Cary - Applied Research Study

 

Project duration: 01.04.2023 – 15.12.2023

Client: E.ON SE

In many countries in Europe the population is aging owing to the simultaneous trends in declining birth rates and declining mortality rates. This development has resulted in a significant increase in demand for nursing and care services which in their current forms cannot adequately meet future demands. In parallel to this, current care models place significant demands on unpaid informal carers, which are often family members who already bear other family and professional responsibilities.

Given this situation, there is a great need for affordable care solutions that can properly support both the informal carer and the care recipient. The vision of Cary is to enable relatives to worry less about elderly people living alone by providing accurate, reliable and actionable behavioral insights that allow them to be proactive about the elderly’s wellbeing while still respecting their privacy.

In this project, the Chair for Energy System Economics (FCN-ESE) supports E.ON as part of the project Cary in developing a functioning prototype for a simple low-cost technical solution for an alarm generation and behavior recognition system in care recipients’ homes that can be installed by the carer.

The solution will consist of four key components that are underpinned by the required infrastructure to allow Cary to operate as a complete solution:

  1. Energy consumption data measured by a smart meter in the care recipient’s home
  2. Additional sensors/smart plugs for the improvement of system accuracy
  3. Intelligent algorithm which learns care recipient’s daily routines and reports alarms in case of deviations
  4. Smartphone application that provides an overview of the care recipient’s wellbeing and delivers notifications

As part of the first work package FCN-ESE will analyze the results (measured data and diaries) from the first Proof of Concept (PoC) carried out in fall 2022 and derive recommendations for a further PoC planned for summer 2023. Furthermore, suitable models for pattern recognition and anomaly detection for household energy data will be identified.

The second work package comprises the development of pattern recognition and anomaly detection models and the application of the models to the existing data from the first PoC. The accuracy of pattern recognition and alert generation will be evaluated with different technical setups. This includes the variation of meter and sensor device measurements that are included in the model input data. Based on this, recommendations regarding the hardware set up in the care recipients’ households will be derived.

In the third work package a second field test will be performed for which an improved test setup needs to be defined. The model will then be fed with the newly measured data and necessary adjustments of the model which were identified during the field test will be made.