How Humans Digest Millions of Data Points a Minute
A problem that us humans have with using “big data” is how to digest all the interesting information. On wind turbines, there are hundreds of data points being produced a second. A single person may be tasked with monitoring and managing thousands of such machines. Digestion of this data is an important problem to solve if we want to use big data to make better decisions.
Fortunately, this problem has been more or less solved; and teenagers and day traders are probably the most prolific users of the solutions. We have taken these disparate best practices and melded it into our latest software release. Planned to hit platforms of all current customers on October 1st.
“Google Finance on top; Facebook on the bottom”
News Feed. Every day millions of teenagers log on to a combination of Twitter, Facebook, Instagram, to keep up to date on the happenings of their hundreds of friends and interests. This involves receiving operational updates from those they follow, commenting, assigning priority, and redirecting to other colleagues. A key to the usefulness of this “news feed” format is that it delivers a hierarchy of the incoming material: so that users can quickly get up to date on whats “most important”. This is what makes Facebook functional amongst all the noise and data.
We have adopted many of these concepts in how we display and allow users to interact with wind turbine operational data. We effectively convert each SCADA alert, work order, etc, into a post, that is stored on each turbine page. These posts contain all the relevant information on the event. An operator can go to their homepage to see a listing of the most relevant posts happening across all the machines he or she follows. (Relevance is an algorithm based on start time, downtime, component effected, fault category, accuracy of information, etc). The operator can see what other comments or actions technicians have made, and alongside have an unparalleled set of corresponding data.
There are three types of “posts” that are created in real-time:
Machine generated operation posts from SCADA or work orders are decoded and displayed. This is a tool to not only better understand and view the history of a given asset, but also act upon that knowledge and communicate immediately.
User generated posts can be created, assigned, and prioritized as operators identify an issue and want to communicate to those responsible.
Predicted operation posts are uncovered via Fluitec Wind’s accurate predictive analytic system. These posts are delivered to allow operators to plan ahead and optimally forecast.
All of the above can be found on the operator’s online portal; and settings are available to provide email alerts according to specific needs. Such as, “email me if any of the turbines I follow is down for more than 10 minutes”.
One may recognize that looking, commenting, and digesting a friend’s family vacation pictures, is conceptually similar to digesting a set of borescope pictures from Turbine A4.4. Applying some of the same tools that allow such social activities to be efficient and enjoyable can now help wind turbine monitors improve performance.
Stock Charts have been a vehicle to display years of variable data, with nuance, and provide an ability to effectively compare.
Our latest software incorporates stock chart concepts to display real time SCADA data alongside operation events. Just as a stock chart alongside important news allows an investor to understand the history of an asset. We importantly improve the ability to forecast the performance of the asset, by providing predicted events as well. Imagine if an investor could be alerted to when the next earnings miss would happen? We can do that for wind turbine O&M costs.
We are excited about our ability to take known big data handling methods and apply them to reduce the cost of renewable energy. We are at the forefront of making data usable for machines and humans.
It's one small step for machines, but one giant leap for those reading this.