Applying quality control charts to the analysis of single-subject data sequences

Techniques from the field of quality control can be used to classify the quality of individual samples of physical or cognitive performance. After stable baselines have been established for an individual, deviations in performance can be evaluated using control charts. The effectiveness of this approach in evaluating cognitive performance was tested using databases collected under a variety of risk factors. The sensitivity and specificity characteristics of Shewhart, cumulative-sum (CUSUM), and exponentially weighted moving average (EWMA) control charts were determined for a total of 174 trials involving 10 participants and 23 cognitive performance assessment measures. The most effective technique in each case was typically a function of the specific performance measure and the type of performance change being evaluated. Sensitivity and specificity for the best techniques were as high as 100%. This study demonstrated the usefulness of quality control charts as a tool to evaluate individual participant performance over time. Actual or potential applications of this research include readiness-to-perform screening of industrial workers in order to improve the health and safety of the workforce.     

基于大数据车辆模拟运行及日志收集仿真研究

中国汽车保有量呈井喷式上涨,需要面对越来越多的车辆违法行为,如超速行驶、遮挡牌照、套牌、占道等等。针对这些问题,传统的车辆稽查方式显得力不从心,存在数据单一、实时性差、离线分析为主等缺点。为减轻交管部门的车辆稽查压力,构建了以大数据实时分析技术为基础的车辆模拟运行及日志收集仿真模拟系统。该仿真模拟系统采用Flume、Hadoop框架和Redis集群,参照城市街道布局设计,不仅对海量车辆的超高并发度运行和城市交通进行了模拟,还对车辆日志信息进行了实时收集。以此为依据,构建的车辆模拟运行场景,可以有效地提高车辆稽查效率。该系统为数据量较大的行业提供了一种解决问题的思路,用同样的方法可以为城市旅游、城市产业分布等进行场景模拟,进而调整旅游服务、产业调整等方案。     

Selection tests for visual inspection on a multiple fault type task

Although individual differences between inspectors is the most significant factor in differences in inspection performance there are no suitable selection tests. The task was analysed to indicate the required mental characteristics and from this a battery often selection tests was devised. These are used with a simulated task developed on a computer plotter to include the search and decision components of inspection. Presentation was by sets of slides which contained 2, 4 or 6 fault types and slides were classified as fault-free, acceptable fault, or reject. Subjects were 48 students and 18 industrial inspectors. Results were analysed by multivariate analysis of covariance and showed no difference between the student and industrial subjects. The best predictor was a simplified version of the task. Very good predictive ability was obtained with the Embedded Figures Test, and other good predictors were lobe size and mental imagery. If these tests were not used good predictive ability was obtained with a subset of the WAIS test. Eysenck's test of extroversion, and Gordon's test of Mental Imagery Control.     

Key human factors and their effects on human centered assembly performance

Today in the increasingly competitive market, consumers prefer to have a great variety of products to choose from; this preference is often coupled with demands for a relatively smaller lot size, shorter lead time, higher quality and lower cost. Consequently, manufacturing companies are being forced to consistently increase flexibility and responsiveness of their production systems in order to accommodate changes of the fluctuating market. Among various forms of production systems, human-centred manufacturing systems can offer such a capability in dealing with product variations and production volumes as human workers can always adapt themselves to perform multiple tasks after a learning process. However, human performance can also be unpredictable and it may alter due to varying psychological and physiological states, which are often overlooked by researchers when designing, implementing or evaluating a manufacturing system. This paper presents a study aiming to address these issues by exploring human factors and their interactions that may affect human performance on human-centred assembly systems. The study was carried out based on a literature review and an industrial survey. Critical system performance indicators, which are affected by human factors, were evaluated and the most significant human factors were identified using the fuzzy extent analysis method. The research findings show that experience is the most significant human factor that affects individual human performance, compared to age and general cognitive abilities in human-centred assembly. By contrast, both human reaction time and job satisfaction have the least effect on human performance. The significance of ageing on human performance was also studied and it was concluded that average assembly time of human workers rises by average 1% per year after the age of 38 years old.     

Testing the Value of Checklists in Code Inspections

Checklists are an important part of code and design inspections. Ideally, they aim to increase the number of faults found per inspection hour by highlighting known areas of previous failure. In practice, although some researchers have quantified checklists' benefits, the conclusions' statistical robustness hasn't been as well represented. The author subjects checklists' effectiveness to formal statistical testing, using data from 308 inspections by industrial engineers over a three-year period. The results showed no evidence that checklists significantly improved these inspections. Further analysis revealed that individual inspection performance varied by a factor of 10 in terms of faults found per unit time, and individuals found on average about 53 percent of the faults. Two-person teams found on average 76 percent of the faults.     

