Bearings have a vital role in nearly all rotating machines. Making adequate bearings is very important that satisfy all needs which emerge both in manufacturing and during operation. In former times bearings were examined by only humans, however human inspection is instable and time consuming. In this article, we are investigating a machine learning system that could make more accurate measurements regarding geometry, shape, color, surface defects, deformations and other failures by image acquisition. To achive higher resolution, magnifying of the surface with optical microscopes and scanning electron microscope (SEM) is inevitable. With these methods even tiny failures can be detected. Machine learning methods have beeen developed such as artificial neural networks (ANN) and support vector machines (SVM). Bearing manufacturing failures, image processing techniques are presented in this article besides artificial neural network system that can percieve manufacturing defects approximately 90% efficiency according to our experiments. Recent research is connected to a manufacturing of bearings in a real company in Hungary.
Az EFOP–3.6.1–16–2016–00022 számú, „Debrecen Venture Catapult Program” elnevezésű pályázat „Műszaki kutatói kapacitás bővítése, kutatási szolgáltatások fejlesztése, tudásnégyszög kiépítése a mérnökképzésben” című alprogramjában alakult egy „Mérnöki és innovációs készségeket fejlesztő kutatócsoport”. A kutatócsoport vállalta, hogy készségfejlesztő foglalkozásokat dolgoz ki középiskolások számára matematika, fizika, ábrázoló geometria és informatika témakörökhöz kapcsolódóan. Jelen cikk a számos kifejlesztésre került foglalkozás közül a „Játékos algoritmizálás” című foglalkozást mutatja be, mely foglalkozás során számítógépes játékkal játszva fejlesztjük a diákok algoritmizálási készségét.
This work aims to demonstrate that machine learning is suitable for solving engineering problems. In this work Support Vector Machine (SVM) method has been chosen to investigate a specific engineering optimization problem.Engineering rubber parts must have a predefined load-displacement characteristics under load. This characteristics depends on three parameters of the geometric shape. The aim is to find the rubber part geometry resulting the minimum difference between the desired and the initial load-displacement characteristics. This area can be named as the difference of work (∆W). The objective is to minimize this area, therefore we have to find a regression function for delta work.SVM can be used for classification and regression as well. The linearly non-separable cases can be solved by using the "kernel trick", and using an ε-insensitive loss function the regression task can be solved, too. The application of the method has 4 steps. First, we determine the training points and generate the set of training data. Then we choose the Kernel function and determine hyper-parameters. We search the optimal hyper-parameters using grid search and n-fold cross-validation to avoid overfitting. Finally, we executing the regression with the optimal hyper-parameters, and we get the resulted regression function.In this example 27 training points have been selected and we used Radial Basis Function kernel. After the grid search we get the optimal value of the three hyper-parameters. With the resulted regression function can we calculate the minimum of the work difference and finally we can determine the optimal geometric parameters.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.