Energy Data Hackdays 2020, the results!
The excitement before the new edition of the Energy Hackdays in Brugg was a bit special this year. Besides the usual sweet little heart pinch of the leap into a new group, the discovery of the challenges and the satisfaction of seeing this particular event repeating for the second time in Brugg, there was happiness but also respect about having the Energy Hackdays taking place mostly on site at the Hightech Zentrum Aarau.
So we met in person and as far as we can say, it has been worth it! 13 really ambitious and technical challenges met 85 participants who were nonetheless ambitious and highly qualified! Two big themes emerged this year and predictions based on machine learning was one of them. Predicting performance, usage patterns, anomalies or even failure, in order to plan, use and maintain infrastructure more accurately. Reaching these goals of course allows a much better resource and production management.
The other big topic covered by several challenges was the question of visualization and interfaces, especially for smart-meters: How to help users, scientists, producers or end-consumers to read flows of data and allow them to interpret and decide or react appropriately to a given data supported information? How can they analyse and control different aspects of their infrastructure or installation? Tangent to this topic were challenges that attempted to allow a market overview for the consumer, in this case the market of E-Car charging stations, or to visualize the overall live electricity consumption of Switzerland.
As far as I can judge and from what I heard from the challenge owners, the results blew us away! While the project descriptions might be a bit less accessible to the public than some from past hackdays, the approaches and results certainly correspond to a present need in the energy industry and comfort us in the conviction that hackathons and collaborative work with Open Data do support high-end innovation.
We were also very lucky to welcome the team of Campus 21 who harvested the visions of some of the participants for the future of Open Energy Data.
See you all next year!
The 13 projects developed during the hackdays
Decrease gas peak boiler runtime due to better storage operation: heat demand forecast, improved storage control, better storage operation.
Evaluate and optimize trade-offs in the design of battery storage for PV systems, so our customer can select, whether they want the most economical battery solution or maximise their autarky. Our tools calculate the maximized economic benefit over lifetime.
Read your Smart Meter through the local Customer Information Interface (CII) and visualize your consumption. Design a dashboard with the most useful information.
In order to develop the GIS platform of the Swiss Federal Office of Energy (SFOE) further: Add price information to the charging stations and find the cheapest option around for electric car drivers.
Creating a platform for strategic decision making based on data from the Energy Science Center of ETH Zürich.
We analysed the charging patterns of private vs public e-cars charging stations. This could provide good hints for a further automated customer segmentation, help prediction of behavior changes for the load-curve vs renewable electricity production & help customers optimize their charging habits.
Smart Meter Additional Use Cases: Novel energy certificate assesses where and how strongly building / user behaviour causes deviation from theoretical / optimum behaviour.
Machine Learning Wind Turbine Power Curve Prediction: we compared constructor provided production projections with actual production curves with the goal to improve site-specific performance prediction of wind turbines.
– Development of machine learning algorithms (or tools/aps) for improved site-specific performance prediction of wind turbines.
– Development of alternative algorithms e.g. Artificial Neural Networks
– Inputs: wind velocity, turbulence intensity, shear factor (alpha)
Help meet the Paris Convention goals to achieve net 0 by 2050, less than 2 tonnes CO2 per person! We want to raise awareness around energy use and consumption by putting Switzerland on the map at electricitymap.org and put its open data API to use.
The goal is to create a decision support tool for asset managers, using AI to predict how power transformers will fail, and what to watch out for.
We developed EDA and algorithms for the Anomaly Detection in Smart Meter Data challenge. We developed several approaches for detecting anomalous days based on mean and std of the readings during the day and for detecting single anomalous readings.
These models can be integrated in the second part of the challenge
Create a model for Smartmeter Anomaly detector and their visualization.
Empower citizens to use their energy data. Using the smartmeter’s CII beyond visualisation to steer local consumption.We developed a concept and PoC roadmap to provide a “universal” adapter from smart meters to home IoT platforms.