Centre for Genomic Regulation (CRG) Featured PhD Programmes
University of Cambridge Featured PhD Programmes
University of Hull Featured PhD Programmes

Statistical Learning Methods for Real-time Diagnostics of Gas Turbine Combustion System


   Department of Mathematical Sciences

This project is no longer listed on FindAPhD.com and may not be available.

Click here to search FindAPhD.com for PhD studentship opportunities
  Dr D Zhou  Applications accepted all year round  Funded PhD Project (Students Worldwide)

About the Project

Start date:

January 2022, April 2022, July 2022

Full-time/part-time availability:

Full-time (3 years)

Name of primary supervisor:

Dr. Diwei Zhou

Supervisor's contact details:

Link to supervisor's online staff profile page:

https://www.lboro.ac.uk/departments/maths/staff/diwei-zhou/

Secondary supervisor(s):

Dr. Eve Zhang

Link to supervisor's online staff profile page:

https://www.lboro.ac.uk/departments/aae/staff/eve-zhang/

Project detail:

Statistical Learning (SL) is the process of gaining understanding by constructing models of observed data with the intention to use them for prediction. This project aims to develop novel statistical learning methods and algorithms to understand and predict the nonlinear combustion behaviour in gas turbines (GT). For this engineering application, we focus on tackling three SL challenges (drawing robust inference for complex engineering models, quantifying the uncertainty in models for large-scale GT combustion data and computing under limited computation power).

Conventional techniques for GT diagnostics based on minute data are only efficient in detecting gradually emerged faults, such as compressor fouling. This project will process and analyse the high-resolution data of GT combustion system, aimed at accelerating the GT diagnostics capability and, as part of the real-time GT performance monitoring platform, to prepare for the transition to hydrogen as a fuel for GTs, where it is crucial to manage the safety processes of hydrogen fuel in real-time. Digital twins will be constructed representing the underlying thermodynamic behaviour of the combustion system based on the high-resolution gas path sensor readings. This advanced digital twin system can be used to model the nonlinear combustion behaviour, with estimated measurements and model uncertainties, to detect abnormal sensor readings corresponding to anomalous GT combustion response, i.e., novelty detection. The developed intelligent diagnostic system will be validated using sensor data collected in the university lab environment, as well as real-world data collected for condition monitoring of gas turbines. Working alongside global company Uniper, this project is an exciting opportunity for a forward thinking and imaginative individual to participate in the study and analysis for real-time diagnostics for GT combustion systems.

The student will gain experience and skills in a range of SL methods include Bayesian Predictive models, Neural Network, Robust Statistics, variable selection and model assessment methods.

MA/DZ-Un2/2022

How to apply

All applications should be made online. Under programme name, select 'Mathematical Sciences'. Please quote the advertised reference number MA/DZ-Un2/2022 in your application. To avoid delays in processing your application please ensure that you submit the minimum supporting documents.

Entry requirements:

The successful candidate will have at least a 2:1 BSc (Hons) in a relevant mathematical/statistical or computer science discipline. Coding experience in Matlab or Python is desirable.

English language requirements

Applicants must meet the minimum English language requirements. Further details are available on the International website.


Funding Notes

Fee band:
UK: £4,500; international: £18,100
Search Suggestions
Search suggestions

Based on your current searches we recommend the following search filters.

PhD saved successfully
View saved PhDs