Currents: AI & Energy Insights - June 2024
Welcome back to Currents, a monthly column from Reimagine Energy dedicated to the latest news at the intersection of AI & Energy. Every last Sunday of each month I’m sending out an expert-curated summary of the most relevant updates from the sector. The focus is on major industry news, published scientific articles, a recap of the month’s posts from Reimagine Energy, and a dedicated job board.
1. Industry news
Apple revealed a new home electricity analysis feature through their Home app, coming up with iOS 18. Initially, the feature will only be available to Pacific Gas and Electric customers in California, but hopefully expanding to other countries in the future. Find the full info within the iOS 18 preview notes.
What I’m thinking: Last month, we saw Ikea deploying a similar feature in their smart home app. Intra-day electricity price volatility and increasing rates of home electrification are sparking interest into understanding our home electricity consumption and how much we can save by consuming flexibly. It's impressive to see what can be achieved when software companies can use standardized APIs to access customer utility data. Kudos to Apple and PG&E for paving the way. This should now become the standard for more and more utilities that don’t want to see their customers switch to more data-friendly competitors.
Ento launches Ento Strategy, an AI-powered solution that creates data-driven energy and decarbonisation strategies for large building portfolios. The product utilizes data that’s already on the Ento platform and creates a tailored plan to reach long-term energy and carbon emission goals. Read more about Ento Strategy here and watch the extended product demo to see it in action.
What I’m thinking: I’ve been leading the development of this product for the past year, and I’m excited to finally be able to bring it out to the market. We conducted extensive product discovery with our existing customers and consistently incorporated their feedback to ensure it meets the practical needs of energy managers and sustainability teams. With new regulations coming into effect this year and next, I hope this product will help organizations managing large building portfolios to plan their decarbonization roadmaps.
Octopus Energy lands in Texas and seals key deals to strengthen demand-side flexibility for its customers. Octopus Energy has been on my radar for the last few years due to their role as an innovative utility with a large focus on renewables, demand-side flexibility, and using data to support the services they provide to their customers. This past month has been particularly significant as they got started in the U.S. market, announced signing the biggest battery-leasing deal in the world, and built new integration with Tesla Powerwalls that will allow customers to respond flexibly to prices throughout the day.
What I’m thinking: Octopus is among the utilities that understand the energy system of the future will be decentralized, flexible, and built upon renewables. They are creating the building blocks to support this transition, and it’s great to see them building powerful partnerships and promoting demand-side flexibility across different markets.
NVIDIA is enhancing their Earth-2 digital twin with hyper-local information. This will allow them to create high-resolution weather and climate predictions at street level. At the core of Earth-2 is CorrDiff, a two-step generative AI downscaling capable of producing numerous high-resolution weather scenarios quickly and cost-effectively. More info on Earth-2 can be found here.
What I’m thinking: High-resolution, street-level weather forecasts are crucial for unlocking the power and efficiency of grid-interactive buildings, among many other applications. It’s exciting to see NVIDIA dedicate some of their incredible talent and computing power to solving climate-related issues. We need more AI focused on building things that have a positive impact on society.
2. Scientific publications
Sky image-based solar forecasting using deep learning with heterogeneous multi-location data: Dataset fusion versus transfer learning. This study explores the use of deep-learning models for solar forecasting using ground-based sky images and sensor data. The most promising approach among the ones tested was pre-training models on a large, diverse source dataset and then fine-tuning on the target local dataset via transfer learning.
What I’m thinking: I’m a fan of using transfer learning and it’s exciting to see it applied in the renewable energy field. I also like seeing more research being done on using various data sources (e.g. time series data + images) and creating models that can handle all of them simultaneously. These insights seem to be moving us toward a “foundation model” for solar forecasting, which can be pre-trained on a large global dataset and easily adapted to local sites with minimal fine-tuning.
RECA: A Multi-Task Deep Reinforcement Learning-Based Recommender System for Co-Optimizing Energy, Comfort and Air Quality in Commercial Buildings. This research paper presents the design and implementation of a human-centric recommender system for co-optimizing energy consumption, comfort, and air quality in commercial buildings. The recommender can either suggest an occupant to move, multiple occupants to move together somewhere, or multiple occupants to separate into different locations.
What I’m thinking: We definitely need to involve humans in the building energy optimization loop, but I believe it’s very difficult to simply suggest users move around. Our location in the office is usually determined by factors other than thermal comfort, which need to be taken into account. I think machines should adapt to humans rather than the other way around. Still, I find this research very relevant because it helps us understand the interaction between thermal comfort and energy savings optimization, which will be key in building the HVAC optimization systems of the next decade.
3. Reimagine Energy publications
Due to our latest product launch and a few other engagements I’ve only been able to write one article this month, you can find it here:
On a positive note, I’m working on a new Python tutorial that I hope to release during the next weeks so stay tuned for that!
4. AI in Energy job board
This space is dedicated to job posts in the sector that caught my attention during the last month. I have no affiliation with any of them, I’m just looking to help readers connect with relevant jobs in the market.
Data Scientist, Energy Analytics at Google
Senior Data Scientist at Modo Energy (there’s a few positions open between London and the U.S.)
Research Associate/Senior Research Associate in Optimisation and Reinforcement Learning at University of New South Wales
PhD scholarship in Energy Management of Energy Flexibility Assets in Sports Facilities at DTU - Technical University of Denmark
Conclusion
With so much going on in the sector it’s not easy to follow everything. If you’re aware of anything that seems relevant and should be included in Currents (job posts, scientific articles, relevant industry events, etc.) please answer to this email or reach out to me on LinkedIn and I’ll be happy to consider them for inclusion!