Are renewable energy systems getting everything out of AI?

Are renewable energy systems getting everything out of AI?

The idea of artificially intelligent decision-making is not new. Replicating or superseding the human thought process using computers was already a theoretical concept way before computers were even built. So why the recent buzz about artificial intelligence (AI)? At the very core, this is due to only one development: better computing power.

Artificial intelligence used to be reserved for algorithms that perform some relatively intelligent task. This boils down to being able to interpret a certain scenario and acting accordingly in an atonomous way. A very simple example is the computer moving the paddle in Pong (the classic 70’s arcade game). Although the AI’s are likely to beat humans at some stage in the game, one would hardly call this AI very intelligent. It can’t do anything other than play Pong!

Pong, the 1970’s arcade classic. In the end, the computer always wins.

At the time Pong came out, the theoretical knowledge on AI had already progressed a lot. Algorithms were theorized, which were much more akin to the way humans make decisions, these days called artifical neural networks. The application however, was not feasible by a long shot. And it hasn’t been until quite recently, when the available computer chips had improved sufficiently to support neural network calculcations. Ever since, different types of neural networks have broken record after record in performing artificial intelligence tasks.

So how can this recently rediscovered type of AI help to achieve smarter energy systems?

  1. Automated pattern recognition. Hidden patterns in a potentially steady stream of data, are often impossible for humans to detect. While being very well-trained in figuring out intricate patterns in images and sounds, it becomes increasingly harder for us when we’re interpreting more abstract data. Deep learning focuses on unleashing these pattern recognition capabilities of the human brain, by using network architectures inspired by the human sensory system. This allows for pattern recognition in more complex streams of data, like for example energy production and consumption.
  2. Better autonomous decision-making. One of the desired goals of many artificially intelligent systems, is not only to understand what is going on in the system, but also to act accordingly. Some, especially crucial, systems trigger faults if a wrong decision is made. Surely an artificial intelligence should be able to take this into account! This can be achieved by a set of methods called reinforcement learning. The AI is ‘rewarded’ for making good decisions and ‘punished’ for making bad ones. If combined with deep learning the decisions can very well qualitatively exceed human standards, which is especially interesting for autonomous energy systems.
  3. Increased fault tolerance. Regardless of decision-making being done by humans or computers, some decisions may trigger faults in the system. This is naturally highly undesirable. In most cases, faults should be prevented at all costs. It is not always easy to figure out all potential ways in which a system may fail and prepare for them. A different way of looking at it, is by instigating an artificially intelligent attacker, which will try to trigger faults in the system. This will allow the systems operators (human or artificial) to learn how to deal with potentially fault-triggering mechanisms and increase the fault tolerance.

Sobolt is surfing the ‘new AI’ wave almost since the beginning and is able to apply all of these technologies in practice. This goes further than making nice visualisations of data or automating solutions. It is really about replacing the necessity for human intelligence with well-equipped artificially intelligent subsitutes.

Are you trying to bring down costs, raising accuracy or increasing autonomy of your (renewable) energy system? Or are you just looking for innovations that can bring your product or system to the next level? We’re always happy to sit down with you and find out wherethe opportunities lie.

Contact us now, or send an e-mail to Danny Hetharia to set a meeting.

Comments are closed.