IMEKO Event Proceedings Search

Page 3 of 891 Results 21 - 30 of 8905

Viola Gallina, Zsolt János Viharos, Gabor Nick, Maik Frye, Andreas Kluth, Adam Szaller, Robert H. Schmitt
Factory of the Year Prize – A Benchmarking

The first factory of the year prize was granted more than 60 years ago in the USA. Since then, a considerable number of countries joined this way and several best factory assessment methods and awards have been developed on national, regional, and international levels. These competitions give the possibility for benchmarking the companies. However, in the era of industry 4.0 maturity models have emerged for evaluating individual enterprises' readiness. These models support the companies in the individual strategy development. But the companies are always interested in their results and achievements compared to their competitors. But setting up a benchmark for a part of an industrial sector might be challenging. Therefore, combining the factory of the year evaluation concept with the maturity assessment might be advantageous. In the paper both of the approaches are analysed and it is discussed how they might be linked in a meaningful way.

Gabriele Patrizi, Alessandro Bartolini, Lorenzo Ciani, Marcantonio Catelani
Failure analysis of a smart sensor node for precision agriculture

Nowadays, the use of big data analysis and IoT (Internet of Things) technologies is growing tremendously within companies and organizations in several different application fields. In this scenario, smart farming refers to monitoring of environmental conditions and soil parameter to improve farm productivity, to optimize soil conservation, to save water and to limit plant diseases. During the design of such innovative IoT technologies it is fundamental to carry out a reliability and failure analysis of the device. This could allow to introduce adequate diagnostic solutions to improve the system’s availability. In this work, a failure analysis using FMEA (Failure Modes and Effects Analysis) approach of a smart sensor node for precision farming has been developed. The results of the analysis allow to improve the design of the device introducing a diagnostic-oriented prototype able to solve the major criticalities arisen during the FMEA.

Paul-Alexander Vogel, Anh Tuan Vu, Hendrik Mende, Shrey Gulati, Tim Grunwald, Robert H. Schmitt, Thomas Bergs
Machine learning-based predictions of form accuracy for curved thin glass by vacuum assisted hot forming process

Thin glass is applied in numerous applications, appearing as three-dimensional smartphone covers, displays, and in thin batteries. Nonisothermal glass molding has been developed as a hot forming technology that enables to fulfil demands of high quality yet low-cost production. However, finding optimal parameters to a new product variant or glass material is highly demanding. Accordingly, manufacturers are striving for efficient and agile solutions that enable quick adaptations to the process. In this work, we demonstrate that machine learning (ML) can be utilized as a robust and reliable approach. ML-models capable of predicting form shapes of thin glass produced by vacuum-assisted glass molding were developed. Three types of input data were considered: set parameters, sensor values as time series, and thermographic in-process images of products. Different ML-algorithms were implemented, evaluated, and compared to reveal random forest and gradient boosting regressors as best performing on the first frame of the thermographic images.

D. Buonocore, M. Carratù, M. Lamberti
Classification of coffee bean varieties based on a deep learning approach

In this article, a coffee beans fraud detection based on a deep learning approach is proposed, which has been achieved after classifying the two coffee varieties to distinguish them in a real-time industrial scenario. The coffee bean quality is typically defined by visual inspection, which is subjective, needing significant effort and time, and susceptible to fault detection. For these reasons, a different method is required to be objective and precise. Therefore, object detection techniques were employed to automatically classify the coffee bean samples according to their specie using an own dataset consisting of over 2500 coffee beans. Furthermore, a convolutional neural network (CNN) based on the YOLO algorithm was employed to categorize the coffee beans automatically. The result of this study has revealed that the object detection technique could be used as an effective method to classify coffee bean species and discover food fraud.

Balázs Scherer
Card Level Management Solution for the BRAINE edge framework

The objective of the BRAINE (Big data pRocessing and Artificial Intelligence at the Network Edge) project is to boost the development of an Edge framework focusing on energy efficient hardware and AI empowered software, capable of processing Big Data at the Edge, supporting security, data privacy, and sovereignty. The BRAINE’s Edge framework is designed to be flexible and support many types of data processing and storage cards. Because of this flexibility and the wide range of heterogeneous cards a sophisticated low-level board/card management functionality is needed. The board management architecture of BRAINE is a hierarchical design, which consists of a central board management node (BMC: Board Management Controller) and a distributed low- level board management hardware present on all the nodes (BMMC: Board Management MicroController). This paper describes the functionalities and presents implementation details of the low-level board management controller the BMMC.

