Thermal error measurement and modelling in machine tools. Part II. Hybrid Bayesian Network—support vector machine model

Prediction accuracy of machine tool thermal error significantly depends on the structure of the error model. Machine tool thermal error varies considerably depending upon the specific operating parameters adopted. Most error models developed thus far generally employ neural networks to map temperature data against thermal error. However, it is very important to account for the specific conditions as well within the model. This paper presents a hybrid Support Vector Machines (SVM)–Bayesian Network (BN) model that seeks to address this issue. The experimental data is first classified using a BN model with a rule-based system. Once the classification has been effected, the error is predicted using a SVM model. The hybrid thermal error model thus predicts the thermal error according to the specific operating conditions. This concept leads to a more generalised prediction model than the conventional method of directly mapping error and temperature irrespective of conditions. Such a model is especially useful in a production environment wherein the machine tools are subject to a variety of operating conditions.     

基于数据驱动的数控机床自适应迭代学习控制

数控机床位置伺服系统受加工环境、零件形状和机床机电特性等变化因素的影响,其零件加工是一个典型的非线性、时变和不确定动力学变化过程,因此,建立其精确机制模型很困难。针对相同零件批量加工过程呈现的重复运行特点,基于被控对象的等价数据模型,提出一种基于数据驱动的自适应迭代学习控制方法。所提控制方法采用沿迭代轴的动态线性化方法,通过最小化控制目标函数,仅利用数控机床位置伺服系统的输入输出数据,实现学习控制增益的自适应更新,克服传统P型迭代学习控制方法固定增益的问题,并经过严格理论分析保证了该方法的收敛特性。仿真结果表明:提出的数据驱动自适应迭代学习控制方法,相比传统P型迭代学习控制方法,平均绝对误差和最大绝对误差分别减小了46%和56%。     

Using unsupervised incremental learning to cope with gradual concept drift

Current computational approach to incremental learning requires a constant stream of labelled data to cope with gradual environmental changes known as concept drift. This paper examines a case where labelled data are unavailable. Inspired by the performance of the human visual system, capable of adjusting its concepts using unlabelled stimuli, we introduce a variant to an unsupervised competitive learning algorithm known as the Leader Follower (LF). This variant can adjust pre-learned concepts to environmental changes using unlabelled data samples. We motivate the needed change in the existing LF algorithm and compare between two variants to enable the accumulation of environmental changes when facing unbalanced sample ratio.     

基于深度迁移学习的矿井通风机轴承故障诊断

针对实际应用中矿井通风机轴承负样本少导致故障诊断率低的问题，提出一种基于深度迁移学习的矿井通风机轴承故障诊断方法。组合卷积神经网络(CNN)与双向门控循环单元(BiGRU),并采用随机森林(RF)分类器替换CNN的Softmax层，构建CNN-BiGRU-RF诊断模型，提取轴承更深层次故障特征以便于故障识别；利用源域数据对模型训练，确定模型结构参数；最后，引入迁移学习将模型迁移至目标域，使用目标域有标签数据微调模型参数，构建目标域诊断模型进行故障分类。实验结果表明：在矿井通风机轴承负样本稀少情况下，所提方法的故障识别平均准确率在94%以上，与其他方法相比，具有更好的诊断精度和泛化能力。     

移动机器人神经网络补偿计算力矩控制

针对存在动力学不确定建模项、建模误差及外界干扰的移动机器人,设计RBF神经网络补偿计算力矩控制算法。基于反步法设计运动学辅助速度控制率。根据动力学理想名义模型,基于计算力矩法设计一般的力矩控制器。在此基础上,建立具有不确定建模项、建模误差及外界干扰的移动机器人动力学模型,基于计算力矩法设计带有RBF神经网络补偿的力矩控制器,神经网络的权值由自适应律给出。最后,利用Lyapunov理论证明了系统的稳定性。仿真结果表明：神经网络对系统不确定项具有良好的逼近性能,相比于一般的计算力矩控制,所提出的神经网络补偿计算力矩控制算法具有更好的跟踪性能,控制系统具有更好的鲁棒性。     

