Využití umělé inteligence v diagnostice kompozitních materiálů

Abstract

The thesis proposes the possibility of using elements of artificial intelligence to the prognosis of natural frequencies of laminated plates. Experimental data were obtained from measurements of frequency and size of plates using ESPI method. The proposed method allows the prognosis data without time-consuming measurements. The thesis further deals with the model using neural networks, which describes the oscillations of the tires on the basis of data obtained from the labeling of tires and other metrics. Part of the thesis deals with the synthesis method of neural networks using genetic algorithms, especially from the point of view of speeding up and improving the process of iteration using massive parallelization on GPU.