Predictive model-based quality inspection using Machine Learning and Edge Cloud Computing

The supply of defect-free, high-quality products is an important success factor for the long-term competitiveness of manufacturing companies. Despite the increasing challenges of rising product variety and complexity and the necessity of economic manufacturing, a comprehensive and reliable quality inspection is often indispensable. In consequence, high inspection volumes turn inspection processes into manufacturing bottlenecks.In this contribution, we investigate a new integrated solution of predictive model-based quality inspection in industrial manufacturing by utilizing Machine Learning techniques and Edge Cloud Computing technology. In contrast to state-of-the-art contributions, we propose a holistic approach comprising the target-oriented data acquisition and processing, modelling and model deployment as well as the technological implementation in the existing IT plant infrastructure. A real industrial use case in SMT manufacturing is presented to underline the procedure and benefits of the proposed method. The results show that by employing the proposed method, inspection volumes can be reduced significantly and thus economic advantages can be generated.     

The effects of multitasking on quality inspection in advanced manufacturing systems

Although the industrial quality inspection task has been extensively studied, the effect of multitasking on the performance of the operator in a hybrid inspection system is still unknown. The experiment described in this article compared the quality inspection performance for participants performing a single task, 3 multiple tasks, and 5 multiple tasks. The results of this research indicate that the performance of the operator in the quality inspection task while multitasking in an advanced manufacturing system will be determined by the interaction between the number of different types of defects that can be presented at the same time in the inspected parts and multitasking. The best performance will be obtained when the load created by additional tasks minimizes the monotony of the quality inspection task without interfering with the processing resources needed for the memorized quality criteria. © 2001 John Wiley & Sons, Inc.     

Feedback strategies for visual search in airframe structural inspection

Feedback of information has consistently shown positive results in human inspection, provided it is given in a timely and appropriate manner. Feedback serves as the basis of most training schemes; traditionally this has been performance feedback. Other forms of feedback which provide strategy information rather than performance information may have a role in improving inspection. This study compared performance feedback and cognitive feedback in a realistic simulation of an aircraft structural inspection task. Performance (time, errors) feedback showed the greatest improvements in performance measures. Cognitive feedback enhanced efficiency measures of search strategy. When cognitive feedback consisted of visual representations of the path and the coverage of the search sequence, subjects also were able to use this task information to improve their search performance.     

Affects on visual search performance of individual differences in fixation time and number of fixations

Abstract

The purpose of this study was to provide a basis for improving individual visual performance of inspectors. The relationship between the correct count rate and eye movements of subjects when they counted dots arranged on samples presented for different lengths of time were analysed mainly to determine individual differences. Subjects' eye movements were measured with a corneal reflectance eye camera and analysed frame by frame with a video motion analyser. It was found that accuracy of visual inspection does not depend on length of search time and that a fast search time is not incompatible with a slow search speed. Furthermore, fixation time and number of fixations were considered the main factors governing accuracy of visual inspection. When limited time is allowed for search, a search strategy of prolonging the fixation time leads to high performance and consequently shorter inspection time. Several other findings were obtained which appear important in obtaining accurate information rapidly.     

Industrial inspection performance depends on both viewing distance and oculomotor characteristics.

The influence of viewing distance and individual differences in the oculomotor resting states (dark vergence and dark focus) on both inspection performance and visual fatigue was investigated by asking 24 college students to perform a visual inspection task for 40 min twice, once with a 20 cm viewing distance and once with a 60 cm viewing distance. During the task participants visually searched enlarged images of contact lenses for six different types of defects. Lens images were presented at a constant visual angle (17.5 degrees) on a CRT. Inspection performance varied significantly across the two viewing distances: participants spent on average more than twice as much time inspecting individual lenses in the near condition (mean = 18.32 s lens(-1)) than in the far condition (mean = 7.25 s lens(-1)). Participants also reported greater visual fatigue in the near condition than in the far condition. At both viewing distances, participants who experienced relatively large inward shifts in dark vergence also tended to be slow inspectors. Neither inspection performance nor measures of fatigue were related to dark focus. These results support existing evidence against the use of near viewing distances and suggest that an oculomotor mechanism links inspection performance and visual fatigue to viewing distance.     

Does situation awareness add to the validity of cognitive tests?