Hybrid modelling of 7449-T7 aluminium alloy friction stir welded joints

Abstract

Many finite element models use adjustable parameters that control the heat loss to the backing bar, as well as the heat input to the weld. In this paper, we describe a method for determining these parameters with a hybrid artificial neural network (ANN) coupled thermal flow process model of the friction stir welding process. The method successfully determined temperature dependent boundary condition parameters for a series of friction stir welds in 3·2 mm thick 7449 aluminium alloy. The success of the technique depended on the method used to input thermal data into the ANN and the ANN topology. Using this technique to obtain the adjustable parameters of a model is more efficient than the conventional trial and error approach, especially where complex boundary conditions are implemented.     

三轮全向移动机器人里程计在线校正方法研究

传统里程计校正方法常使用离线校正手段,在地面环境发生改变的情况下,校正效果较差。为解决上述问题,提高机器人定位精度,以三轮全向移动机器人为平台,提出一种结构简单、鲁棒性强的在线里程计校正方法。该方法通过扩展卡尔曼滤波算法处理传感器数据,以得到机器人的实时位姿信息和速度信息,结合三轮全向移动平台的动力学模型,及时修正里程误差。在V-REP中设计仿真实验,实验结果表明:采用所提方法,校正后的里程精度有了极大改善,并克服了离线校正方法受地面环境影响的问题。在已有实际平台上验证了该算法的有效性。     

Linux防火墙中获取网络流量的关键技术

本文详细介绍了在Linux平台上,应用流量分析技术,使防火墙能截获网络中传送的流量信息。程序通过系统内部函数的调用,完成了数据包的获取,使得防火墙由传统的被动式防护向主动式的追踪检测转变,从而使内部网络更加安全可靠。     

Learning in presence of class imbalance and class overlapping by using one-class SVM and undersampling technique

ABSTRACT

The class imbalance problem engraves the traditional learning models by degrading performance and yielding erroneous outcomes. It is the scenario where one of the class representation is over-shadowed by other classes in a data space. Presence of class imbalance can cause a grave difficulty as misclassification cost of minority class tends to be very high. Presence of overlapping cases along with the case of imbalanced data, can lead to create grim situation for effective learning. In this study, an in-depth analysis of the effects of class imbalance and class overlapping in conventional learning models has been presented. A data level approach is adapted with one-class SVM-based anomaly detection to detect the cases of data overlapping while an adapted Tomek-link undersampling algorithm is defined to treat both overlapped and imbalanced cases. The proposed model evolves to eliminate borderline, redundant and overlapping cases with the account of Tomek-link pair and sparse neighbourhood. The proposed method has been evaluated with six state-of-the-art models for seven binary and two multiclass datasets, with respect to three standard learning models. The proposed model has been evaluated with cost-sensitive learning and extreme learning based approaches for imbalanced class learning. The proficiency of the proposed method over state-of-the-art models is established through experimental analyses.     

A weighted pattern matching approach for classification of imbalanced data with a fireworks-based algorithm for feature selection

ABSTRACT

Learning a classifier from imbalanced data is a challenging problem in Machine learning. A dataset is said to be imbalanced when the number of instances belonging to one class is much less than the number of instances belonging to the other class. Classifiers that proves efficient on standard data fail when the data is imbalanced as they are over trained by the majority class instances. Since class imbalance is a common characteristic of real-world data, the need for better classifiers becomes essential. This paper proposes a novel instance-based classification algorithm called Weighted Pattern Matching based Classification (PMC+) for classifying imbalanced data. PMC+ classifies unlabelled instances by computing the absolute difference between the feature values of the instances in the dataset and the unlabelled instance. PMC+ employs a simple classification procedure with weights and shows reasonably good performance. To improve the performance of PMC+, Fireworks based Feature and Weight Selection algorithm based on the idea of PMC+ has been proposed. PMC+ is evaluated on 44 binary imbalanced datasets and 15 multiclass imbalanced datasets. Although PMC+ does not employ a resampling or cost-sensitive method, experiments show that PMC+ is effective for classification of imbalanced data. The results of the experiments were validated using various non-parametric statistical tests.     

