Use this URL to cite or link to this record in EThOS:
Title: Frequency analysis of linear and nonlinear systems for applications in fault detection and medical diagnosis
Author: Zhang, Sikai
ISNI:       0000 0004 8509 2644
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Access from Institution:
The frequency analysis is highly demanded to process the industrial or medical data for fault detection and diagnosis, especially when the investigated machine or human tissues are stimulated by the periodic signals. The PhD research work aims to develop the new methods for system frequency feature extraction and selection of features for machine learning oriented classification and to apply these methods in the fault detection and medical diagnosis. To analyse the system characteristics with input-output data, novel modelling and model frequency feature extraction method is proposed. The method is effective in revealing the physically meaningful characteristics of systems. To select the useful features for machine learning oriented classification, an orthogonal least squares based feature selection method is proposed. Compared to traditional methods, the proposed method has a faster computation speed and higher accuracy. These novel methods are then applied to two real-world problems, which are wind turbine fault detection and preterm birth prediction. In the wind turbine fault detection application, the results show that the modelling and model feature extraction method is powerful in the damage sensitive feature extraction, while traditional methods do not work well. In the preterm birth prediction, the proposed methods can extract and select features from the magnetic impedance spectroscopy data of the pregnant women's cervix tissue. The results demonstrate that the magnetic impedance spectroscopy data have the capability to predict the spontaneous preterm birth. These application studies demonstrate that the proposed methods have great potential to be used in many engineering system fault detection and medical diagnosis related applications.
Supervisor: Lang, Zi-Qiang Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available