Prof Rajesh Sundaresan (Professor, Robert Bosch Centre for Cyber-Physical Systems, Department of Electrical Communication Engineering)
The project aims to provide personalised feedback to about 25,000 electricity consumers in the town of Aluva, Kerala. The goal of these personalised feedback inputs is to effect positive changes to consumer behaviour. The data from these 25,000 will compared with the consumption of a control group of another 25,000 from the same town. If our feedbacks are found effective, this will lead to an opportunity to expand the use of this framework to other utilities and provide them opportunities for savings.
The challenge is that we cannot instrument these 25,000 houses with smart meters rightaway because of budgetary reasons. So personalised feedback has to be based on cheaper means.
We obtained two years consumption data from the Kerala State Electricity Board for these consumers. We have used this data to cluster the households into various abstract categories. We have surveyed representative households – we have completed about 1,100 surveys of households – to get information on the number of individuals in the household, age-groups, basic appliances and numbers, building material types, floor, etc. The unsurveyed users have been associated with a certain number of ‘nearest neighbour’ surveyed consumers, and the factors are imputed based on the factors of these nearest neighbours. We have also arrived at disaggregation algorithms to identify consumptions for various categories, such as lighting, coooling, heating, refrigerator, others. The latest reports sent to consumers along with bills will contain this input and is the third in the set of reports sent to consumers along with their bills. The first report introduced the programme to the consumers, while the second report indicated comparisons within the cluster.
Additionally, to aid in the diaggregation, we have actually installed loggers in a few of the surveyed homes. We are still trying to make sense of the data that have been collected since the appliance signatures seem to be very different from data collected locally.
We are also trying to build an activity model for each individual in each household so as to enable us to simulate a city-scale system for modeling electricity consumption. This is still in its preliminary stages.
Exploiting appliance state constraints to improve appliance state detection Conference Forthcoming
Proceedings of the 8th ACM International Conference on Future Energy Systems (ACM eEnergy), 16.-19.05.17, Hong Kong (China), Forthcoming.
Edge conductance estimation using MCMC Conference Forthcoming
Proceedings of the 54th Annual Allerton Conference on Communication, Control, and Computing, 27.-30.09.16, Allerton (USA), Forthcoming.