Browsing by Author "St. Leger, Aaron"

Now showing 1 - 4 of 4

Metadata only
A Comparative Study of Programming Languages for a Real-Time Smart Grid Application
(IEEE, 2023-11-13) Rooney, Michael P.; Rao, Nakul; Liebers, Nicholas; St. Leger, Aaron; Matthews, Suzanne J.
With security an increasing concern, SCADA system designers should consider the programming language used to implement critical smart grid applications. In this paper, we compare the performance of an anomaly detection workflow implemented in a common programming language used in SCADA systems (C) to equivalent implementations in three less commonly-known languages (Numba Python, Cython, and Rust). We benchmark our implementations on two real-world datasets of synchrophasor data and compare their performance on two Arm-based single board computers. Our results demonstrate that the Numba Python implementations achieve real-time performance in many contexts that pure Python counterparts cannot. In all tested scenarios, the Rust implementations achieve real-time performance while consuming similar amounts of power to their C counterparts. Our results suggest that SCADA designers should take a closer look at Numba Python and Rust for performant WAMS applications.
Metadata only
Blockchain for Power Grids
(IEEE, 2019-04) Banks, Christian; Kim, Samuel; Neposchlan, Michael; Velez, Nicholas; Duncan, Kate J.; James, John; St. Leger, Aaron; Hawthorne, Daniel S.
Sharing information is an important part of regulating and maintaining efficient and safe power grids. This project's goal is to develop a way of using blockchain technology to share transaction information among different power grids in a secure, controlled, monitored, and efficient manner. The biggest concern regarding the data is integrity. By leveraging blockchain technology, the data will be reliable and resilient to attacks, such as man-in-the-middle and data spoofing attacks. The Hyperledger Fabric implementation provides a permissioned network in which power grids will act as nodes that maintain ledger information. By using a distributed ledger to validate transactions though the process of consensus, the system can share information in a manner that is more secure and transparent than traditional information sharing systems in which data is less secure and takes longer to validate. The additional layers of security and speed that Hyperledger technology provide help to prevent issues, such as power grid failures, that could stem from the latency or integrity issues involved with traditional methods of validating, processing, and reacting to shared data.
Metadata only
Comparing the Performance of Numba and CUDA for Historical Analysis of Synchrophasor Data
(IEEE, 2024-02-19) Rao, Nakul; Liebers, Nicholas; St. Leger, Aaron; Matthews, Suzanne J.
Modern Wide Area Monitoring systems (WAMS) incorporating Phasor Measurement Unit (PMU) technology are producing big datasets. Historical analysis of PMU data is beneficial in development of online WAMS applications, quantifying baseline normal performance, and discovering anomalous events. Energy and time-efficient computational techniques are beneficial for historical analysis of PMU data. Application workflows that include historical analysis typically combine higher-level (but slow) languages like Python with faster (but older) languages like C. This paper compares the performance of Numba Python and C for historical analysis of PMU data, on both the CPU and GPU. We augment a known PMU anomaly detection scheme with linear state estimation, implement it separately in Numba and C, test the approaches on two real-world datasets, and measure their performance on the CPU and GPU of the NVIDIA Jetson Xavier single board computer, varying the available power modes. Results demonstrate that while Numba is significantly faster than traditional Python, simplifies application development, and holds promise for PMU applications, there is a noticeable performance gap between Numba and C on the GPU.
Metadata only
Implementing a Full-state Feedback Laboratory Exercise in an Introductory Undergraduate Control Systems Engineering Course
(ASEE, 2019) Bluman, James E.; St. Leger, Aaron; Korpela, Christopher M.
Many mechanical engineering undergraduate students find the study of control systems engineering to be one of the more challenging subjects that they encounter. These challenges include working in the Laplace and frequency domains, learning new analysis techniques, as well as the breadth of topics that are typically covered in an introductory control systems undergraduate class. The challenges faced by instructors consist of deciding which material to include, balancing the depth and breadth of understanding various topics, selecting the best learning activities for each technique, and providing meaningful hands on experimentation in a predominately theoretical course. Fortunately, control systems engineering is amenable to instruction through laboratory exercises, where students can try different control techniques and observe their effectiveness nearly in real-time. Some effort is required to adequately link theory to experimentation in a theoretical introductory course. In this paper, we describe the implementation of a new full-state feedback laboratory exercise which was designed to illustrate the efficacy of full state control of a fourth order system. The general process of modeling, simulating the system, controller development, then deployment and evaluation in the lab is a common pedagogical process in control systems engineering education. The importance of visualization, in the context of using information technology, is discussed in Bencomo (2003). The laboratory exercise in view utilizes the same aforementioned process with an emphasis on visualizing system performance in state feedback control. The students first complete a pre-lab exercise which covers the modeling, control design, and simulation. Then they utilize commercially available software-hardware package that allows them to deploy their design and observe its real world performance. Specifically, the prelab begins by requiring them through modeling the dynamics of the electro-mechanical system. Furthermore the students then design the controller gains in a full state feedback in order to achieve a desired transient response. They then model the system in SIMULINK prior to coming to the lab, and analyze the effectiveness of their control design. The pre-lab assignments are submitted by the students, graded by the instructor, and then returned in the laboratory. In the laboratory, the students walk through a series of exercises beginning with the open loop response and ending with full state feedback in a closed loop sense. The intermediate steps allow the students to observe the improvements in the response of the system. The students are also introduced to signal processing requirements, for example the need to filter a differentiated signal. The novelty in this exercise lies in the procedural implementation of state feedback (no feedback, partial state feedback(s), and full state feedback with estimation) and evaluation of performance. Specifically, through visual observation of system performance and quantification of system performance through data acquisition and analysis. The full paper will provide the details of the laboratory including implementation instructions and lessons learned through conducting this laboratory exercise with students.