Collaborative energy management app to prevent blackouts
- Daniela Leite (Research Scientist)
- Diego Baptista (Research Scientist)
- David Ringgenberg (Fullstack Developer)
- Mugeeb Hassan (Fullstack Developer)
The logic behind our project builds upon the fact that individual's consumption patterns can be highly influential on the state of the overall system. Thus, failures can be easily avoided by consumer and provider cooperation towards energy reduction.
Our project is presented in the shape of an app. It consists of an user interface where different metrics and actions are presented. These metrics/actions regard overall energy consumption, by-room/by-ambient consumption, and outage alerts. The app is used as a tool to ask specific users helping the energy provider to avoid energy overcharges. When a blackout is predicted, the provider contacts the users to voluntarily inquire them to reduce their consumption. The users then decide whether they want to cooperate with the provider by turning off some the energy consumption sources.
For example, let's assume that the provider predicts that the consumed energy at a particular region of a city would reach values that are extremely closed to those of energy generated by them. This means that the energetic system might suffer a collapse, which would translate on power outages across different parts of the city. The company sends then a message to their costumers to ask them for help. This message comes in the form of an alert. Different users are contacted, and then it is left to them to decide how to respond to the alert. Some of them decide to cut off the energy coming into the kitchen, or the living room, depending on whether they need it those parts of the house or not. Users get updates about the overall consumption levels. Once it decreases, the alert disappears.
- Technologies:
- Fast API.
- ReactNative, for the mobile app.
- Design:
- Figma for the UI, UX.
- Canva, for the video.
- Programming languages:
- TypeScript and JavaScript.
- Python for the backend: jupyter-notebook, numpy, scipy, matplotlib, pandas.
A challenge we ran into was related to the specificity of the energy consumption measurements. The data set provided did not contain rich temporal consumption information. The consumption values consisted of the aggregation of the different households over long periods of time. This meant that short-time actions were unfeasible. To overcome this issue, we thought of obtaining more regular measurements by installing sensors at the household levels.
Another issue we faced concerned the exactitude of the consumption measurements within particular households. The provided values consisted of general energy categories.
- We learned that blackouts are a latent risk for cities around the world.
- We got a better understanding of the energy ecosystem in Switzerland.
- We got a feeling about the energy measurements, categories in regular households.
- We got to know about used technologies to develop apps.
- We understood about basic predictive models, and how machine learning can be useful to cope with real life problems.
Our future plans about the project regard:
- The predictive model: we would like to use more suitable machine learning algorithms to forecast both individual's costumer and collective consumption
- Integration with third party systems, to obtain data for sensors.
- Reward systems: we would design a more attractive reward system for the costumer, to further incentivize the cooperation with the company
- Selling it to users and companies.