This site uses cookies to enhance navigation. If you continue with your visit, you accept the use of cookies and privacy policy. Accept

IK4 Doctorados

 

CEIT

Multicultural center which develops applied research programs to improve the innovative capacity of companies.

 
DOCTORATES

CEIT

Supervised and unsupervised learning techniques for detecting technical losses (leaks) and non-technical losses (fraud) and water consumption forecasting for water delivery systems

CEIT

START: 09/2016

OPEN

OBTAINING DOCTOR DEGREE: 10/2019

Description

The sustainable and integrated management of water resources is one of the most complicated global issues for drinking water distributors. Fresh water is a limited and valuable resources, and it must be preserved and used accordingly. Given the fact that water is a scarce resource and the other issues that have arisen due to population growth and the effects of climate change, water must be managed in an efficient manner, where the primary goal is ensuring that the maximum amount of water that enters the system ends up at its consumption points while minimizing losses to the greatest degree possible.

The doctoral thesis will focus on the development of new algorithms for data analysis and machine learning for water consumption forecasting and leak detection, which will improve the operation and management of water delivery systems.

Specifically, the candidate will study:

1. Unsupervised learning techniques based on time-cycle analysis, defining for such purposes measurements ofdistance between cycles of consumption that are specifically designed for the case study at hand.

2. Tools and methods based on graph theory and brain connectomicsin order to discover functional patterns of cause and effect.

3. Models that are able to automatically detect seasonalities or concept drift, in such a way that they are capable of adapting to the changes and produce a high quality forecast, balancing the Pareto between plasticity (adapting to change by taking advantage of previously trained models) and stability (model learning without overadjustment to specific changes in the dataset).

The models developed will be validated using public domain datasets, and potentially with real data after securing the collaboration of entities such as URA or Consorcio de Aguas.

Department or unit of the Associates Technology Center

Water and Health

Investigation line

Big data, Data analytics and machine learning

Start date planned

09/2016

Obtaining date of the doctor degree

10/2019

Requeriments

Degree in Telecommunications Engineering with a Master’s in Telecommunications Engineering /Degree inExact Sciences / Degree in Mathematics with a Master’s in Modelling or Mathematical, Statistical or Computational Research or a Master’s in Telecommunications Engineering.

IK4 Research Alliance SELECT IN ORDER OF PREFERENCE THE OFFERS YOU WISH TO APPLY FOR

With number 1 as the first option, 2 as the second and 3 as the last option.

Remember, only click "Send" once you have configured your options correctly. Click on the "Save" button to add more options, or to change their position.