A5101484129- Smart Grid Energy Management
- Building Energy and Comfort Optimization
- Anomaly Detection Techniques and Applications
- Reinforcement Learning in Robotics
- Indoor and Outdoor Localization Technologies
- Smart Grid Security and Resilience
- Adversarial Robustness in Machine Learning
- Microgrid Control and Optimization
- Power System Optimization and Stability
- Network Security and Intrusion Detection
- Advanced Graph Neural Networks
- Optimal Power Flow Distribution
- Traffic Prediction and Management Techniques
- Fault Detection and Control Systems
- Transportation Planning and Optimization
- Domain Adaptation and Few-Shot Learning
- Advanced Bandit Algorithms Research
- Context-Aware Activity Recognition Systems
- Air Quality Monitoring and Forecasting
- Model Reduction and Neural Networks
- Human Mobility and Location-Based Analysis
- Complex Network Analysis Techniques
- Gaussian Processes and Bayesian Inference
- Adaptive Dynamic Programming Control
- Power Systems and Renewable Energy
Virginia Tech
2018-2025
Monash University
2021-2024
Chongqing University
2023
Sichuan Provincial Architectural Design and Research Institute (China)
2023
University of California, Berkeley
2014-2020
Henan Provincial Health Bureau
2020
Lawrence Berkeley National Laboratory
2016-2019
Berkeley College
2017
University of California System
2016
Alstom (France)
2012-2014
WiFi fingerprinting-based indoor positioning system (IPS) has become the most promising solution for localization. However, there are two major drawbacks that hamper its large-scale implementation. First, an offline site survey process is required which extremely time-consuming and labor-intensive. Second, RSS fingerprint database built vulnerable to environmental dynamics. To address these issues comprehensively, in this paper, we propose WinIPS, a WiFi-based non-intrusive IPS enables...
Sensing by proxy (SbP) is proposed in this paper as a sensing paradigm for occupancy detection, where the inference based on “proxy” measurements such temperature and CO2 concentrations. The effects of occupants indoor environments are captured constitutive models comprising coupled partial differential equation-ordinary equation system that exploits spatial physical features. Sensor fusion multiple environmental parameters enabled framework. We report experiments conducted under simulated...
Non-intrusive presence detection of individuals in commercial buildings is much easier to implement than intrusive methods such as passive infrared, acoustic sensors, and camera. Individual power consumption, while providing useful feedback motivation for energy saving, can be used a valuable source detection. We conduct pilot experiments an office setting collect individual data by ultrasonic acceleration WiFi access points, addition the monitoring data. PresenceSense (PS), semi-supervised...
To ensure grid efficiency and reliability, power system operators continuously monitor the operational characteristics of through a critical process called state estimation (SE), which performs task by filtering fusing various measurements collected from sensors. This study analyzes vulnerability key operation module, namely ac-based SE, against potential cyber attacks on data integrity, also known as false injection attack (FDIA). A general form FDIA can be formulated an optimization...
Graph representation learning plays a vital role in processing graph-structured data. However, prior arts on graph heavily rely labeling information. To overcome this problem, inspired by the recent success of contrastive and Siamese networks visual learning, we propose novel self-supervised approach paper to learn node representations enhancing self-distillation with multi-scale learning. Specifically, first generate two augmented views from input based local global perspectives. Then,...
The detection of anomalies in multivariate time series data is crucial for various practical applications, including smart power grids, traffic flow forecasting, and industrial process control. However, real-world usually not well-structured, posting significant challenges to existing approaches: (1) existence missing values along variable dimensions hinders the effective modeling interwoven spatial temporal dependencies, resulting important patterns being overlooked during model training;...
Smart buildings today are aimed at providing safe, healthy, comfortable, affordable, and beautiful spaces in a carbon energy-efficient way. They emerging as complex cyber-physical systems with humans the loop. Cost, need to cope increasing functional complexity, flexibility, fragmentation of supply chain, time-to-market pressure rendering traditional heuristic ad hoc design paradigms inefficient insufficient for future. In this paper, we present platform-based methodology smart building...
Building control is a challenging task, not least because of complex building dynamics ad multiple objectives that are often conflicting. To tackle this challenge, we explore an end-to-end deep reinforcement learning paradigm, which learns optimal strategy to reduce energy consumption and enhance occupant comfort from the data building-controller interactions. Because real-world policies need be interpretable efficient in learning, work makes following key contributions: (1) investigated...
We investigate the important problem of certifying stability reinforcement learning policies when interconnected with nonlinear dynamical systems. show that by regulating partial gradients policies, strong guarantees robust can be obtained based on a proposed semidefinite programming feasibility problem. The method is able to certify large set stabilizing controllers exploiting problem-specific structures; furthermore, we analyze and establish its (non)conservatism. Empirical evaluations two...
A method is presented to learn neural network (NN) controllers with stability and safety guarantees through imitation learning (IL). Convex conditions are derived for linear time-invariant systems NN by merging Lyapunov theory local quadratic constraints bound the activation functions in NN. These incorporated IL process, which minimizes loss, maximizes volume of region attraction associated controller simultaneously. An alternating direction multipliers based algorithm proposed solve...
Non-intrusive presence detection of individuals in commercial buildings is much easier to implement than intrusive methods such as passive infrared, acoustic sensors, and camera. Individual power consumption, while providing useful feedback motivation for energy saving, can be used a valuable source detection. We conduct pilot experiments an office setting collect individual data by ultrasonic acceleration WiFi access points, addition the monitoring data. PresenceSense (PS), semi-supervised...
In numerous reinforcement learning (RL) problems involving safety-critical systems, a key challenge lies in balancing multiple objectives while simultaneously meeting all stringent safety constraints. To tackle this issue, we propose primal-based framework that orchestrates policy optimization between multi-objective and constraint adherence. Our method employs novel natural gradient manipulation to optimize RL overcome conflicting gradients different tasks, since the simple weighted average...
We develop a data driven, partial differential equation-ordinary equation model that describes the response of carbon dioxide (CO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> ) dynamics inside conference room, due to presence humans, or user-controlled exogenous source CO . conduct three controlled experiments and tune whose output matches measured concentration when known inputs are applied model. In first experiment, amount gas is...