Photo of Dr Andrew P Duncan

Dr Andrew P Duncan

Associate Research Fellow

Email:

Telephone: 01392 724075

Extension: (Streatham) 4075

Dr Andrew P Duncan has extensive technical software engineering experience (over 30 years) in the electronics and computing industry mainly concerning test and measurement of electronic and photonic systems. He holds a PhD in Computer Science (2015) and an MSc in Applied AI (2010) both from University of Exeter, an MSC in Cybernetics (1978) and a BSc in Electronics (1974) both from University of London.

His PhD Thesis is entitled “The Analysis and Application of Artificial Neural Networks for Early Warning Systems in Hydrology and the Environment” and is based on work carried out for a number of projects relating to prediction of urban flooding using Artificial Neural Networks (ANN):

FRMRC2 http://emps.exeter.ac.uk/engineering/research/cws/research/flood-risk/ and

UKWIR RTM http://emps.exeter.ac.uk/engineering/research/cws/research/urban-drainage/case_study/  and for bathing water quality:

BACTI https://emps.exeter.ac.uk/engineering/research/cws/research/urban-drainage/bacti/

In his thesis he develops a novel technique for using the values of the learnt weights from ensembles of ANNs to infer the degree of relevance of predictors (input features) used for the models. This then allows simpler, often better-performing, models to be constructed using only a reduct of the relevant input features. Multi-output ANN models are also constructed that allow modelling and prediction of flooding at multiple sewer nodes from a single ANN.

He is now Associate Research Fellow at University of Exeter Centre for Water Systems and specialises in software development and research into machine-learning and nature-inspired computation using Big Data for water and environmental applications. His project contributions to date include: EPSRC-funded FRMRC2 – Flood Risk Management Research Consortium 2 (2010-11); UKWIR-funded RTM project (2011-12) – looking into potential applications of machine-learning based flood prediction models for real-time control of urban drainage networks; Environment Agency-funded BACTI project (2012-13) – developing artificial neural network (ANN)-based predictive models for water quality at designated bathing beaches as required by the EC Bathing Water Directive (2006); EU FP7 and Indian Government joint-funded SARASWATI and WATER4INDIA projects (2014-2016) investigating options for improved water and wastewater treatment in India and BaMoS pilot project (2015) to investigate potential for adding satellite-based remote sensing data to the BACTI ANN models.

From 1 Feb 2014 he took up a position as Associate Research Fellow in CWS and commenced work on The Saraswati Project - a joint Indian Government and EU FP7-funded multi-agency project to research and develop improved, sustainable wastewater treatment and re-use solutions for India. My work is to research and apply Evolutionary Many-Objective Optimisation (EMOO) techniques and compare with Multi Criteria Analysis (MCA) for the selection and design of treatment trains and water redistribution systems that simultaneously meet many financial, technical, environmental and social objectives. University of Exeter Saraswati project page here

 

Supervisors: Prof. Dragan Savic (Water Systems); Dr. Edward Keedwell (Computer Science); Prof. Fayyaz Memon (Saraswati)

UoE Co-workers: Rebecca Austin; Dr Albert Chen; Dr Bidur Ghimire; Dr Michele Guidolin; Mike Gibson; Matthew Shere; Dr Seyed MK Sadr; Nicholas Sanders; 

 

Google+