机器学习辅助下的五轴数控铣削刀轨优化

针对铣削加工过程中刀具挠度变形的自动补偿问题,提出了一种用于五轴数控加工的刀轨自优化方法.首先,该方法从铣削加工材料去除仿真中获得工艺条件,且将计算出的切削条件与相应的形状误差测量相关联;其次,采用基于统计学习理论的支持向量回归(Support Vector Regression,SVR)来预测所产生的形状误差,并进行...     

油液颗粒过滤器动态过滤特性仿真与实验研究

现有油液颗粒过滤器的过滤模型采用基于时间平均概念的理论过滤比,没有考虑系统温度、流量波动等动态运行因素对过滤比的影响。为此,基于滤前、滤中和滤后的系统理论建立油液颗粒过滤器的动态过滤理论模型,研究基于动态过滤比的油液颗粒过滤器过滤特性。与传统的基于恒定过滤比的方法相比,所提方法考虑重复过滤和重复污染的影响,能够实现对油液颗粒过滤器过滤效果的准确表征。通过机械传动系统污染控制系统的仿真分析与实验研究,验证所提方法的有效性。结果表明：油液颗粒过滤器动态过滤比随着重复过滤率和重复污染率的增大而不断减小;当污染控制系统流量较大、压力较高和温度较低时,油液颗粒过滤器的动态过滤比较大。     

基于梯度奖励的深度强化学习移动机器人路径规划

针对目前深度强化学习移动机器人路径规划中稀疏奖励导致的效率低、收敛慢等问题,提出一种梯度奖励政策。使用区域分割将环境分割为缓冲区、探索区、临近区以及目标区,奖励的动态变化可以逐步缩小机器人的探索范围,同时在安全区域内也能获得正向奖励。首先输入机器人当前的位置坐标,经过神经网络后估计4个动作的Q值,随后通过去首动态贪婪策略达到最大化探索,最后采用基于均方误差的优先经验回放抽取样本进行梯度下降更新网络。实验结果表明：在小范围环境内探索效率可提升近40%,在大范围环境下成功率高于80%,而且在提高探索效率的同时增强了鲁棒性。     

Machine tool component error extraction and error compensation by incorporating statistical analysis

The extraction of component errors of a machine tool’s axes is a critical step for the synthesis of 3D volumetric error mapping, which is a prerequisite to improve the machine tool accuracy by numerical error compensation. This paper presents a method for the extraction of machine tool component errors from a statistical point of view. First, the B-Spline mathematical model is established to represent the component error function, and the least-squares fitting method to measured data points is presented. Then, statistical analysis is used to select the B-Spline model with proper flexibility, so as to separate repeatable errors from random errors in the measured data. Finally, based on the component error extraction method, numerical error compensation experiments were conducted on the XY-plane of a high precision machine tool by using a cross-grid scale system. According to the statistical analysis of the experimental data, all repeatable errors except the dynamic errors caused by machine tool control system were compensated for.     

基于多体系统理论的精密加工中心综合误差建模

基于多体系统理论,针对一台含有工作台站的精密加工中心,建立其综合误差模型,分析并解决了交换工作台站的运动误差对综合误差的误差传递问题,改进了各轴热漂移误差的综合误差建模方法。运用该误差模型进行实时补偿,能有效提高加工中心的加工精度。     

数控机床位置伺服系统的无模型自适应迭代学习控制

数控机床位置伺服系统在加工过程中受负载、摩擦和电路系统响应特性等因素影响,很难精确建立其加工过程动力学模型。针对批量零件加工过程中的重复执行过程,设计了一种数据驱动的无模型自适应迭代学习控制方案。该方案借助沿迭代轴的动态线性化方法,将数控机床位置伺服系统加工动力学过程等价转化成一个虚拟的迭代数据模型,并根据设计的迭代学习控制律和参数估计律构建数控机床位置伺服系统的无模型自适应迭代学习控制方案。仿真结果表明:该迭代学习控制方案基于数控机床重复运行的特点,仅利用位置和电机电流信息,完成了对零件加工过程的改善,提高了加工精度。     

