Specialising in both data driven and mathematical approaches for water resources and with a skill set eminently applicable to other domains. HydroSmart delivers expert application of data mining, pattern recognition and modelling for classification/ prediction. A key focus for the business is the issue of 'big data' and obtaining 'signal from the noise' using data science. Additionally, sustainability and the 'water-energy-food' nexus (with strong links to transport) is seen as a core area. Primarily, the focus is on data to action using data science.
Data science is an emerging discipline which combines analysis, programming and business knowledge and uses new and advanced techniques and technologies to work with complex data.
Hydroinformatics has emerged over the last decade to become a recognised and established field of independent research within the hydrological sciences. Hydroinformatics is concerned with the development and hydrological application of mathematical modelling, information technology, data science (e.g. Data mining and knowledge discovery, Big data and deep learning techniques) and computational intelligence tools (e.g. prediction and classification). It provides the computer-based decision-support systems that are now becoming increasingly prevalent for use by consulting engineers, water service providers and government agencies to implement solutions such as Smart Networks.
In a game changing period of transformation, smart networks are at the forefront of investment plans for UK water companies. Technological advancements allow water companies to gather more information about their networks and assets than ever before. With the industry seeking Totex solutions in AMP6, water utilities are investing in solutions and new technologies that improve customer service, enhance efficiency and drive resilience. A smart water network is a fully integrated set of products, solutions and systems that enable water utilities to remotely and continuously monitor and diagnose problems, pre-emptively prioritize and manage maintenance issues, and
remotely control and optimize all aspects of the water distribution network using data-driven insights
Big data and deep learning
The availability and affordability of sensing, smart systems, data storage and transmission technologies mean that water utility companies are increasingly enabled to collect more data than ever before. This information revolution era opens up hereto unseen possibilities in the creation of tools in the Engineering of the Future. Deep learning is when big data intersects with machine learning. These techniques allow the tackling of problems that exceed human understanding. Deep learning’s important innovation is to have neural nets learn categories incrementally, attempting to model lower-level categories (like letters) before attempting to acquire higher-level categories (like words). Deep learning excels at this sort of problem. Next-generation formulations of deep artificial neural networks allow for the direct transition from data to action. Such state of the art algorithms include unsupervised feature extraction as part of the data driven learning.
Data mining and knowledge discovery
Water sector data sets have not in general been analysed by, or used as input to, leading edge computer science based techniques: in particular in the field of artificial intelligence. It is neither cost effective nor efficient to rely on detailed human monitoring of raw sensor outputs. AI-based systems can automate mundane tasks involved in the process, as well as presenting more ‘intelligent’ information to the operator. State of the art clustering techniques are able to find clusters and relationships in data in an unsupervised manner (without a specific label or output value). Data is grouped into categories based on some measure of inherent similarity or distance. The team has experience in data mining of large water quality and asset datasets with advanced clustering algorithms (such as Self Organising Maps and DBSCAN).
Data analytics: prediction
Predictive analytics involves using the patterns of past behaviour to predict behaviour in the future. Predictive analytics encompasses a variety of statistical techniques from predictive modelling, machine learning, and data mining that analyse current and historical facts to make predictions about future (or unknown) events.
Predicting future values of time series is one such application useful in many water resource domains. Since there is a set of temporal ordered observations for which (working on the assumption that) there exist serial correlations along the series, previous observations can be used to predict future values. The task is essentially one of function approximation i.e. to approximate the underlying continuous valued function producing the time series.
The team has expertise in applying various Artificial Neural Networks, SVM and other algorithms for prediction.
Data analytics: classification
There are other ways in which analytics can be used than for prediction, from understanding customers to optimising a business process. In machine learning and statistics, classification is the problem of identifying which category a new observation belongs, on the basis of a training set of data containing observations (or instances) whose category membership is known (hence described as supervised learning). An example would be assigning a given pipe to a high risk or low risk category for water quality issues based on asset characteristics (pipe material, age, diameter, etc.).
Classification can be viewed as a special case of function approximation using some type of discriminant for the decision.
The team has experience in appling a range of algorithms to water industry problems such as Artificial Neural Networks, Fuzzy Logic and Decision Trees, and including Ensemble methods.