This first of its kind technology, named AquaML, combines state-of-the-art machine learning technology with a single, low-cost pressure sensor. The simple, trained system can differentiate between different water-using devices being used in a home or premises, generating water usage data from each, and providing the “data disaggregation” that utilities and building service managers have long demanded. AquaML further demonstrates Cambridge Consultants’ talent for moving machine learning beyond research and into real-life, physical projects that can impact millions of lives.

Water is a hugely valuable commodity but its scarcity is increasing in many regions of the world, and the problem will only worsen. The OECD project that world water demand will increase by 55% between 2000 and 2050[1], and by 2060, Singapore’s total water demand could almost double. Singapore is at the front line of water innovation, and has become a globally renowned hub for technology partnerships as it seeks to guarantee its own water security.

One of the central challenges facing the water industry is in understanding exactly how water is being used once it has been delivered to customers’ premises. Access to detailed usage information will unlock new insight, helping utilities to plan for future demand. Improved measurement will enable greater control, while waste from leaks, drips and misuse can be detected and minimized to improve efficiency. All of this is made possible by a single sensor and cutting-edge machine learning algorithms.

Cambridge Consultants is at the cutting edge of advances in machine learning, applying these transformative technologies to a wide variety of industries. The water industry is one of many that are ripe for disruption. AquaML combines a single pressure sensor and a sophisticated machine learning algorithm. When water flows through a pipe to supply an outlet it creates a unique detectable pressure profile that the machine learning algorithm first learns and then uses to identify water usage from separate devices, for example from taps, washing machines, showers or toilets. AquaML records the volume of water used and even produces a real-time display of the water usage by outlet. Previously, the only way of gaining this usage data was to install sensors at each device, perhaps hundreds, which would be both expensive and complex to maintain.

Data generated by AquaML allows a water utility company to understand how water is used within individual households. Beyond simply knowing how much water an apartment block uses, a utility can now break down that usage by showers, toilets and individual sinks.

“Water is a precious resource and must be harnessed in the most sustainable and efficient way possible,” said Wang Bin, VP Industrial and Energy, Cambridge Consultants. “By applying the latest machine learning technology, in a completely new approach, we can provide detailed insights into precise usage of water on a scale that has never before been viable.”

[1] https://www.oecd.org/env/indicators-modelling-outlooks/oecdenvironmentaloutlookto2050theconsequencesofinaction-keyfactsandfigures.htm

編集者へ