Applying machine learning to Energy demand and distribution optimization in Kaduna Electric.

At the moment, Kaduna electric has a poor means of energy distribution, they basically share the time at which they transmit power to different areas, it is as bad as you having power for 4hrs and you friend on the opposite street living in the dark for those 4hrs (it is popularly called power shedding).

A way to fix this would be them to build a model that helps optimize the distribution of power based on demand, for instance in the mornings people going to work use stuff like the water heaters, irons, toasters etc (a certain period of time, e.g. 6:00am-7:30am). Times like these means a higher demand for energy in the morning and factories (which factories? some factories run for 24 hrs, some offices like banks run 9am — 5pm) for instance that consume high energy would not be supplied a lot of energy at that point in time.

A time series data of energy consumption (across residencies, offices, factories) in Kaduna would be needed and analysis gotten from this data would be used as the labels for building a machine learning model. that would have the ability to predict the amount of energy needed at a given point in time which would optimize power distribution, having this also means they would also be able to predict the amount of energy to generate(or to demand from the central grid).

It is noteworthy to add that to get this data IoT sensors would aid us in reading the demand and the amount of consumption from different geographical locations. In other words, these sensors would tell us how much power is being consumed by each building/consumer, hence knowing the demand at a given time.

This can result in more income for them, because they can always channel the electricity to places where it is being utilized most at every given point in time. Additionally it gives them information that will enable them to schedule and plan the expansion of the grid in future to meeting growing energy demand in the town.

Special thanks to Nnamdi Judges for editing this article and for lending his expertise on data collection with IoT.

Connect with the Raven team on Telegram

Source: Deep Learning on Medium