Speaking at the PPDM Oklahoma City data management luncheon, Jason McKittrick introduced a joint production analytics offering from Microsoft, OSIsoft and Neal Analytics. Following a quick tour of Excel’s pivot tables, graphics and (Bing) data maps, McKittrick showed how Microsoft’s Power BI can be used to link on site data with business intelligence functionality in the cloud via a ‘data management gateway.’
The Microsoft tools now link to the plant or platform via OSIsoft’s PI System, notably with a new ‘self-service BI for PI offering.’ Other novelties on the horizon include machine learning in the Microsoft Azure ‘data factory,’ a PI integrator for Azure and an ability to combine textual information with real time operational data for KPIs in-context.
Excel is presented as the client of choice for what proved to be a rather bewildering amount of slideware. But the machine learning self-service data science available in PI System 2015 looks pretty compelling with interfaces for R, Python and even open computer vision, OpenCV!
Bill Barna then showed how some of the above components have been leveraged in Neal Analytics’ oil and gas predictive analytics offering. This includes tank level monitoring and forecasting using Azure machine learning. Neal uses neural networks, Poisson and ‘decision forest regression’ along with model cross validation to ensure a model generalizes well to new tanks. Presentations avaliable from PPDM.
In an interesting announcement from a different slide deck, Microsoft reports that Rockwell Automation is using Microsoft Azure ‘Internet of Things’ to connect disparate systems in the petroleum supply chain. Rockwell client Hilcorp Energy monitors ‘cloud-connected’ submersible pumps from from the control rooms.
© Oil IT Journal - all rights reserved.