Putting AI + Blockchain into the Electrical Fuse Box
Electrical load panels, a.k.a. fuse box or circuit breaker panel, has not changed very much (at all) for nearly 60 years — primarily because they haven’t really needed to. The electrical panel has the main purpose of distributing utility power through circuits and utlimately to energy-demanding devices. Over 95% of homes or businesses use an electrical panel and is simply a must-have piece of equipment to distribute electrical energy.
Most people don’t realize, or even think about, how the electricity is generated, transferred through power lines, connected to the home/building and powers the devices they need. There is an expectation that if they flip a switch or press the ON button the “thing” should just turn on. However, when we lose power from an outtage in the area or there’s a problem in the electrical grid network, you better believe people start to panic from the blackout.
Even the U.S. Department of Energy accepts that the current network of transmission lines, substations, tranformers and other components making up the electrical grid we all depend on is a patchwork job, stretching its capacity again and again alongside population and growing energy demands and that we must look to new technologies to help create a smarter grid, of which of course, HEMS platforms derived from that demand.
Home energy management devices (HEMS)
Imagine for a moment that you don’t have to remember to turn lights on or off, manage your AC unit or really manage cumbersome devices that suck energy from your home.
Not really hard to imagine, right?
Consumers now have the ability to purchase smart home HEMS products with the ability to do just that. HEMS are basically platforms that combine hardware and software systems to enable users to monitor/manage energy usage and production in order to improve building performance.
If you were to go to CES as we did earlier this year, you would see smart home apps and interfaces for just about everything — many dealing with automating your energy needs. You can tell that there is indeed a race going on to be the platform who has a control panel on your wall or the go-to app in your pocket.
As much as it was a pleasure to meet the innovators and startups when going from booth to booth, we soon realized that the industry is doing excactly what history has done in the past, it’s fulfilling demand by placing patchwork on top of patchwork on top of patchwork — just like the traditional grid system — it’s just digital patchwork now. And it’s already getting messy.
What does that mean??
The concern we have for these new innovations is mainly the compliance and standardization of which the code these technologies rely on. Without diving into the tech terms in this article, think of it like this…what if your energy provider (e.g. PG&E, Excel, etc.) told you that the stove you just picked out and brought home is not compatible with the energy standard in your home. Furthermore, that unless you chose a less desired {insert brand A} over {insert brand B} you would be restricted from self-empowering tools to manage energy costs.
This scenario of course would not go well with consumers, but it is exactly what we see in this new movement toward consumer-facing HEMS products to monitor and manage one’s energy. It’s not universal for all, especially utility-facing.
This is where we decided to do something about it.
Who better to see how an industry repeats itself than a company who has been involved in that history for nearly a quarter of a century.
Simmitri, Inc. a 23-year old company evolved from a roofing and energy efficiency family-owned business to being one of the first solar roofing providers in Northern California and now joining the blockchain revolution, is looking to disrupt the industry by taking a new approach that NO ONE is doing — which is taking a new look at not a more attractive bell or whistle with smart home HEMS devices, but the boring old electrical panel.
When we initially discussed how Simmitri could bring energy efficiency to our clients, we knew the ONLY way we invisioned that would be through our experience with electrical panels. — Jonathan Garcia (COO) | Simmitri
With the meteoric rise of artificial intelligence and blockchain in 2017, Simmitri started to dive in.
Simmitri brought on the brightest in the AI space — leaders of the AI division at Google, scholars across Ivy League schools, electrical engineers, renewable energy inventors and renowned blockchain developers — all to collectively insert their data science backgrounds into how AI + Blockchain would function within our tokenized economy.
After the concept phase was well underway and flushed out, Simmitri hosted a mixer for utility providers to provide feedback on what the macrogrid needed in order to perform better from microgrids. Undoubtedly, the path we were on with the electrical panel was making a whole lot of sense from a utility point of view.
What emerged from all of our R&D, is a newly-revamped, modern century electrical fuse box. With keeping a similar shape and maintaining basic functions of an electrical panel, we decided that it also needed a cosmetic facelift as well as an internal overhaul. Basically, it’s now a computer, hard-lined directly into the electrical network of the building.
