DEELT - Departamento de Engenharia Elétrica

URI permanente desta comunidadehttp://www.hml.repositorio.ufop.br/handle/123456789/5266

Navegar

Agora exibindo 1 - 2 de 2

Applied sensor fusion : tuning parameters of CF and KF by means of evolution strategies.
(2020) Santos, Talysson Manoel de Oliveira; Barroso, Márcio Falcão Santos; Ricco, Rodrigo Augusto
From the point of view of metrology, inertial sensors acting separately present undesirable performance in the measurement of angular position. In order to provide measurements with greater precision and accuracy, the measures of each of these sensors are typically fused by means of filters. The performance parameters of these filters are hard to tune and several works have been using exhaustive search algorithm or manual experimental tests to tuning these parameters. However, the exhaustive search algorithm usually requires a large computational effort and adjusting parameters manually does not guarantee that the estimated parameters are optimized. In this work, it has been proposed the tuning of the Complementary Filter (CF) and the Kalman Filter (KF) through the heuristic method Evolutionary Strategies. Experimental results have shown that our method is a useful tool that considerably reduces the time to find the tuning of the FC and the FK. In addition, the use of the tuned FC and FK improved significantly metrological characteristics of the system. The use of Bland and Altman’s statistical method show that the measurements of the angular position have a good agreement with the actual angular position of a servo motor.
A low-cost wireless system of inertial sensors to postural analysis during human movement.
(2019) Santos, Talysson Manoel de Oliveira; Barroso, Márcio Falcão Santos; Ricco, Rodrigo Augusto; Nepomuceno, Erivelton Geraldo; Alvarenga, Érika Lorena Fonseca Costa de; Penoni, Álvaro C. O.; Santos, Ana F.
The dynamics of the human body has generated considerable recent research interest among scientist devoted to reducing the number of injuries and for performance improvement. In these studies, the investigation is usually addressed by means of commercial devices based on video recordings. However, these systems based on video recordings are usually expensive and require suitable laboratories for their use, which makes it unfeasible to collect data for activities outside controlled environments. In this work, we have shown that it is possible to present similar results with a much lower sampling rate, focusing on the evaluation of minimum and maximum values of the gait. As a result, it has been possible to develop a wearable, compact, portable, low-cost, wireless and embedded system to simultaneously analyze the three-dimensional angular position in eight points. This technology can be used in many sorts of environments. It is also possible to access information in real time with reliable and accurate measurements by means of simple modelling for the use of fusion techniques implemented in the microcontroller. Tests were conducted to evaluate the metrological characteristics of the system using the Complementary Filter (CF) and the Kalman Filter (KF). An algorithm of evolutionary strategies tuned both filters, providing errors of less than 5% for static situations in the measurement of the angular position over the entire system utilization range. Our results have been compared with the commercial system Qualisys Motion-Capture. The statistical method elaborated by Bland and Altman has been used. We have found our method yields a motion analyses in good agreement with results using post-processed video.