Abstract
This paper analyzes Hat, an open-source framework for developing event-driven component-based SCADA applications, and discusses possibilities to add various analytical tools to such platforms. As a part of the contribution, an open-source component called Artificial Intelligence Model Manager (AIMM) has been developed and integrated into a Hat-based SCADA platform. AIMM is extensible through various plugins, allowing the addition of various models for advanced analytics e.g., machine learning tools, statistical tools, etc. The paper describes AIMM architecture and provides a use case in which state estimation was performed in a medium-voltage distribution grid. This case study demonstrates that it is possible to extend component-based SCADA systems with components for advanced analytics with minimal fundamental system changes.
Highlights
Supervisory Control and Data Acquisition (SCADA) systems have undergone many changes over the various generations of software development trends
We propose the addition of a new component, Artificial Intelligence Model Manager (AIMM), which specializes in advanced analytics
AIMM is integrated into an event-driven component-based SCADA system, and there is little research analyzing the integration of advanced analytics into such an architecture to solve problems the power industry faces
Summary
Supervisory Control and Data Acquisition (SCADA) systems have undergone many changes over the various generations of software development trends. Event-driven component-based architecture principles are not without drawbacks, such as higher data flow complexity and greater communicational overhead Despite these shortcomings, they find application in various industrial systems, including the example used in our case study, power grid management. Another group of techniques, focused on inference from historical data, has seen a large increase in usage These mainly include methods from the field of machine learning, such as neural networks, various types of linear models, or deep learning. We refer mainly to the increasing use of Bayesian statistics, which allows inferred processes to be modeled using methods such as structured time series, or Gaussian process regression, and optimized using approaches such as Markov chain Monte Carlo or variational inference [6] In these datadriven techniques, we include methods from the field of soft computing, such as fuzzy logic or evolutionary computation. We use a method based on weighted quadratic loss optimization, but other advanced analytical methods such as deep learning or Bayesian statistics can be used
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
