Methods Estimating the Degree of Soil Contamination by Heavy Metals in Precision Farming

Abstract

Створено методи грубого і тонкого оцінювання стану області ґрунту забрудненої важкими металами при точному землеробстві. Для грубого оцінювання розроблена конструкція узагальненого параметра, що враховує всі фактори забруднення. При тонкому оцінюванні ступеня забруднення розроблено принцип побудови математичної моделі об'єкта і кваліметричної шкали.

Recognition of the Type of Marine Ship Based on Comparison with Normalized Reference Parameters of Radiolocation Signals

“…Normalization allows you to obtain information about the shape or structure of the RPRS signal, regardless of the actual amplitude values, which allows you to use it in recognition tasks. Here normalization of signals means the application of the theory of reduction for a significant decrease of the input data of RPRS [2] and the formation of a vector of normalized parameters of the reference signal P = (T, S, Fc 1 , Fc 2 ,… Fc if , … Fc Nf ), where T is the duration of the RPRS signal; S is the unit of normalized signal amplitude; Fc 1 ,…,Fc Nf -discrete values of histogram-normalized amplitudes; N f is the number of duration intervals of the normalized RPRS signal.…”

“…The value of the Fс j signal form-codes can be written in symbolic form, where each value of Fс if corresponds to a hexadecimal number with alphabet '0'…'F', for example, 16 values of the vector Fс = ( 14,14,11,7,11,10,2,7,15,6,9,9,8,8,4,2) is interpreted as the character form-code 'EEB7BA27F6998842', in which the hexadecimal letters correspond to the decimal numbers: 'A' = 10, 'B' = 11, 'C' = 12, 'D' = 13, 'E' = 14, 'F' = 15. The use of symbolic form-codes can be useful for significantly reducing the volume of the database of normalized signal parameters with little time spent on unpacking real values of form-codes.…”

“…It should be noted that in most cases the results of identification of MS without a teacher are less accurate compared to methods based on training and the use of reference samples. Implementations of these two groups of algorithms are described in sufficient detail, for example, in [1,2]. Meanwhile, the identification and recognition of marine ships remains an actual problem despite the fact that the use of deep learning methods with the help of neural networks could provide higher accuracy for resolution of such tasks.…”