Metrological evaluation of heterogeneous surfaces obtained by water jet cutting technology using artificial intelligence elements

Abstract

This paper deals with the design and construction of a neural network for predicting the results of roughness parameters for heterogeneous surfaces. At the same time, it demonstrates that other statistical methods, especially regression analysis, fail in this respect, and their results cannot be used reliably. The samples produced using waterjet cutting were used to obtain the necessary data for constructing the neural network. Its heterogeneity characterizes this surface. This paper describes these samples, the parameters of their creation, the laboratory measurements, the complete construction of the neural network and the subsequent comparison of the results with regression functions. This paper aims to design a functional neural network that will best describe the roughness pattern of the surface under study. This neural network will predict this waveform based on the input variables and prove that this advanced statistical method completely exceeds the capabilities and predictive value of conventional regression analyses.