Dr. Hadis Karimipour is the director of Smart Grid Lab in the School of Engineering at the University of Guelph. Before joining the University of Guelph, she was a postdoctoral fellow in University of Calgary working on cyber security of the smart power grids. She is currently an Assistant Professor at the School of Engineering, Engineering Systems and Computing Group at the University of Guelph, Guelph, Ontario. Her research interests include large-scale power system state estimation, cyber-physical modeling, cyber-security of the smart grids, and parallel and distributed computing. The overall objective of her research is to investigate and overcome the challenges associated with the real-time monitoring and control of the large scale systems and smart grids. The main areas of her research includes:
Smart Grid Cybersecurity
The term “smart grid” in the context of power systems refer to a modernized electricity generation, transmission and distribution infrastructure. A smart grid can be described as a power system having bidirectional communications facilitated through the use of advanced sensing/metering devices and advanced control technologies. As a result of deployment of new smart grid technologies, the electric power industry is faced with new and changing cybersecurity threats, vulnerabilities, and the need for requirements applicable to the smart grid, both broadly and in specific areas such as applied cryptography, and cybersecurity for microgrids. The research plan is to conduct research that will enable the development of industry standards and guidance in order to successfully implement secure Smart Grid technologies.
Machine learning (ML) is a category of algorithm that allows software applications to become more accurate in predicting outcomes without being explicitly programmed. The basic premise of machine learning is to build algorithms that can receive input data and use statistical analysis to predict an output while updating outputs as new data becomes available. Things like growing volumes and varieties of available data, computational processing that is cheaper and more powerful, and affordable data storage are the main motivations toward machine learning developments.
Risk Mitigation for Dynamic State Estimation
Power systems are monitored by supervisory control system which gathers all sorts of measurements from different parts of the power system. The information provided by the system may not always be true due to the presence of errors in the measurements, telemetry failures, etc. Moreover, the collected set of measurements may not allow direct extraction of the parameter of interest. Therefore, the control centre relies on Dynamic State Estimation (DSE) to estimate/predict the states of the system. DSE is a key Energy Management System (EMS) function that model the time varying nature of the system, thus its security and reliability is paramount for power system functionality.
GPU Parallel Programming
There exist two major processor architectures: the multi core computer processing unit (CPU) and the many core graphic processing unit (GPU). The GPU is composed of hundreds of computational cores known as stream processors (SPs), unlike the CPU with a limited number of arithmetic cores. The popularity of the GPUs in the field of high-performance computing is due to their ability to provide computational power for massively parallel problems at a reduced cost. Utilizing the massively parallel architecture of the GPU, by assigning separate tasks to individual compute unified device architecture (CUDA) threads, computationally intensive sections of the state estimation program can be converted into a CUDA kernel (functional program which generates many threads for data parallelism). Therefore, all the task can be off-loaded and executed in parallel utilizing thousands of threads, thereby accelerating the process of data processing and cybersecurity analysis significantly.
Prospective Graduate Students
I am always looking for motivated and enthusiastic students with strong background in electrical engineering. The required skills for potential graduate students include:
- Strong background in control systems, smart grid and cyber-security
- Algorithm development and programming (C/C++)
- Experience using MATLAB/Simulink environment
- Background in detection and estimation theory, and machine learning
- Strong oral and written communication skills
Project 1: Machine Learning Based Anomaly Detection in Smart Grids
Nowadays, critical infrastructures are joined to the Internet for centralized control and management which created a great potential for unauthorized access and exposed these systems to the same vulnerabilities that plague traditional computer systems and networks. This project aimed at exploring the viability of machine learning methods in terms of their ability to identify anomaly or cyber-attacks in smart grids. Multiple learning methods will be evaluated using a dataset of Remote Terminal Unit communications, which included both normal operations and data injection attack scenarios. We will prove that machine learning algorithms can leverage non-linear relationships between smart power grid measurements to differentiate between malicious, and non-malicious disturbances. The intent of this project is to determine an optimal algorithm that is accurate in its classification to provide reliable decision support to a power system operator.
Candidates should have a strong academic background, proficiency in mathematical modeling, anomaly detection, computer programming and simulation.
Project2: Robust controller design for cyber-attacks detection and elimination
Despite all of the efforts towards protecting smart grid against cyber-attack, they are leaving irreparable damages on the system. The goal of this project is to combine advanced control strategies (feedback control and sliding mode control) with cyber-security analysis to enable rapid diagnosis, and appropriate responses to any event. The effect of renewable resources and storage on power system operation will be examined, including balancing authority functions, automatic generation control, and market operation. This will improve fault detection rates and significantly reduce misclassifications.
Candidates should have a strong academic background, proficiency in mathematical modeling, control system design, computer programming and simulation, and familiarity with renewable energy technologies.
Email : hkarimi-@-uoguelph-dot-ca
Phone: +1 (519) 824-4120 ext. 52506
Address: University of Guelph-50 Stone Road East,
Guelph, Ontario, Canada N1G 2W1