Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.784716
Title: Upper extremity electromyography signal feature extraction and classification
Author: Bin Wan Daud, Wan Mohd Bukhari
ISNI:       0000 0004 7970 2637
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The human upper forearm (UFA) consists of several muscles. These muscles are used by researchers to identify the hand movements based on their acquired EMG signal. The precision of EMG signal features and parameters proportionally vary with muscle signal, features and fatigue. The major challenge for the study is to identify fundamental manifestation in the EMG signal in muscle localisation in the human UFA regions, so that EMG based control performance is improved. This can be achieved through improvement of data collection, features extraction and classification. Hence, a fundamental study is performed by investigating the signals acquired from the human UFA to discover muscle characteristics and to establish the inter-relationship between the forearm and upper arm muscles. The principal objective of this study is to investigate the research challenges stated earlier via non-invasive EMG acquisition. Therefore, experimental protocols for data collection are designed to achieve the study objectives. Using new features extracted from muscle inter-relationship, feature reductions and specific classifier, a linear discriminant analysis (LDA) is trained to detect possible errors in classification decisions. Non-stationary conditions in real life applications for normally limbed human are taken into account in the data collection strategy, such as different levels of maximum voluntary contraction (MVC) so that the classification is robust and accurate estimations are achieved. The proposed study contributes towards the enhancement of data collection strategy, extraction of best features and parameters, and optimal classification accuracy for the control strategy. Furthermore, it is established that the relationship between human UFA muscles is contributed from the movement with an accuracy of >90%. This provides an additional insight into the interrelationship between both muscle regions (forearm and upper arm), which is unique in this study.
Supervisor: Tokhi, M. O. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.784716  DOI: Not available
Share: