Tytuł pozycji:
Possibilities of neural networks using in the design, simulation and verification of the controlled gas nitriding process
The article presents the possibility of neural networks application to design and simulate the growth kinetics of class 1 nitrided layers in steel 32CDV13 and 42CrMo4 (40HM), using data obtained from analytical models. The study analyses unidirectional multilayer neural networks with one hidden layer, with approximation properties. The algorithm developed takes into account the average thickness of the layer of iron nitride. This parameter is most frequently used for the classification of nitrided layers, especially for anticorrosion layers. As a result of research and discussion stated: the neural networks with approximating properties used allowed to build models, well-fitted to the data obtained using analytical models, taught structures of neural networks can be used in systems estimating the results of the nitriding process. The duration of the first stage of the process and the value of the potential in the second degree determine the thickness of the iron nitride layer obtained after the nitriding process. The value of the potential in the second stage also determines the intensity of limiting the thickness of the iron nitride layer. Nitriding decreases the impact strength of steel regardless of the thickness of the subsurface iron nitride layer. The iron nitride layer on the steel increases its resistance to frictional wear. Its resistance to friction wear increases with the increase of the thickness of this layer.
Celem pracy była ocena możliwości wykorzystania sieci neuronowych do projektowania i symulacji kinetyki wzrostu warstw azotowanych na stalli 32CDV13 i 42CrMo4 (40HM), wykorzystując dane uzyskane z modeli analitycznych. W części weryfikacji eksperymentalnej, określenie udarności i odporności na zużycie przez tarcie tych stali po procesie regulowanego azotowania gazowego.
Opracowanie ze środków MNiSW w ramach umowy 812/P-DUN/2016 na działalność upowszechniającą naukę (zadania 2017).
