Predicting and Determining the Time Between Metrological Failures of Smart Systems for Precision Farming

Abstract

Introduction. Solving the problem of forecasting and determining the operating time for metrological failure and conducting the first calibration of smart systems of precision land use is possible by solving the problem of self-calibration of smart sensors that are part of these smart systems (SS). This problem is solved and described in [1]. The purpose of the paper is the methodology for dynamic prediction and determination of the time between metrological failures (MF) and the first verification of SS designed for precision farming. Results. The article describes a method patented in Ukraine for measuring the SS operating time for a MF (dynamic prediction method) based on a synthesized probabilistic-physical model (PP-model) of SS MF described by a multi-parameter Kondratov – Weibull distribution function (DF) with controlled (flexible) parameters. The proposed model describes the relationship between the normalized error and the parameters of the metrological reliability (MR) of the SS. It is shown that the dynamic regression PP-models of MF are a combination of the capabilities of regression models using flexible multi-parameter DF, with the possibility of using dynamic (spatio-temporal) processes covering different trends in the change in the values of normalized errors and their uncertainty bands, confidence level, time frame, acceptable boundary conditions, etc. Dynamic regression models of MF SS make it possible to understand the relationship between DF variables and allow the possibility of studying metrological problems (“scenarios”) of the “what if …” type. The dynamic regression method is a set of techniques for reciprocating approximation of the values of the shift parameter of the dynamic PP-model of MF to the predicted value of the shift parameter of the static PP-model of MF SS, as well as methods for assessing the reliability and accuracy of forecasting and determination. The article describes the essence of a new method for determining the operating time of the SS in the MF using the PP-model of the MF based on the Kondratov – Weibull DF. For the first time, a graphical portrait of the PP-model of SS metrological failures in the combined system of scales (coordinates) has been developed and presented - with the scales "probability of metrological failure Pξ" and "normalized error ξx" and separate or combined scales of "interval time scale tx " and "calendar time scale". The procedure for determining the time of the first verification is described, the advantage of non-periodic verifications is noted in order to save costs for their implementation. The possibility of occurrence of "conditional misses" in determining the error and time of operation on the MF during one or another verification is shown. Their existence is established only after the subsequent verification, analysis of the obtained data, and drawing the curve of the DF on a graphical portrait. It is recommended to choose the time between verifications as a multiple of one year, and to carry out verifications on the same day and month of the year. Conclusions. The dynamic regression method is an effective and versatile method due to the high accuracy of forecasting and determining the operating time in the MF. It can also be implemented using MF PP- models based on the DF of Kondratov - Cauchy, Kondratov - Laplace and others. Keywords: smart sensor, self-calibration, wireless sensor systems, methods of redundant measurements, problems of metrological support.