New automatic feeding and orienting system

A now and versatile automatic feeding and orienting system is described and analysed both experimentally and theoretically. The system is particularly suitable for orienting laminar parts and can utilize both the geometry and mass distribution of the part as orienting aids. Unlike the vibratory bowl feeder, where orientation is generally achieved by the rejection of incorrectly oriented parts, in the system described, orientation is achieved principally by the use of re-orienting techniques. Further, the performance of this system is readily predictable.     

Flexible parts feeding for flexible assembly

A key obstacle to the economic application of flexible automation is parts feeding. Traditionally, dedicated devices such as vibratory bowl feeders have performed parts feeding. Such devices have a high cost, and are dedicated to a single component geometry and, consequently, the number of dedicated feeders required for a planned variety of products is equal to the number of distinct part types. This results in a high capital cost that can only be justified in a dedicated mass production environment, where there are long production runs and few product changes. Dedicated feeding methods are also usually too inflexible for robotic assembly systems where there is a wide range of component types in small batches being produced at lower rates. Flexible feeders capable of feeding many different part types are now commercially available but they are invariably large and costly. This paper describes a compact, low cost belt feeder prototype based on Pherson et al.'s (1983) concept but, which is capable of feeding complex geometries using modern sensor technology for part recognition, a standard non-active orientation blade, and a novel method for handling cylindrical parts.     

Piezo actuated vibratory feeding with vibration control

This research investigated the effect of decoupling the motion of a conventional vibratory feeder in order to develop a one-axis vibratory sliding mode for feeding tiny delicate parts that could ultimately be adapted to two axes. This research investigated a new type of vibratory drive by employing a piezoelectric actuator. In a decoupled design, the action of the horizontal and vertical vibration would be separately controlled to achieve a desired part trajectory and consequently the feed rate of the part. A computer control scheme was used to develop voltage input waveforms to control these actuators to produce movements that cause the parts to move forward relative to the track and overcome backsliding and other inefficient feeding motions. There are two advantages to this type of design. First, it can handle the parts in a more controlled manner by keeping them in constant contact with the track while maintaining a high conveying speed. This results in good protection for fragile parts. Secondly, this concept results in feedback control schemes that could allow the computer to adjust the drive waveforms to optimize feeder performance. Due to the non-sinusoidal nature of driving waveforms, the higher harmonic components can excite undesired deformation of the responding trajectory. A self-tuning control based on dynamic system identification was employed to counteract the higher harmonic effect. A one-dimensional hardware model is built and experiments were conducted to verify the precision and feasibility of the controlled actuation.     

新型压电直线式振动送料器设计

目的为了满足现代工业领域中对轻、薄、小产品平稳输送的要求,设计一种利用压电双晶片振子惯性力作为驱动源的新型直线式振动送料器。方法在对振动送料器的工作原理进行分析的基础上,建立压电振子动力学模型,利用Matlab软件分析振子长度和质量对冲击力与变形量的影响,研制送料器样机并进行性能测试。结果当振动频率为209.5～214.5Hz时,送料器具备输送物料的能力;当系统共振时,输送速度最快,随着电压的增加,送料速度呈线性关系增加。结论与同型号的电磁振动送料器相比,研制的压电直线式振动送料器消耗电流仅为其14%,工作噪音下降了17 dB。与同型号日本压电送料器相比,输送速度提高了10%。     

Investigation of vibratory bowl finishing

Although vibratory finishing has been in commercial use for over 40 years, relatively little fundamental information is known about the process, and industrial practice remains largely based on empiricism and trial and error. An experimental investigation into the vibratory bowl finishing process was conducted using material removal rate and surface roughness as the dependent variables. Bowl performance can best be described in terms of acceleration and depends primarily on the feed weights used, while bowl loading has a relatively minor effect. Furthermore, small changes in bowl loading resulting from normal media wear can be neglected. Results showed that material removal rates were constant over time for aluminium, brass and steel and were sensitive to hardness and bowl acceleration. Surface roughness saturated after a fixed period and was primarily a function of material composition. Surface roughness did not appear to be sensitive to bowl acceleration.     

Hidden Two-Stream Collaborative Learning Network for Action Recognition

The two-stream convolutional neural network exhibits excellent performance in the video action recognition. The crux of the matter is to use the frames already clipped by the videos and the optical flow images pre-extracted by the frames, to train a model each, and to finally integrate the outputs of the two models. Nevertheless, the reliance on the pre-extraction of the optical flow impedes the efficiency of action recognition, and the temporal and the spatial streams are just simply fused at the ends, with one stream failing and the other stream succeeding. We propose a novel hidden twostream collaborative (HTSC) learning network that masks the steps of extracting the optical flow in the network and greatly speeds up the action recognition. Based on the two-stream method, the two-stream collaborative learning model captures the interaction of the temporal and spatial features to greatly enhance the accuracy of recognition. Our proposed method is highly capable of achieving the balance of efficiency and precision on large-scale video action recognition datasets.     

