El rehabilitasyonu için geliştirilen harici iskeletin optimal tasarımı
Loading...
Date
2019
Authors
Ertan, Hulusi Bülent
Mohammadzadeh, Mohammad Hassan Gol
Journal Title
Journal ISSN
Volume Title
Publisher
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
Bu tez çalışması, robotik el rehabilitasyon sistemi tasarım ve uygulamasını hedefleyen bir proje kapsamında gerçekleştirilmiştir. Beyin damarlarını etkileyen kazaların yol açtığı sakatlıklar hemiplejik olduğundan, projede işlev kaybı olan elin dış iskelet tarafından yönlendirildiği bir sistem üzerine yoğunlaşılmıştır. Bu çalışmada robotik harici iskelet, görsel uyaran yazılımı ve ayna nöron sistemi aktivitesini gözlemleyen ölçüm sistemi ile senkronize bir biçimde çalışmaktadır. Bu tez kapsamında harici iskelet mekanizmasının, ileri ve tersine aktarım açıları ve hedeflenen kinematiği kullanan bir çoklu amaç fonksiyonu ile eniyilenmesi hedeflenmiştir. Çoklu gövde dinamiği modelleri ve sanki-statik yapıdaki kinetik modeller, tasarım ve benzetim amaçlı geliştirilmiş ve kullanılmıştır. Projemizde insan parmağının eklemlerindeki pasif torklar tam ve kesirli mertebe diferansiyel denklemlerle modellenmiştir. Parmağın dış iskelet ile bütünleşik benzetimlerinde tam ve kesirli mertebeden pasif torklar uygulanmıştır. Kesirli mertebe torklar, parmakta spastisite sonucu oluşan anomalinin modellenmesi için önerilmiştir. İnce kavrama hareketini sağlamak amacıyla PID ve FEL tipi denetimciler tasarlanmış ve model üzerinde sınanmıştır. Uyarlamalı bir kontrolcü olan FEL'in, kesirli mertebe pasif torklar etkin olduğunda PID'den daha başarılı sonuçlar verdiği gözlenmiştir. Direkt ve tersine sanki-statik modellerin, pasif torkların kestirimine yönelik kullanımı da tez kapsamında sunulmuştur.
This thesis is a part of the research project which aims to design and implement a robotic hand rehabilitation system. The disabilities caused by the cerebral vascular accidents are hemiplegic. Therefore, the system is designed to make the impaired hand be driven by the exoskeleton. This system consists of a robotic hand exoskeleton which is synchronized with the visual stimulus software and the monitoring system for the activity of the mirror neuron system. Focus of this thesis is on optimizing the exoskeleton mechanism using a multi-objective cost function in terms of the forward and backward transmission angles and the desired kinematics. Mathematical models based on the multibody dynamics approach are used for the design and simulation purposes. In addition, the quasi-static models are utilized. The passive torques in the joints of the human finger are modeled with the ordinary and fractional order mathematical terms. In the simulations, both the integer and the fractional order passive torques are implemented. The fractional order model is mainly used to represent the anomaly due to the spasticity. Two control strategies, namely the Proportional-Integral-Derivative (PID) and Feedback Error Learning (FEL) types are designed and evaluated with simulations to control the exoskeleton system during the pinching motion. It is shown that the adaptive controller, FEL, copes with the fractional order passive torques better than the PID controller. It is shown that the inverse and direct quasi-static models are used to estimate the passive torques.
This thesis is a part of the research project which aims to design and implement a robotic hand rehabilitation system. The disabilities caused by the cerebral vascular accidents are hemiplegic. Therefore, the system is designed to make the impaired hand be driven by the exoskeleton. This system consists of a robotic hand exoskeleton which is synchronized with the visual stimulus software and the monitoring system for the activity of the mirror neuron system. Focus of this thesis is on optimizing the exoskeleton mechanism using a multi-objective cost function in terms of the forward and backward transmission angles and the desired kinematics. Mathematical models based on the multibody dynamics approach are used for the design and simulation purposes. In addition, the quasi-static models are utilized. The passive torques in the joints of the human finger are modeled with the ordinary and fractional order mathematical terms. In the simulations, both the integer and the fractional order passive torques are implemented. The fractional order model is mainly used to represent the anomaly due to the spasticity. Two control strategies, namely the Proportional-Integral-Derivative (PID) and Feedback Error Learning (FEL) types are designed and evaluated with simulations to control the exoskeleton system during the pinching motion. It is shown that the adaptive controller, FEL, copes with the fractional order passive torques better than the PID controller. It is shown that the inverse and direct quasi-static models are used to estimate the passive torques.
Description
Keywords
Mekatronik Mühendisliği, Mechatronics Engineering
Turkish CoHE Thesis Center URL
Fields of Science
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
0
End Page
95