PhD Thesis: Water demand forecasting using machine learning on weather and smart metering data
PhD Supervisors: Professors Zoran Kapelan, Slobodan Djordjevic and Jan Hofman
PhD Industrial Supervisor: Dr Chris Hutton, Wessex Water
Accurate forecasts of demand are a key input to understand future supply-demand balance. While long-term predictions can be used to develop pro-active strategies that will secure water for the future, short-term forecasts assist in the operational and financial management of the system.
A major part of demand forecasting is understanding the underlying processes and identifying the determining factors that drive water consumption. The current research aims to utilise a combination of statistical methods and machine learning techniques in order to identify patterns and establish relationships between water demand and a variety of factors that are suspected to influence it.
In order to achieve this, an extensive dataset comprising of high-resolution consumption data (derived from smart meters), household characteristics, socio-economic factors, and weather variables became available. The methodology adopted is based on a systematic approach that evaluatesthe relationship between water consumption data and explanatory factors for different temporal and spatial scales and aggregations of households (based on household characteristics and socio-economic data), while implementing uncertainty and risk based planning
Maria was a WISE CDT student from 2015-19 and undertook her PhD research within the Centre for Water Systems at the University of Exeter, whilst collaborating closely with Wessex Water. Maria’s previous academic background was an MEng in Civil Engineering from the National Technical University of Athens (NTUA). Whilst at university she completed a year-long internship as a research assistant at the Berlin Centre of Competence for Water (Kompetenz Zentrum Wasser Berlin), where she worked on project SEMA (SEwer deterioration Model for Asset Management strategy). Following her return to Greece she undertook a four month internship as a data analyst for a Greek start-up company. Maria’s research interests lie at the intersection between civil and software engineering, and include water demand pattern recognition and forecasting, statistical analysis and machine learning.