Michael

Michael Yan-Jia ZHAO

 Ph.D. Candidate at Center for Intelligent and Networked System (CFINS)
 Department of Automation, Tsinghua University, Beijing, China, 100084
 Telephone: +86-10-62792491 (Lab)
 Email: zhaoyj@mails.tsinghua.edu.cn
 MSN: michael_zyj@hotmail.com

Research Interests

Performance Evaluation and Optimization of Complex Systems

Discrete Event Dynamic Systems (DEDS) are wide-existed systems in the world. In practice, many man-made systems can be modeled as DEDS, such as manufacture system, communication network, and transportation system. This kind of systems is driven by events and cannot be described through differential equations. We can only evaluation the performance through simulation. Thus, it is a practical and valuable research direction to combine the simulation and optimization effectively. My current research is based on schedule problem in manufacture system.

Stochastic Optimization

Optimization through simulation can be modeled into stochastic optimization. In other words, the objective function and the constraints relate to some stochastic variables, which furthermore can only be observed through simulation and the probability characteristic of which is unknown. Prof. Xi-Ren CAO and Prof. Yu-Chi HO developed Perturbation Analysis (PA) in 1980s, which helps to extract the gradient information from one simulation. Prof. Ho developed Ordinal Optimization (OO) in 1990s, which helps to lessen the computation burden in stochastic optimization through exponential convergence of the order and goal softening. My current research efforts are mainly focused on the perturbation analysis and its combination with Markov decision processes (MDP), event-based optimization, and schedule problem in manufacture system.