Wyniki wyszukiwania - BazTech

1

Application of Artificial Neural Networks in the Analysis of Mechanisms Destroying Forging Tools

Hawryluk M., Mrzygłód B., Gronostajski Z., Głowacki M., Olejarczyk-Wożeńska I.

Archives of Metallurgy and Materials

|

2020

|

Vol. 65, iss. 1

193--200

EN

This article discusses the results of studies using the developed artificial neural networks in the analysis of the occurrence of the four main mechanisms destroying the selected forging tools subjected to five different surface treatment variants (nitrided layer, pad welded layer and three hybrid layers, i.e. AlCrTiSiN, Cr/CrN and Cr/AlCrTiN). Knowledge of the forging tool durability, needed in the process of artificial neural network training, was included in the set of training data (about 800 records) derived from long-term comprehensive research carried out under industrial conditions. Based on this set, neural networks with different architectures were developed and the results concerning the intensity of the occurrence of thermal-mechanical fatigue, abrasive wear, mechanical fatigue and plastic deformation were generated for each type of the applied treatment relative to the number of forgings, pressure, friction path and temperature.

2

Low cycle fatigue studies of WCLV steel (1.2344) used for forging tools to work at higher temperatures

Hawryluk M., Dolny A., Mroziński S.

Archives of Civil and Mechanical Engineering

|

2018

|

Vol. 18, no. 2

465--478

EN

The paper presents the results of low cycle fatigue studies of WCLV tool steel aimed at determining the diagrams of low cycle fatigue and verifying them based on the analysis of durability of the selected forging tool. Based on the macro- and microstructural tests as well as numerical modelling of industrial forging processes, the conditions for the implementation of laboratory fatigue tests were determined. The samples underwent periodical uniaxial loading (tension–compression conditions) at four levels of amplitude of total strain (0.5; 0.8; 1.0; 2.0%), at three temperatures (20, 300 and 600 °C), based on the macro- and microstructural tests as well as numerical modelling of industrial forging processes. For the determination of the hysteresis loop based on the obtained fatigue results, the Ramberg–Osgood equation was applied. The fatigue diagrams in the bilogarithmic system were approximated by the Manson–Coffin–Basquin equation, and the diagrams of low cycle fatigue in the ɛa–Nf and σa–Nf system were obtained. The results of the laboratory tests concerning the fatigue strength of WCLV tool steel were preliminarily verified based on the analysis of the punch used to forge a lid forging, exhibiting a good agreement. The performed comparative analysis of the results of the fatigue tests and the numerical analysis combined with the studies of the microstructure revealed the possibility of their application in the aspect of forging tools’ durability as well thermo-mechanical fatigue strength.

3

Durability analysis of forging tools after different variants of surface treatment using a decision-support system based on artificial neural networks

Mrzygłód B., Hawryluk M., Gronostajski Z., Opaliński A., Kaszuba M., Polak S., Widomski P., Ziemba J., Zwierzchowski M.

Archives of Civil and Mechanical Engineering

|

2018

|

Vol. 18, no. 4

1079--1091

EN

This article concerns a decision-support system based on artificial neural networks (ANN) enabling analysis and forecasting of the durability of forging tools used in the hot forging process of a cover forging – a sealing element of the driveshaft in road freight vehicles. The process of knowledge acquisition, adopted neural network architecture and parameters of the developed network are presented. In addition, 3 variants of a hybrid layer (gas nitrided layer GN + PVD coating) were applied to the selected tools (punches applied in the 2nd top forging operation): GN/AlCrTiN, GN/AlCrTiSiN, and GN/CrN, in order to improve durability, and the resultant tools were also compared to standard tools (with only gas nitriding) and regenerated tools (after repair welding regeneration). The indispensable knowledge about the durability of selected forging tools (after various surface engineering variants), required for the process of learning, testing and validation for various neural network architectures was obtained from comprehensive, multi-year studies. These studies covered, among other things: operational observation of the forging process, macroscopic analysis combined with scanning of tools’ working surfaces, microhardness measurements, microstructural analysis and numerical modeling of the forging process. The developed machine-learning dataset was a collection of approx. 900 knowledge records. The input (independent) variables were: number of forgings manufactures, pressing forces, temperature on selected tool surfaces, friction path and type of protective layer applied to tool. Meanwhile, output (dependent) variables were: geometrical loss of tool material and percentage share of the four main destructive mechanisms. Obtained results indicate the validity of employing ANN-based IT tools to build decision-support systems for the purpose of analyzing and forecasting the durability of forging tools.