一种基于VGGish神经网络的水声目标识别方法

水声目标智能识别是水声装备智能化的重要组成部分,深度学习则是实现水声目标智能识别的重要技术手段之一。当前水声目标智能识别经常面临数据集较小带来的训练样本量不足的情况,针对小数据集识别中存在的因过拟合导致模型泛化能力不足,以及输入的水声信号二维谱图样式不统一的问题,文章提出了一种基于VGGish神经网络模型的水声目标识别方法。该方法以VGGish网络作为特征提取器,并在VGGish网络前部加入了信号预处理模块,同时设计了一种基于传统机器学习算法的联合分类器,通过以上措施解决了过拟合问题和二维谱图样式不统一问题。实验结果显示,该方法应用在ShipsEar数据集上得到了94.397%的识别准确率,高于传统预训练-微调法得到的最高90.977%的准确率,并且在相同条件下该方法的模型训练耗时仅为传统预训练-微调方法的0.5%左右,有效提高了识别准确率和模型训练速度。     

A neural network-based algorithm that searches for the measuring points of a rule surface

The purpose of this study is to use a proposed neural network-based algorithm to explore the determination of the recommended measuring points for a rule surface. The task of measuring a rule surface starts from the rule surface design blueprint. Mesh grid data on the designed rule surface were selected. The pattern recognition capability of the back-propagation neural network is explored in this article. The network learning was successfully performed by the learning and testing of the network, the support of a designated acceptable perpendicular error value, a learning model in which training examples were gradually added and the adjustment of the number of training examples according to the network structure.     

Improved neural network model for reverse engineering

While conventional engineering transforms engineering concepts into real parts, in reverse engineering real parts are transformed into engineering models. The construction of a surface from three-dimensional (3D) measuring data points is an important problem in reverse engineering. This paper presents a reconstruction method for the sculptured surfaces from the 3D measuring data points. The surface reconstruction scheme is presented based on a neural network. The reconstruction of the existing surfaces is realized by training the network. A series of measuring points from existing sculptured surfaces is used as a training set. Once the neural network has been trained, it serves as a geometric model to generate all the points that are needed. However, the learning rate for the neural network is relatively slow, and the learning accuracy is often unacceptably low. In this paper, to improve the performance of the neural network, a pre-processor is proposed before the input layer. The pre-processor maps the input into the larger space by generating a set of linearly independent values. The effect of the pre-processor is to increase modelling accuracy, and reduce learning time. Based on this method, experimental results are given to show that the reconstructed surfaces are faithful to the original data points. The proposed scheme is useful for regular or irregular digitized data.     

A computer analysis of orientation devices for vibratory bowl feeders

Currently, vibratory bowl feeders are widely used in industry to feed small parts in assembly processes. The actual orientation of parts is achieved by various devices used in these feeders. However, each type of device performs only a limited function. The objective of this project was to develop a systematic approach which both analyses the geometrical and topological properties of parts designed in a CAD system and selects appropriate orientation devices to feed the designed parts.     

Neural Networks in Applied Statistics

Artificial neural networks are computer algorithms or computer programs derived in part from attempts to model the activity of nerve cells. They have been applied to pattern recognition, classification, and optimization problems in the physical and chemical sciences, as well as in other fields. We introduce the principles of the multilayer feedforward network that is among the most commonly used neural networks in practical problems. The relevance of neural network models for the applied statistician is considered using a time series prediction problem as an example. The multilayer feedforward neural network uses a nonlinear function of the predictors to obtain predictions for future time series values. We illustrate the considerations involved in specifying a neural network model and evaluate the accuracy of neural network models relative to the accuracy obtained using other computer-intensive, nonmodel-based techniques.     

深度卷积神经网络在滑动轴承转子轴心轨迹识别中的应用

针对传统旋转机械智能识别方法需要人为提取特征及诊断精度低的问题,基于深度学习的强大学习能力,提出一种深度卷积神经网络故障诊断模型(Deep Convolutional Neural Network Fault Diagnosis Model,DCNN-FDM)用于轴心轨迹识别。该模型包括输入模块、特征提取模块及分类模块三部分。原始图像输入模型后,经过输入模块的二值化处理及最近邻插值,统一变为尺寸大小为32×32的单通道图像;经特征提取模块中两组交替的卷积层和池化层作用,得到图形特征;最后,这些特征经全连接层的扁平化处理而张成一维向量,输入到softmax分类器中进行分类。利用奇异值差分谱方法,对实测轴心轨迹进行提纯,得到4类轴心轨迹样本集用于DCNN-FDM的训练与预测。结果表明:所提模型较传统的浅层学习模型的识别效果好,可实现转子故障的精确诊断,识别率达到97.09%。最后通过全连接层的主成分可视化分析,验证了模型具备自适应特征学习能力。     

Classification and Research of Skin Lesions Based on Machine Learning

Classification of skin lesions is a complex identification challenge. Due to the wide variety of skin lesions, doctors need to spend a lot of time and effort to judge the lesion image which zoomed through the dermatoscopy. The diagnosis which the algorithm of identifying pathological images assists doctors gets more and more attention. With the development of deep learning, the field of image recognition has made longterm progress. The effect of recognizing images through convolutional neural network models is better than traditional image recognition technology. In this work, we try to classify seven kinds of lesion images by various models and methods of deep learning, common models of convolutional neural network in the field of image classification include ResNet, DenseNet and SENet, etc. We use a fine-tuning model with a multi-layer perceptron, by training the skin lesion model, in the validation set and test set we use data expansion based on multiple cropping, and use five models’ ensemble as the final results. The experimental results show that the program has good results in improving the sensitivity of skin lesion diagnosis.     

