Incerteza de medição: uma abordagem probabilística na avaliação de projetos de eletrocardiógrafo

Abstract

Measurement uncertainty is a widespread concept applied in technical and scientific work, mainly to estimate the precision of measurement results, initially, and to evaluate the conformity of products and processes, secondly. However, in these same technical and scientific works, uncertainty analysis is little used during the design and development phases of a measurement system. In this thesis, this uncertainty is presented as a parameter that adds information and, consequently, knowledge about an object of study, initially, and as a parameter that indicates possibilities for improvements on what is studied, in a second moment. In this thesis, a probabilistic methodology is proposed that addresses measurement uncertainty as a parameter for decision-making through a systematic synthesis and analysis for electrocardiograph projects. This methodology consists of applying the Monte Carlo method to calculate measurement uncertainty as a parameter to guide improvement implementations and an adaptive filter algorithm, the NLM-SC filter, which has as a control parameter a predefined target uncertainty value. -established. This NLM-SC adaptive filter produces better results, compared to the NLM method already known in the literature, when applied to ECG signal processing. This methodology is evaluated, through computer simulation, in an electrocardiograph project. For this simulation, the computational model of the measurement system is divided into two modules: the pre-amplifier and the final stage. From the results obtained, it is concluded that the pre-amplifier module has a greater influence on the measurement results than the final stage module. It can also be concluded that the main source of ECG measurement uncertainty is related to the measurand, and using resistors with an accuracy of 0.1% reduces the uncertainty of the system as a whole from 2.13% to 0 .82%, relative to the measurement result, for a specified confidence level of 95%. The methodology proposed in this thesis is evaluated based on the measurement uncertainty applied in the measurement system design phase, contributing to the state of the art on this topic. Furthermore, the NLM-SC adaptive filter developed and the identification of strategic locations for improvement actions in the design of an electrocardiograph constitute important contributions to the area.