4

A durability analysis of forging tools for different operating conditions with application of a decision support system based on artificial neural networks (ANN)

Hawryluk M., Mrzygłód B.

Eksploatacja i Niezawodność

|

2017

|

Vol. 19, no. 3

338--348

PL

W pracy przedstawiono wyniki badań, dotyczące, procentowego udziału mechanizmów niszczących dla dwóch typowych wariantów eksploatacji narzędzi kuźniczych: smarowanych i chłodzonych oraz bez smarowania. Prezentowane wyniki pochodzą z opracowanego przez autorów systemu wspomagania decyzji (SEPEK-2) działającego w oparciu o sztuczną sieć neuronową. Wiedza o analizowanym zagadnieniu trwałości narzędzi kuźniczych, potrzebna do procesu uczenia sztucznej sieci neuronowej zawarta była w zestawie danych uczących, pochodzących z kompleksowych badań, zrealizowanych w warunkach przemysłowych. Zestaw danych uczących obejmował zbiór 450 rekordów wiedzy. W pracy przestawiono proces pozyskiwania wiedzy, przyjętą architekturę sieci neuronowej oraz parametry opracowanej sieci. Przeprowadzona globalna analiza wyników generowanych przez opracowany system, dla trwałości traktowanej jako zwiększająca się liczba odkuwek (od 0 do 25000 sztuk), wykazała że dla narzędzi smarowanych i chłodzonych dominującym mechanizmem jest zmęczenie cieplno-mechaniczne, a nie zużycie ścierne, które rzeczywiście dominuje w procesach kucia dla narzędzi niechłodzonych i niesmarowanych. Należy podkreślić, że zdecydowana większość opracowań z tego obszaru przypisuje, że to zużycie ścierne jest dominujące, a jak wykazały wyniki badań i analiz dla wybranych reprezentatywnych procesów kucia, przy wykorzystaniu systemu wspomagania decyzji opartego o SNN, to zmęczenie cieplno-mechaniczne jest dominujące w tych procesach. Jednakże ze względu na łatwą mierzalność oraz popularnie stosowane modele zużycia ściernego, bazujące na modelu Archarda, to właśnie zużyciu ściernemu przypisuję się największy udział, choć w rzeczywistości dla narzędzi smarowanych i chłodzonych zmęczenie cieplno-mechaniczne wzmaga ów efekt przypisywany zużyciu ściernemu. Natomiast ogólnie przyjęty pogląd jest słuszny, w przypadku narzędzi niesmarowanych. co potwierdziły także analizy przy wykorzystaniu SNN.

EN

The paper presents the results of research concerning the percentage participation of destructive mechanisms for two typical variants of exploitation of forging tools: lubricated and cooled, and without lubrication. Discussed results come from the developed by the authors the decision support system (SEPEK-2) based on artificial neural network. The knowledge about the durability of forging tools needed for learning artificial neural network was included in the training data set, from comprehensive studies, carried out in industrial conditions. Set of training data set included 450 records of knowledge. The paper presents the process of acquiring knowledge, adopted neural network architecture and parameters developed network. Carried out a global analysis of the results generated by the developed system for the durability of forging tools treated as the maximum number of produced forgings to their destruction (from 0 to 25,000 items), showed that for the lubricated and cooling tools the dominant mechanism is thermo-mechanical fatigue, and do not abrasive wear, which actually dominates in the process of forging tools for uncooled and unlubricated tools. It should be emphasized that the overwhelming majority of studies in this area is attributed that to abrasive wear is dominant, and as shown by the results of research and analysis for the selected representative forging processes, with the use of decision support system based on ANN, the fatigue a thermo-mechanical is dominant in these processes. However, due to the easy measurability and commonly used models wear, based on the model of Archard, it is abrasive wear assigned the largest participation. In fact, for the tool lubricated and cooled tools a thermo-mechanical fatigue intensifies this effect attributed to abrasive wear. While the generally accepted view is correct, in the case of tools unlubricated, as confirmed by the analysis using ANN.