Online Program

Return to main conference page
Thursday, February 15
PS1 Poster Session 1 and Opening Mixer Thu, Feb 15, 5:30 PM - 7:00 PM
Salons F-I

Wavelet-Based Methods for Data-Driven Monitoring (303585)

View Presentation View Presentation

*Achraf Cohen, University of West Florida 

Keywords: statistical process monitoring, wavelets analysis, data-driven methods.

Statistical process monitoring techniques are important to understand a process variation and determine its current state. The rise of big data technologies has contributed to process large volume, wide variety and velocity of data available in industrial systems. Statistical process monitoring serve as an efficient alternative tool where other approaches (model-based, knowledge-based) may fail to provide satisfactory results, such as in complex systems where it is often tough or impossible to come up with an analytical model. Wavelets based methods are popular nowadays for the goal of fault detection and diagnosis, their significant advantages consist of reducing noise, extracting features, and reducing dimension. Because of their often superior performance, wavelets based methods are widely used for process monitoring in various applications (chemical, electric, image processing, manufacturing, biomedical, etc.). In this work, I will be talking about statistical process monitoring generations, particularly wavelets based methods. I will focus on control charts and machine learning techniques. Illustrative examples will be also presented.