OBJECTIVE: Does adding situation awareness (SA) to a battery of cognitive tests improve prediction? BACKGROUND: Identifying variables that predict skilled performance in a complex task aids in understanding the nature of skill and also aids in the selection of operators to perform that task. SA is thought to be an important predictor of performance. SA is often thought to be based on underlying cognitive mechanisms. METHOD: Three performance measures taken from the Federal Aviation Administration's (FAA's) Air Traffic Scenarios Test, the low-fidelity simulation component of the FAA's controller selection battery, were used as criterion variables in a hierarchical regression. After predicting performance based on a battery of cognitive (e.g., intelligence, working memory, spatial memory) and noncognitive tests (e.g., cognitive style, personality, demographics), we added measures of SA. RESULTS: SA did provide increases in prediction, but only when measured with the Situation Present Assessment Method, an on-line query method. When the same questions were asked off line, SA did not enter the model in two cases and improved prediction by only 2% in the third. CONCLUSION: Thus, some measures of SA do show incremental validity, even against a backdrop of a large number of cognitive variables. APPLICATION: On-line measures of SA can be a worthwhile addition to standard batteries of tests used to predict performance in cognitively oriented industrial tasks.     

Empirical assessment of the impact of structural properties on the changeability of object-oriented software

The changeability of software can be viewed as the quality of being capable of change, which among others implies that the task of changing the software requires little effort. It is hypothesized that structural properties of the software affect changeability, in which case measures of such properties can be used as changeability indicators.Ways in which structural properties of the software can be measured are described and empirically validated based on data collected from an industrial Java development project. The measures are validated by using them as candidate variables in a multivariate regression model of the actual effort required to make modifications to the evolving software system.The results suggest that some measures that combine existing structural attribute measures with a weighting factor based on the relative proportion of change in each class can explain a large amount of the variation in change effort. This constitutes initial, empirical evidence that the proposed measures are valid changeability indicators. Consequently, they may help designers to identify and correct design problems during the development and maintenance of object-oriented software.     

基于轻量化深度学习网络的工业环境小目标缺陷检测

工业环境下表面缺陷检测是质量管理的重要一环,具有重要的研究价值.通用检测网络(如YOLOv4)已被证实在多种数据集检测方面是有效的,但是在工业环境的缺陷检测仍需要解决两个问题:一是缺陷实例在表面占比过小,属于典型的小目标检测问题;二是通用检测网络结构复杂,很难部署在移动设备上.针对上述问题,提出一种基于轻量化深度学习网络的工业环境小目标缺陷检测方法.应用GhostNet替代YOLOv4主干特征提取网络,提高网络特征提取能力及降低算法复杂度,并通过改进式PANet结构增加YOLO预测头中高维特征图比例以实现更好的性能.以发动机金属表面缺陷检测为例进行实验分析,结果表明该模型在检测精度(mAP)提升5.83%的同时将网络模型参数量降低83.5%,检测速度提升2倍,同时满足缺陷检测的精度和实时性要求.     

Time-sharing revisited: Test of a componential model for the assessment of individual differences

Abstract

Time-sharing ability as an individual differences variable in dual task performance was examined using a componential model. Five proposed components were assessed: (1) serial processing ability; (2) an internal model of the system dynamics; (3) performing heterogeneous operations; (4) adaptation to rapidly changing dynamic conditions; (5) parallel processing ability. The approach combined methodologies from experimental psychology and from individual differences research. Forty subjects were given four single task pretests, and then performed a compensatory tracking task in various dual task combinations administered during six sessions over a period of three days. At the conclusion of the experiment the subjects had to perform three different dual task transfer tasks. The results of a factor analysis and a series of stepwise multiple-regression analyses revealed two important dimensions of individual differences in dual task performance: (1) individual differences in cognitive style linked to the concept of field dependence/independence; (2) individual differences in time-sharing ability. The individual differences in cognitive style were identified by the first derived factor. It was defined by the measures that represent serial and parallel processing. Based on the opposing signs, positive for the serial processing measures and negative for the parallel processing measures, it was concluded that this factor is bipolar, representing two distinct strategies. Further-analyses suggested that the strategy differences may be linked to individual differences in field dependence/independence. Individual differences in time-sharing ability were identified by the second derived factor of the factor analysis and through multiple-regression analysis. Based on the results the concept of a ‘process-specific’ time-sharing ability was introduced.     