Updated CO2/H2S internal corrosion model of natural gas pipelines based on flow field calculations

The ‘A line’ of the Sichuan Natural Gas East Transportation in China was used as the subject of the study, and an existing CO₂/H₂S corrosion model was utilised to predict the uniform corrosion rate. Then, the flow parameters were simulated by computational fluid dynamics based on turbulence theory, and the influence on the corrosion rate of the pipeline was analysed in a detail to more accurately describe the corrosion problems of natural gas pipelines that contain CO₂/H₂S. After that, an updated CO₂/H₂S corrosion model under the influence of flow filed was proposed by modifying the existing CO₂/H₂S corrosion model. The actual condition was calculated by the updated CO₂/H₂S corrosion model. Results show that flow parameters, namely, velocity, turbulent kinetic energy and phase distribution, affect pipeline corrosion. The flow parameters did not change significantly at the small scale changes in the pipeline (5 and 15°) of a broad and smooth flow channel of the large diameter gas transport pipeline. The shape of corrosion often appears in the form of an elliptical sheet. The corrosion location and the corrosion rate calculated by the updated model are consistent with the wall thickness detection data in the site conditions, which verified that the updated CO₂/H₂S corrosion model is valid. The updated CO₂/H₂S corrosion model influenced by the flow field can predict the corrosion distribution and the corrosion rate of the three-dimensional key positions in natural gas pipelines.     

Interactive natural language acquisition in a multi-modal recurrent neural architecture

ABSTRACT

For the complex human brain that enables us to communicate in natural language, we gathered good understandings of principles underlying language acquisition and processing, knowledge about sociocultural conditions, and insights into activity patterns in the brain. However, we were not yet able to understand the behavioural and mechanistic characteristics for natural language and how mechanisms in the brain allow to acquire and process language. In bridging the insights from behavioural psychology and neuroscience, the goal of this paper is to contribute a computational understanding of appropriate characteristics that favour language acquisition. Accordingly, we provide concepts and refinements in cognitive modelling regarding principles and mechanisms in the brain and propose a neurocognitively plausible model for embodied language acquisition from real-world interaction of a humanoid robot with its environment. In particular, the architecture consists of a continuous time recurrent neural network, where parts have different leakage characteristics and thus operate on multiple timescales for every modality and the association of the higher level nodes of all modalities into cell assemblies. The model is capable of learning language production grounded in both, temporal dynamic somatosensation and vision, and features hierarchical concept abstraction, concept decomposition, multi-modal integration, and self-organisation of latent representations.     

TA15钛合金动态再结晶晶粒生长模型

从理论上分析TA15钛合金动态再结晶晶粒生长驱动力,提出动态再结晶晶粒生长等效反驱动力的概念。基于动态再结晶晶粒生长驱动力,建立（描述动态再结晶晶粒尺寸演变）的晶粒生长速率模型及晶粒尺寸模型。以TA15钛合金动态再结晶晶粒尺寸实验测定数据为例,采用遗传算法(GA)优化尺寸模型参数。结果表明,模型计算结果与实验数据能够达到较好的吻合,平均误差为7.4%。     

退火20MnNiMo合金高温流变行为的一种本构描述

通过20MnNiMo钢多组试样的热压缩实验获得应变速率为0.01～10 s-1、变形温度为1173～1473 K条件下的真应力-应变数据。结合Arrhenius双曲正弦本构方程,通过线性回归分析求解得到不同变形条件下本构模型中的热变形激活能Q,材料常数n、α及结构因子A,从而构建了用于表征20MnNiMo钢流变应力与应变量、温度、应变速率之间内在关系的本构方程。研究结果表明:20MnNiMo钢在热压缩变形过程中发生了明显的动态软化行为,流变应力水平随应变速率的增加而增加,随温度的升高而降低;流变应力的预测值与实验值较吻合,而且预测的最大相对误差仅为7.54%。