The combination of the OSIsoft PI System with MATLAB offers substantial potential for powerful, well-structured data analysis and reporting. A modular integration of the PI Data Archive, PI AF and MATLAB was developed at Stellenbosch University, leveraging the MATLAB .NET API and MATLAB object oriented syntax to interface with AF objects. The integration formed part of a performance monitoring framework developed at Stellenbosch University for Anglo American Platinum in collaboration with OPTINUM solutions of South Africa. In addition to the interface layer that enabled integration of the framework to the PI Data Archive, a second AF-MATLAB interface layer presented classes from the AF SDK as MATLAB classes inside the MATLAB application. A configuration map was designed and developed in Excel as a user-friendly translation from AF elements and attributes to units, sensors and properties in the MATLAB application. The final product is in daily use at Anglo American Platinum concentrator and smelter plants, and generates operational performance reports on weekly and daily data episodes of production at these plants. A case study of PGM concentrator monitoring demonstrates the application of the framework. This integration of the PI Data Archive, PI AF and MATLAB exploits the best features of each of these technology stacks in a customizable, stand-alone tool for advanced data analysis and reporting.
Speaker
JP Barnard
JP BARNARD, PhD is Technical Manager at the Centre for Process Monitoring, Process Engineering, Stellenbosch University, where he acts as program manager, software development manager and administrator of computational infrastructure.
Barnard re-joined Stellenbosch University, his alma mater, in 2008 as technical manager of the newly created Centre for Process Monitoring. Previously, he was employed by Ansoft Corporation (now part of ANSYS) and Mathworks, where he participated in the development of the software products of these companies. Barnard obtained his PhD in Engineering at the University of Stellenbosch in 2000.
He redesigned and developed a MATLAB framework for process monitoring, for Anglo American Platinum. He managed the project, did system design as well as implemented core segments of the solution. As part of the solution, he developed a MATLAB interface to OSIsoft PI AF, targeting the AF .NET SDK.
He leads and oversees the development of data mining tools, written mostly in MATLAB, integration with .NET applications; data historians, such as InSQL and PI System. He leads an R&D project that aims to implement process performance monitoring techniques in AF object templates that leverage MATLAB for advanced data analytics and reporting.
On the side, he is configuring a MATLAB distributed computing cluster in Microsoft Azure, as well as in the local HPC cluster on campus, in collaboration with Microsoft, Mathworks and the IT Unit at Stellenbosch University.
Barnard supports post-graduate students at Process Engineering on their research projects, concerning computational and optimization problems.