卷积神经网络在印刷品缺陷检测的应用

目的 为了改善传统机器检测印刷产品缺陷存在误费率高的不足。方法 提出以卷积神经网络为控制核心的印刷品缺陷检测系统。设计可在实际检测中应用的卷积神经网络,设计在线印刷质量检测系统的硬件结构。结果 对结构相同而训练次数、学习率不同的卷积神经网络进行了缺陷检测的性能对比,验证了该卷积神经网络在学习率小于0.01时,可以获得较好的识别效果;在学习率大于0.05时,网络不容易收敛。网络训练次数越多,精度越高,相应的训练时间也较长。在满足快速性和精确度的条件下,确定了适应某印刷品的缺陷检验网络训练次数为50,学习率为0.005,此时的识别率为90%。结论 经过实验证明,该检测系统具有良好的缺陷识别能力,缺陷类型的分类准确率较高。该系统具有一定的实用价值。     

基于特征参数迁移的滚动轴承故障诊断

针对变工况下的滚动轴承无法获得大量带标签样本数据以及传统深度学习诊断方法识别率低的问题,提出一种基于迁移学习的卷积神经网络模型滚动轴承故障诊断方法.首先,采用短时傅里叶变换处理滚动轴承振动信号获得源域、目标域样本集;其次,利用源域样本预训练卷积神经网络模型;最后,通过目标域样本微调卷积神经网络模型实现滚动轴承故障诊断....     

A neural network system for shape-based classification and coding of rotational parts

The classification and coding of parts for group technology applications continue to be labour intensive and time-consuming processes. In this paper a pattern recognition approach utilizing neural networks is presented for the automation of some elements of this critical activity. As an illustrative example, a neural network system is used to generate part geometry-related digits of the Opitz code from bitmaps of part drawings. It is found to generate codes accurately and promises to be a useful tool for the automatic generation of shape-based classes and codes.     

基于卷积神经网络的典型农作物叶病害识别算法

目前,典型的一些农作物叶病害诊断主要依靠人工,但该方式耗时费力。针对大豆、棉花、水稻、小麦和玉米5类典型农作物的常见叶病害诊断问题,提出了一种基于卷积神经网络的典型农作物叶病害识别方法。从Plantvillage数据库以及其他网站收集典型农作物的叶病害图像,并对这些图像进行预处理,构建了含12 836张的数据集。参照AlexNet框架,构建8层卷积神经网络,采用迁移学习训练网络,最后通过测试集验证网络的识别准确率和损失值。分析不同的卷积神经网络的性能,实验结果表明：本算法对典型农作物的叶病害有良好的识别效果;迁移学习模式下,学习率为0.001时本算法在训练集的识别准确率约为99.47%,在测试集的识别准确率约为96.18%。     

基于改进双流卷积递归神经网络的RGB-D物体识别方法

为了提高基于图像的物体识别准确率,提出一种改进双流卷积递归神经网络的RGB-D物体识别算法(Re-CRNN).将RGB图像与深度光学信息结合,基于残差学习对双流卷积神经网络(CNN)进行改进:增加顶层特征融合单元,在RGB图像和深度图像中学习联合特征,将提取的RGB和深度图像的高层次特征进行跨通道信息融合,继而使用So...     

Improvement of the generalization capability for a pattern-recognition neural network that uses a Gaussian-synapse neuron model

A method for improving the generalization capability for optical pattern recognition by use of a Gaussian-synapse neuron model is discussed. By the dispersive effect of the Gaussian function the input images are blurred and then fed into a multilayer neural network for learning and recognition. The effectiveness of this method is demonstrated in two-dimensional shift- and scale-invariant optical pattern recognition.     

On the application of vibratory disk feeders for silos

Originally vibratory disk-feeders and-reclaimers were designed with rather small diameters. They were mainly applied in chemical process plants, and were mostly flush mounted underneath bins of modest size, say up to 1 m diameter. Over the past decades they have been increasingly used for bins of much larger diameter, some via a hopper and some not. The latter may cause an unfortunate combination of the most important parameters, leading to malfunctioning and even to feeder jam. This work shows that basically vibratory disk feeders should be installed via a hopper, which for practical reasons becomes feasible above diameters of approximately 0.8 m, depending on the material to be reclaimed and the required mass flow. Feeders of considerable size, up to 2.5 m, may very well operate satisfactorily when properly mounted underneath a hopper. Nonfunctioning of vibratory disk feeders is mostly caused by a combination of too bad flowability of the bulk material to be reclaimed; too high a feeder load and wrong silo fill procedures and feeder power-up procedures.