Dynamic classification of flow time series

Real-time monitoring can be a powerful tool for improving the efficiency of water supply systems and reducing losses. Nevertheless, in large systems, it is unrealistic to be permanently attentive to all zones without automatic monitoring tools. 
Baseform develops forward-thinking software for networked water infrastructures that assists water utilities in monitoring the entire network in its daily management and to track down, collect and validate all relevant information produced in a systematic way.

Several algorithms are continuously applied by Baseform to analyze and accurately predict water demand patterns and to establish whether the consumption recorded, during a specific period, is normal or may reveal abnormal events, such as pipe or service connection failures, leaks, or meter malfunctions. 
However, it is advisable to know the characteristics of each flow time series in advance to choose the most suitable event detection mechanisms.

The goal of this project is to develop and compare several algorithms to continuously classify time series based on recent records and derived features. This categorization aims to identify groups of time series with comparable characteristics that should be configured with similar parametrizations of the event detection mechanisms.

Contact: Margarida Azeitona (margarida.azeitona@baseform.com)