基于ADTree改进算法的轮胎大数据质量分析

工业企业在生产制造过程中积累了大量的生产数据.海量的工业数据蕴含了价值巨大的信息,通过分析、挖掘这些工业数据能够提升企业数字化管理与质量数据分析能力.本文以轮胎行业制造大数据的应用为背景,分析了轮胎行业制造大数据的质量分析需求与数据特征,将轮胎生产各个环节的多源异构数据有效整合,经过数据预处理流程,构建了结构化的生产制造与质量检测关联分析数据集.针对传统ADTree算法性能较低的问题,本文使用优化后的自底向上的归纳方法进行了改进,充分利用已知数据,减少了建树时分裂测试评估的计算量.实验证明,改进后的ADTree算法更适用于大数据量的数据挖掘.ADTree的挖掘结果经过整理,可以找出影响轮胎质量的重要因素.     

图像FAST角点检测算法的FPGA实现与系统集成

基于Vivado HLS工具封装了包括FAST算法和AXI接口的FPGA硬件IP,并将该IP集成到工业摄像机的嵌入式系统中,工作摄像机中FAST算法的最终处理结果传输给视觉检测设备主机。利用相同图像的FAST角点检测算法对FPGA算法电路和OpenCV算法程序进行对比实验,完成相同功能的情况下前者的处理速度比后者快约60倍,该功能在工业摄像机中使数据带宽最多减小9%,显示出本文的工作可提高工业视觉检测设备的检测效率和性能。相关技术可应用于其他图像处理算法,也可推广到二维传感器阵列数据的处理。     

Automatic visual inspection: An approach with multi-instance learning

One of the industrial applications of computer vision is automatic visual inspection. In the last decade, standard supervised learning methods have been used to detect defects in different kind of products. These methods are trained with a set of images where every image has to be manually segmented and labeled by experts in the application domain. These manual segmentations require a large amount of high quality delineations (on pixels), which can be time consuming and often a difficult task. Multi-instance learning (MIL), in contrast to standard supervised classifiers, avoids this task and can, therefore, be trained with weakly labeled images. In this paper, we propose an approach for the automatic visual inspection that uses MIL for defect detection. The approach has been tested with data from three artificial benchmark datasets and three real-world industrial scenarios: inspection of artificial teeth, weld defect detection and fishbone detection. Results show that the proposed approach can be used with weakly labeled images for defect detection on automatic visual inspection systems. This approach is able to increase the area under the receiver-operating characteristic curve (AUC) up to 6.3% compared with the naïve MIL approach of propagating the bag labels.     

Investigation of guessability of industrial safety signs: Effects of prospective-user factors and cognitive sign features

This study investigated the effects of prospective-user factors and five cognitive sign features on guessability of safety signs. Sixty Hong Kong Chinese subjects guessed the meanings and rated the sign features of 63 Mainland China safety signs. The prospective-user factors of working or site visit experience at laboratory or construction site, awareness of safety signs in daily life, injury experience due to ignorance of safety signs, and driving experience had significant effects on guessing performance. However, gender, Mainland China visit experience, and family member working at laboratory or construction site had no influence on guessing performance. The guessability scores varied with the five cognitive sign features viz. familiarity, concreteness, simplicity, meaningfulness, and semantic closeness. The guessability of safety signs was better when the signs were familiar, concrete, simple and meaningful and when the signs were associated with the underlying concepts. Overall, the findings showed that prospective-user factors and cognitive sign features are involved in effective communication of safety sign messages.

Relevance to industry

Posting of safety signs is a possible safety precaution measure that can be taken to attempt to reduce accidents and injuries in industrial undertakings. The findings of this study provide useful information for designing more user-friendly safety signs which could act as a reference for interface designers to develop more user-friendly safety signs in the future.     

An exploration of factors that impact individual performance in an ERP environment: an analysis using multiple analytical techniques

This study explores the factors that can impact individual performance when using enterprise resource planning (ERP) systems. Starting from the proposition that organizational performance depends on individuals' task accomplishments, we test a structural model of task–technology fit, ERP user satisfaction, and individual performance in ERP environments. This research utilizes a survey method to examine the perceptions of ERP users. We performed factor and reliability analyses to assess the validity of the survey instrument. Six factors were identified as having an impact on individual performance: System Quality, Documentation, Ease of use, Reliability, Authorization, and Utilization. To explore the relationships among these factors, we conducted regression and multivariate adaptive regression splines analysis, and compared the findings from these two analytical techniques. The study provides evidence that System Quality, Utilization, and Ease of Use are the most important factors bearing on individual performance in ERP environments. Our findings also provide IT managers and researchers with knowledge of how these factors can be manipulated to improve individual performance when using ERP systems.