A. Vulić, G. Bajić, G. Vukoslavović, M. Radević, D. Vidaković, J. Šetina

This paper focuses on the process of establishing the Laboratory for Pressure at the Montenegrin Bureau of Metrology (MBM), the National Metrology Institute of Montenegro, which disseminates the unit of pressure in the field of calibration and conformity assessment in a small, centralized metrology system. Establishment of a traceability chain, development and the continual improvement of measurement methods, participation in interlaboratory comparisons and PT schemes, accreditation according to ISO/IEC 17025, registration of calibration and measurement capabilities (CMC) in BIPM KCDB for pressure, are stages in the development of the Laboratory for Pressure and also in proving its competencies at the international level.

F. Boineau, M. D. Plimmer

Calibrating leak rates using a constant-volume flowmeter consists in measuring the pressure rise rate produced by the leak throughput in a measurement volume. When the leak rates to be determined are low, the latter should be as small as possible, and since it is composed of dead volumes of the tubing, the valves, the gauge, etc., one needs to determine this dead volume accurately. This paper describes an original and simple technique which provides a relative uncertainty in the calibration of the dead volume of a constant-volume flowmeter (~ 60 cm³) of approximately 0.1 % (k = 1). In turn, the bottom end of the calibration range of the flowmeter has been brought down to about 5 × 10-8 Pa∙m³∙s-1.

C. Marchesi, M. Rani, S. Federici, M. Lancini, L. E. Depero

The plastic recycling industry necessitates fast and reliable methods to recognize the different polymer types to improve the recycling capacity. In this contribution, the coupling of a miniaturized Near-Infrared (NIR) spectroscopy technique with a robust data analysis is presented. Comparison of multiple machine learning techniques, such as Support-Vector Machines (SVM), Fine Tree, Bagged Tree, and Ensemble Learning, and chemometric approaches, such as Principal Component Analysis (PCA) and Partial Least Squares – Discriminant Analysis (PLS-DA), were combined to provide a broad overview and a rational means for selecting the approach in analysing NIR data for plastic waste sorting.

M. Cundeva-Blajer

Decisions in conformity assessment, especially in the testing, inspection and certification (TIC) sector are predominantly based on the measurement data. The digital transformation has started to impact the TIC sector. The quality of the data in big data analytics, is not always sufficiently addressed, especially in sectors with traditionally empirical approaches, as TIC. This contribution conducts a survey of possibilities for application of data science in the TIC decision making processes, based on conclusions with complementary usage of experimental “measurements” and the “data science”, with a case study in estimation of the instruments recalibration interval with data fusion approach.

B. Petrovic, S. Strbac

Personal Dosimetry Laboratory of the Public Health Institute of Republic of Srpska (PHIRS) is estimating doses to approximately 1500 occupationally exposed workers to ionizing radiation on annual basis which represents 1/3 of occupationally exposed workers of Bosnia and Herzegovina. It is well known fact that the impact of ionizing radiation to human body and the risk of severe adverts effects are correlated with the magnitude of exposure. Therefore, delivering precise and accurate measurement result is essential in control of occupational exposure. Inter-comparison exercise plays an important role in testing laboratories to demonstrate technical competence. Participation in globally acknowledged and recognized EURADOS inter-comparison exercise for whole body dosemeters in photon fields is being conducted in 2 – years cycle by the laboratory. This paper is focused on the analysis of the EURADOS inter-comparison results. From 2014 to 2020, in total 4 inter-comparison exercises have been conducted, each requesting circa 30 dosemeters to be labelled and dispatched for irradiations. Entire number of the dosemeters is divided into groups for irradiations to a reference values with different irradiation qualities, dose ranges, angle of incidence and one group is kept to monitor the background. Statistical analysis of measured quantity Hp(10) has been reported with response values (arithmetic mean, median, minimum, maximum and coefficient of variation). All response values are inside the trumpet curve defined in ISO 14146, ranging in mean response from 0.85 to 0.92. In addition, results are compared to En value and estimated uncertainty budget. This demonstrates that measured results are in compliance with recommendations on accuracy for radiation protection purposes.

Page 3 of 891 Results 21 - 30 of 8905