Significant memory and computational requirements of large deep neural networks restricts their application on edge devices. Knowledge distillation (KD) is a prominent model compression technique for in which the knowledge trained teacher transferred to smaller student model. The success mainly attributed its training objective function, exploits soft-target information (also known as “dark knowledge”) besides given regular hard labels set. However, it shown literature that larger gap...

In this paper, we propose a novel approach for the data-driven characterization of power system dynamics. The developed method Extended Subspace Identification (ESI) is suitable systems with output measurements when all dynamics states are not observable. It particularly applicable dynamic identification using Phasor Measurement Units (PMUs) measurements. As in case systems, it often expensive or impossible to measure internal components such as generators, controllers and loads. PMU capture...

In this paper, we propose linear operator theoretic framework involving Koopman for the data-driven identification of power system dynamics. We explicitly account noise in time series measurement data and robust approach approximation nonlinear The identified model is used prediction state trajectories system. application illustrated using an IEEE nine bus test

The use of the isolated hybrid power systems is being popular due to continuous increasing gap between demand and supply conventional energy sources intermittent nature non-conventional sources. Normally, source such as wind have induction generator generate electricity but generators require reactive for its operation this continuously changing by variation load power. synchronous used in system generating through diesel supplying partially; therefore, another required fulfill demand. In...

In this paper, we present a novel approach to identify the generators and states responsible for small-signal stability of power networks. To end, newly developed notion information transfer between dynamical system is used. particular, using concept transfer, which characterizes influence various system, are causing instability network. While characterizing from state state, can also describe modes thereby generalizing well known participation factor while at same time overcoming some...

Stability analysis of power system is a problem immense importance in community. Identification the cause for instability relevant and has been studied widely. In this work we provide novel approach, using concept information transfer dynamical system, to identify states generators which are most responsible given network. Our developed notion physically motivated previously shown capture true causality influence. paper, use measure characterize causal interactions influence particular,...

We articulate the reason why dynamic state estimation is needed to push boundaries of modal analysis electric power grids in real-time operation. Then, we demonstrate how unravel linear and nonlinear modes by using extended mode decomposition along with estimates synchronous generators' rotor angles speed deviations from nominal speed. The estimated are associated electromechanical oscillations that take place continuously because imbalances between generation demand. numerical simulations...

Knowledge Distillation (KD) is a model compression algorithm that helps transfer the knowledge in large neural network into smaller one. Even though KD has shown promise on wide range of Natural Language Processing (NLP) applications, little understood about how one compares to another and whether these approaches can be complimentary each other. In this work, we evaluate various algorithms in-domain, out-of-domain adversarial testing. We propose framework assess robustness multiple...

In this paper, we propose a novel approach for the data-driven characterization of power system dynamics. The developed method Extended Subspace Identification (ESI) is suitable systems with output measurements when all dynamics states are not observable. It particularly applicable dynamic identification using Phasor Measurement Units (PMUs) measurements. As in case systems, it often expensive or impossible to measure internal components such as generators, controllers and loads. PMU capture...

This study aims to analyze a novel indirect photoacoustic sensor (PAS) using Machine Learning techniques. The studies focus on understanding the sensor's repeatability, influence of temperature and humidity microphone output voltage, applicability models accurately describe behavior. To behavior, two are carried out in controlled setting. With R2 score 0.964 between voltage gas concentration ppm, first illustrates repeatability for measurements. second looks at how affect voltage. For this...

This study explores the temporal response of an indirect photoacoustic CO <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> sensor, focusing on combined temperature and humidity variations. The intricate relationship between resolution, environmental conditions, sensor repeatability is investigated. Through Studies 1 to 4, impact feature differences explicit values under varying resolutions assessed. A resolution 700 seconds seen at...

Sensor measurements are mission-critical for monitoring and controlling power systems because they provide real-time insight into the grid operating condition; however, confidence in these insights depends greatly on quality of sensor data. Uncertainty is an intrinsic aspect measurement process. In this paper, we develop analytical method to quantify impact uncertainties numerical methods that employ Koopman operator identify nonlinear dynamics based recorded particular, interval each...

Knowledge Distillation (KD) is a model compression algorithm that helps transfer the knowledge of large neural network into smaller one. Even though KD has shown promise on wide range Natural Language Processing (NLP) applications, little understood about how one compares to another and whether these approaches can be complimentary each other. In this work, we evaluate various algorithms in-domain, out-of-domain adversarial testing. We propose framework assess robustness multiple algorithms....

In this paper, we present a novel approach to identify the generators and states responsible for small-signal stability of power networks. To end, newly developed notion information transfer between dynamical system is used. particular, using concept transfer, which characterizes influence various system, are causing instability network. While characterizing from state state, can also describe modes thereby generalizing well known participation factor while at same time overcoming some...