Gainesville, FL 32611
University of Wisconsin-Madison
Abstract: Big Data Analytics for Performance Improvement in Complex Systems
In modern complex systems, massive quantities of temporally and spatially dense data are frequently generated due to the rapid advancements of sensor technology and communication networks. Such a data-rich environment poses new and significant challenges of analysis in the following aspects: (i) the access and efficient handling of the rich and heterogeneous data streams that may be contaminated by noises, (ii) the recognition of system knowledge that describes numerous components, complicated interactions, and ever-changing dynamics, and (iii) the effective implementation of the acquired knowledge to enhanced control, planning, and coordination of the systems.
This talk concentrates on big data modeling and monitoring to develop systematic data-driven analytics methodologies for process modeling, quality control, and performance improvement in complex systems. In particular, a Nonparametric Anti-rank based Sampling (NAS) strategy will be introduced to online monitor non-normal big data streams in the context of limited resources, where only a subset of observations are available at each acquisition time. In particular, this proposed method integrates a rank-based CUSUM scheme and an innovative idea that corrects the anti-rank statistics with partial observations, which can effectively detect a wide range of possible mean shifts when data streams are exchangeable and follow arbitrary distributions. Two theoretical properties on the sampling layout of the proposed NAS algorithm are investigated when the process is in control and out of control. Comprehensive simulations and real case studies will be provided to illustrate the effectiveness of the proposed method over existing techniques. In addition, extensions and other related works will also be introduced.
Department of Industrial and Systems Engineering at the University of Florida