Imagine a little virtual neighbourhood of houses, apartments and small businesses, each with solar panels or not, a battery or not, and sometimes an electric car. This Application simulates that neighbourhood running over time: watching how each house manages its energy, earns or spends money, and learns to do better.
Enter AppEach house has a small artificial intelligence brain. It takes in six pieces of information about the house's current situation
It then outputs a single number between 0 and 1, essentially a dial from conserve energy to use energy freely.
What makes it interesting is that this brain learns. After every moment in time, it gets a score (a "reward") based on how well it did.
It then quietly adjusts its internal settings to do better next time. No teacher, no manual, just trial and error.
Each house belongs to one of six personality types, ranging from
These types affect things like income, battery size, solar panel capacity, and whether they own an electric car.
Every moment ("tick"), a household goes through a little routine: the sun shines (or doesn't), the brain decides how much energy to use, the car charges if needed, money flows in or out, and the brain learns from what happened. If a house runs completely out of budget, it drops out of the system.
The household can be returned to the system by its neighbours, but the simulation shows a specific amount of time until all houses finish their budgets.
The Grid holds all the houses arranged in rows and columns (editable). Every tick, it steps all houses forward in time simultaneously. It also runs a little energy market between neighbours.
If your battery is low and your neighbour's is full, they can sell you some energy at the current market price. That price itself rises and falls depending on how scarce energy is across the whole neighbourhood.
Explored the simulation?
Share your feedback. It takes under 2 minutes and directly informs the next stage of the project.
Go to SurveyIf you have questions about the research and would like further information, please contact Dr Marilia Bergamo (marilia.lyrabergamo), Associate Professor Craig Hight (craig.hight), Dr Andrea Cassin (andrea.cassin) or Dr Richard Wood (Richard.Wood) all @newcastle.edu.au