Data arrangements to train an artificial neural network within solving the tasks for calculating the Chézy roughness coefficient under uncertainty of parameters determining the hydraulic resistance to flow in river channels

Hydraulic calculations and mathematical modelling of open flows in river channels keep still being among the most topical hydro-engineering today’s problems terms practice. While solving them, independently on research topic purpose, methods used, a number simplifications assumptions are usually accepted applied. Moreover, there is range methodological, structural, parametric uncertainties, which to be overcome require complex empirical pre-researches. First all, these uncertainties relate assessing hydraulic resistances establishing numerical characteristics depend many factors varying spatially temporally.One frequently used integral expressing resistance Chézy roughness coefficient C. However, despite large semi-empirical formulas dependencies calculate coefficient, no ideal way or method determine this characteristic unambiguously. On one hand, while opting for an appropriate formula we need take into account practical experience based comprehensive options analysis considering different equations alternatively represent flows. other fullness comprehensiveness field researches numerous hydro-morphological parameters characterizing various aspects can also have essential role. In particular, accuracy assessment computing data, formulas, indicates that measurements included selected largely determines relative error such calculations.This paper deals with problem data arrangements development general rules formation training test samples train artificial neural networks elaborated compute taking uncertainty channels. The solved example network direct propagation hidden layer sigmoid logistic activation function.