What makes the Simbox really stand out (apart from its design) is that it uses AI as a foundation. We’re developing more than just machine learning, but advanced, human-like AI protocol. We call it Simi, an intelligent assistant with the main purpose of managing your energy needs and efficiently distributing power. It is not meant to record your favorite TV show, play Mozart or entertain you — but to learn the family’s behaviors inside the home and adapt programmatically to improve upon the building’s performance.
Adding Artificial Intelligence into the SimBox
Simi will begin her life with behavioral modification through machine learning, proven to be great for regression (prediction) and classifiers — similar to that of a child who is born and raised in a certain environment. The ML algorithms are the building blocks for the AI, which includes multiple machine learning components such as speech recognition (Recurrent Neural Network), image recognition (Convolutional Neural Network) and knowledge representation.
The really interesting part of Simi is having curiosity and human context allowing “it” to be a “her.” — David Cavaros | AI Developer
Simi uses RNN to process text and to find out the intention (e.g. an action to take, information, etc.) That task categorizes the intention as well as fills in the necessary concepts to complete the request of the energy demand. If the request is ambiguous, contradictory, or more information is needed, it will simply ask for the information it requires to complete its task.
Supporting images for more context will be useful and can be done by training a Convolutional Neural Network (CNN) to map the images to the same concepts as the RNN.
Concepts can be trained to match word embeddings (have them both reside in the same dimensional space). This will make the knowledge concepts map directly to words. That way it will be more fluent for Simi to map words to concepts, and images to be translated to concepts as well. Simi also uses k-means clustering to find the clusters (categories) of the knowledge concepts, and that is the way she has a human-readable way of “understanding” the AI’s memory.
As Simi consumes data points from the network, she unpacks what she believes (references) will be the most energy efficient processing protocol to the user.
Adding Blockchain and SIM Tokens
The distributed platform will utilize smart contracts to increase the transparency, network effect and automation. As the SimBox’s backend and operating system is used for storing information about the user behavior and household electricity usage, it will also analyze multiple layers of the data consumed and produced to initiate various distribution tasks based on those metrics.
Remember that the SimBox is both a macro network and microgrid-facing device. When the smartbox receives communication from the SimCloud Platform in regards to macro demands and requests, it will send signals through the building’s circuits to determine which loads meet the criteria for conservation or production. This is what we call efficient distribution.
Let’s give an example…
Let’s say you as a homeowner have an account with a utility provider where you pay each month to continue using the grid’s services. However, during peak times (usually when people return from work) the utility power lines becomes stressed and sends your SimBox a request to conserve energy in your home. The Simbox will then communicate with the electrical nodes throughout the distribution channels and determine which devices can be powered down, minimalized or where power can be drawn from. In doing this automatically, you would be participating in a Distributed Energy Resource (DER) initiative and would be eligible to earn SIM Tokens based on that scenario.
The need for smart contracts on the blockchain will allow for SIM Tokens to be distributed among Platform participants working as follows:
- Smart Contract stores the registry of all wallets used by SimBox owners.
- Data parameters met and tokens transferred to the smart contract for distribution;
- Each wallet tracks in real-time how many tokens are eligible to receive;
- Simmtiri can initiate tokens to be distributed among all the participants within the ecosystem, based on some certain rules (e.g. saved energy percentage from total, energy pool gamification, etc).
- Rules (e.g. kWh of electricity saved, devices turned off during peak times, etc.) can be defined within the smart contract and SimBox wallets participating in the competition defined;
- Reward tokens or stakes are preloaded to the smart contract;
- Every day metrics on data usage relevant to competition is submitted to the smart contract;
- All the participants can track results in the Platform;
In summary, by adding AI into the blockchain we achieve several benefits to the entire network. It is truly a win-win-win scenario. Our participants are happy because we over compensate the user based on automatic energy efficiency and providers are excited because the chances of blackouts during the “duck curve” are largely minimalized.