AI in the energy sector: innovation and patenting trends
As we transition to a more sustainable energy system, grid management is becoming increasingly complex. Renewable energy sources such as solar installations are often distributed, creating a need to manage electricity flows to supply electricity to end consumers as well as feed it back into the grid. Renewables also create a fluctuating supply that can be difficult to predict. These factors require careful management, and artificial intelligence (AI) has a key role to play.
Roles of AI in the energy sector
A principal use of AI in the energy sector is to forecast supply and demand more accurately. This improves electricity flow planning, enables more efficient use of the grid, and increases financial return from selling power in advance.
For example, in 2019 Google and its subsidiary DeepMind developed a neural network that predicts wind power output 36 hours in advance. Together with other machine learning innovations, Google reported that this increased the commercial value of its wind energy by 20%. The company has stated that it hopes this approach will strengthen the business case for wind power and facilitate more widespread adoption of renewables worldwide.
Accurate forecasting of supply also enables tasks that require a lot of power to be scheduled to coincide with peaks in supply. Google, for example, can schedule its own computationally intensive tasks to coincide with peaks in supply from its wind farms, avoiding the need to buy extra power from the market.
Another common use of AI in the energy sector is predictive maintenance. Traditionally, grid infrastructure is inspected and repaired at fixed times that are scheduled in advance. The actual state of the infrastructure and its environment is not taken into account in this scheduling, and consequently inspections are sometimes too early and repairs are sometimes too late.
In a new, predictive approach, sensors such as cameras and vibration sensors are distributed around the grid to monitor assets such as transmission lines and collect data from which AI solutions can predict whether maintenance is required. This creates smart scheduling of maintenance and helps to prevent serious failures in infrastructure. AI solutions for predictive maintenance can reduce inspection costs by up to 25% and increase uptime by up to 20%.
Another area that energy companies are exploring is the use of AI for enhanced consumer services. For example, Octopus Energy uses AI solutions in its consumer-management platform “Kraken” to manage end-user solar installations and energy needs, from optimal electric vehicle charging and heat pump management to cost-effective, green procurement of additional energy from the grid.
As with all AI-based innovations, AI solutions in the energy system take a significant amount of energy to implement. In 2022 Google reported that it had spent about 15% of its total energy consumption on AI workloads over the previous three years. In 2024, the company reported that it expected its total greenhouse gas emissions to go up before they come down, and that “reducing emissions may be challenging due to increasing energy demands from the greater intensity of AI compute”. In the energy sector AI solutions therefore need to save more energy than they use in order to be viable.
EPO energy sector AI filing trends
The recent surge in AI innovations in all fields of technology is reflected in patent filings. At the European Patent Office (EPO), AI solutions in the energy sector fall into two filing categories:
- Computer technology
- Electrical machinery, apparatus and energy
Both of these are experiencing strong, consistent growth and are in the top five technology areas by European patent filings. In the five-year period between 2018 and 2023, computer technology filings grew by 36%, and electrical machinery, apparatus and energy filings grew by 43%.
Patentability of AI in Europe’s energy sector
In Europe, patentability requirements for AI-based inventions follow the same rules as software inventions. On their own, software and AI inventions are not considered to be patentable, but they can be if they solve a problem that the EPO considers to be technical.
For example, AI-based solutions for predicting weather events, controlling hardware, and making efficient use of resources in a technical system are likely to be patentable. Our patent attorneys will be happy to advise you further on European patentability requirements.
It will be fascinating to watch AI and other solutions emerge in the energy sector as the decarbonisation effort continues and to see these developments work their way through the global patent system.
Useful links
- Machine learning can boost the value of wind energy, Google: dycip.com/machine-learning-wind-google
- Smartening up with Artificial Intelligence: What’s in it for Germany and its Industrial Sector? McKinsey (PDF): dycip.com/ai-germany-industry-mckinsey
- Octopus Energy’s Kraken consumer-management platform website: dycip.com/octopus-kraken-ai
- Measuring the Environmental Impacts of Artificial Intelligence Compute and Applications: The AI Footprint, OECD Digital Economy Papers, November 2022, No. 341: dycip.com/OECD-nov22-341
- Environmental Report 2024, Google (PDF): dycip.com/google-environmental-report-24
- Top 10 technical fields, Patent Index 2023, EPO: dycip.com/epo-top-technical-fields-23