LOS ANGELES, CA — Industrial automation is undergoing a complete transformation due to the emergence of Industrial IoT. It requires new tools, new skills and new ways of extracting valuable insight from the vast amount of information.
In energy systems, finding the efficient operating strategy of energy assets is the biggest challenge due to the complexity of modern systems. Such complexity is introduced by both internal and external factors. Continuous adoption of IIoT and abundance of data drastically increase the performance capacity and our expectation of what can be achieved. The high variability of renewable generation makes it extremely difficult to find the right balance due to the increased number of possible operating configurations.
The latest breakthroughs in Artificial Intelligence research can help address this issues. For example, Reinforcement Learning agents can help discover the most efficient operating strategy and achieve superhuman performance at it. They are not limited by rigid rule-based systems created by domain experts or abstract models that are only as good as the data they were trained on. The way they learn is by experiencing energy assets directly which allows them to adequately work with nonlinearity, and learn non-stationary and hidden environments.
It’s hard to estimate ROI in energy space because even a marginal improvement can mean millions of dollars of savings for both producer and consumer. However, since the integration of AI agents doesn’t require huge capital investments, modernization or personal training, it is definitely the most cost-effective way to rapidly improve performance metrics.
The technical challenges of applying Artifical Intelligence and Reinforcement Learning agents are not very different from any other enterprise software integration. In fact, it may even be easier than installing a new accounting software. The challenges here are primarily psychological because adopting AI agents lead to dramatic changes in the day-to-day operations of energy systems. The leaders need to be agile and prepared and follow the adaptive strategy when choosing to embrace digital transformation.
Like with any new technology, in 2-5 years time, the adoption will become widespread and the opportunity to get the benefits of being a first-mover will be gone. In fact, application of Reinforcement Learning agents will be a common occurrence and we will have to be looking for new ways to increase performance for the benefit of all.