To Recap: All devices have energy dashboards, The assumption is that the device is in a constant error state while the dashboard represents the optimum operational state. Of course the dashboard could malfunction, but for the sake of this study let us assume it does not. The word device must be defined. it is any entity that consumes energy for the purposes of this research topic. The scope of devices therefore would include, automobiles, household appliances, human beings, animals just to name a few. We will fit these use cases in our research as we progress.
My initial direction was to establish standards for interoperability of energy dashboards. The assumption was that dashboards and devices had to interoperate, so do dashboards and sources of information and interoperability standards would therefore also govern the algorithms that adductively establish what the delta between the dashboard and the device is at any given moment. Further the interoperability standards would establish methods to reduce the error and at the same time develop standards of what the acceptable error would be. Finally since he standards should possess the capability to self moderate or self regulate or even change as it learns.
My initial direction was not explicit about the body of research that has gone into machine learning, semantic web and big data particularly since the year 2000. Google’s approach to indexing the web of billions of documents and Tim Berners Lee’s Semantic web of abductive reasoning came around the same time, however it was only 5 or 6 years later that Doug Cutting and Yahoo opensourced hadoop and mapreduce. Doug/Yahoo’s contribution was a key milestone as it made it possible for parallel computations on embarrassingly large datasets available for collaborative development.
Standards of interoperability for energy dashboards are perfect candidates for embassingly large parallel computational challenges. Assuming that all data is on the web and will be accessible, one can see how a dashboard can establish the size of the error between the device and itself and take action to self correct or alert. it is no more than taking what we do today, to the next step. Let us take a use case of device = Human being today. If we assume the error condition is a stain on white shirt. The device does a search, on google, or any other search engine, assimilates the necessary information on how to correct the error or reduce the error and proceeds with the action.
In the use case above Instructions would be the class, “Removal of”, “Stains”, “from white shirt” would be the features. Once the page rank algorithm generates the highest ranking instructions, the user acts upon it by scanning through the highest page ranks, and acting out the instructions.
if our use case = Refridgerator and let us assume the error condition is that the light bulb does not go out when the door is closed, The Refridgerator could follow a similar model as above. The classifier model still work’s the correction of the error condition might require a trigger to a supply/chain process.
From both use cases, we see Big Data and Energy Dashboards will play a large role in the future of automation and Dashboards for a greener built environment.