Bin Zhang, PhD
University of South Carolina,
College of Engineering and Computing
Department of Electrical Engineering
Prof. Bin Zhang
Dr. Zhang obtained the PhD degree in Electrical Engineering at Nanyang Technological University, Singapore. He has over 20 years of experience in prognostics and health management, intelligent systems and control, robotics, machine learning. Particularly, he has developed and applied these techniques to condition monitoring, control, resilience and intelligence to power grids, batteries, manned/unmanned systems. He has published more than 200 technical papers in his areas of expertise. He is/was editors for many prestigious journals and member of steering committee of many international conferences.
Email: zhangbin@cec.sc.edu
Office: SWGN 3A22
Phone: 803-777-8335
FOLLOW
Recent Work
Most recent work
Battery modeling
We built and verified lithium-ion battery equivalent circuit model (ECM) and electrochemcial model (EMs), which include single particle model and simplified first principle model. used it for battery state-of-charge and remaining-discharge-time prediction. The models are used for battery state-of-charge and remaining-discharge-time prediction. We also consider the aging of battery in this process by integrating capacity degradation, i.e., state-of-health.
Power electronics
We worked on fractional order delay and virtual variable sampling (VVS) for repetitive control (RC) and its applications to power electronics control. Fractional order delay approximates the non-integer delay part by building a finite impulse response filter, which is able deal with both period delay unit and phase lead compensation. The VVS approximates a variable sampling delay unit instead of the fixed system delay unit for RC and its filters, in which RC is able to be frequency adaptive. The VVS method can provide much larger adjustable range on reference frequency fluctuation. To further improve the stability and eliminate harmonic distortions efficiently, VVS is integrated with Discrete Fourier transform-based and \(nk\pm m\)-based selective harmonic RC schemes.
Deep learning methods
Out team had developed methods with graph neural network, deep residual convolutional neural networks (DRCNN), deep belief networks (DBN), long-short-term memory (LSTM) networks, parametric rectified linear unit (PReLU)-DBN, multi-layer extreme learning machine, multi-task CNN, and transfer learning for monitoring applications to many systems including batteries, bearing, cable, analog circuits, etc.
Featured Video
add related explanation here.