it was assumed that the mass loss of each unit area was proportional to the frictional work density in the contact area. A numerical example was present to verify the present method. The numerical results of the example are reasonable, and indicate that the high rail wear of the curved track caused by the leading wheelset is much more serious than those caused by the other three wheels of the same bogie.     

An automatic and accurate method for tool wear inspection using grayscale image probability algorithm based on bayesian inference

Accurate, rapid and automated tool wear inspection is critical to manufacturing quality, cost and efficiency in smart manufacturing systems. However, manual inspection of tool wear is still a common industrial practice which is inefficient, prone to human errors and not suitable for digitized manufacturing. Previously reported automatic tool wear inspection methods were inaccurate because they only used the remaining worn boundary (i.e., the partial-absence original tool boundary) to approximate tool wear. The authors discovered the association principle between the change law of the cutting edge grayscale and the relative position of the original and worn boundary, which was used to establish the probability functions to accurately reconstruct the curved original tool boundary via Bayesian Inference. The experiment results reported in this paper proved higher efficiency and accuracy than previous automatic tool wear inspection methods.     

Optimize tool paths of flank milling with generic cutters based on approximation using the tool envelope surface

This paper presents a global optimization method to generate a tool path for flank milling free-form surfaces with a generic cutter based on approximation using the tool envelope surface. It is an extension of our previous work [Gong Hu, Cao Li-Xin, Liu Jian. Improved positioning of cylindrical cutter for flank milling ruled surfaces. Computer Aided Design 2005; 37:1205–13]. First, given initial tool path or tool axis trajectory surface, the grazing points of the tool envelope surface can be calculated. Second, the errors between the tool envelope surface and the designed surface along the normal direction of the tool envelope surface are calculated. Based on this new definition of error, an optimization model is established to get the global optimized tool axis trajectory surface. In order to simplify the calculation, two variants of this method based on the least square criterion are proposed to solve this model. Since this method is really based on the tool envelope surface, it can reduce the initial machining errors effectively. The proposed method can be used not only for cylindrical cutters and conical cutters, but also for generic cutters with a surface of revolution. In addition to ruled surfaces, it also can be used for machining non-ruled surfaces. Finally, several examples are given to prove its effectiveness and accuracy. The generated tool paths and calculated grazing points for test are available in supplementary files for the readers’ convenience in verifying this work in different CAD/CAM systems.     

基于二次指数平滑预测的虚拟机调度方法研究*

针对数据中心的高能耗问题,提出了一种基于负载感知和预测的虚拟机调度方法,采用二次指数平滑法预测物理主机资源负载情况,利用MMT和MM相结合的策略选择待迁虚拟机,使用资源最佳适配策略（BRF）选择目标物理主机。该调度方法的预测模型能提高迁移触发准确率,随着调度轮数的增加,对资源需求互补的虚拟机会被整合到相同物理主机上,从而减少迁移次数;最后,通过CloudSim仿真平台与FT_MMT、CDLC、AR_MMT调度策略进行了对比,结果表明该调度方法在能耗节约、迁移次数方面均有提升。