Engineering Commons^™

A Data-Driven Multivariate Process Monitoring Platform For Knowledge Discovery And Model Building In Industrial Applications, Estelle E. Seghers

LSU Master's Theses

In industrial chemical manufacturing processes, the amount of raw data generated can add complexity in the analysis and understanding of the process dynamics. Being able to properly interpret this data can help improve plant operation, especially regarding safety and profitability. This research has culminated in FastMan-JMP, a platform proposed for monitoring of industrial processes and optimization of the offline data-driven model-building process as part of the process monitoring workflow. FastMan-JMP is a tool developed in Python to apply various data mining and machine learning techniques quickly and easily to better understand valuable patterns and hidden trends in process data. One …

On I/O Performance And Cost Efficiency Of Cloud Storage: A Client's Perspective, Binbing Hou

LSU Doctoral Dissertations

Cloud storage has gained increasing popularity in the past few years. In cloud storage, data are stored in the service provider’s data centers; users access data via the network and pay the fees based on the service usage. For such a new storage model, our prior wisdom and optimization schemes on conventional storage may not remain valid nor applicable to the emerging cloud storage.

In this dissertation, we focus on understanding and optimizing the I/O performance and cost efficiency of cloud storage from a client’s perspective. We first conduct a comprehensive study to gain insight into the I/O performance behaviors …

Process Monitoring And Data Mining With Chemical Process Historical Databases, Michael Carl Thomas

LSU Doctoral Dissertations

Modern chemical plants have distributed control systems (DCS) that handle normal operations and quality control. However, the DCS cannot compensate for fault events such as fouling or equipment failures. When faults occur, human operators must rapidly assess the situation, determine causes, and take corrective action, a challenging task further complicated by the sheer number of sensors. This information overload as well as measurement noise can hide information critical to diagnosing and fixing faults. Process monitoring algorithms can highlight key trends in data and detect faults faster, reducing or even preventing the damage that faults can cause. This research improves tools …

A Study Of Distributed Clustering Of Vector Time Series On The Grid By Task Farming, Arun B. Nayar

LSU Master's Theses

Traditional data mining methods were limited by availability of computing resources like network bandwidth, storage space and processing power. These algorithms were developed to work around this problem by looking at a small cross-section of the whole data available. However since a major chunk of the data is kept out, the predictions were generally inaccurate and missed out on significant features that was part of the data. Today with resources growing at almost the same pace as data, it is possible to rethink mining algorithms to work on distributed resources and essentially distributed data. Distributed data mining thus holds great …

Data Mining And Knowledge Discovery: A Guided Approach Base On Monotone Boolean Functions, Vetle Ingvald Torvik

LSU Doctoral Dissertations

This dissertation deals with an important problem in Data Mining and Knowledge Discovery (DM & KD), and Information Technology (IT) in general. It addresses the problem of efficiently learning monotone Boolean functions via membership queries to oracles. The monotone Boolean function can be thought of as a phenomenon, such as breast cancer or a computer crash, together with a set of predictor variables. The oracle can be thought of as an entity that knows the underlying monotone Boolean function, and provides a Boolean response to each query. In practice, it may take the shape of a human expert, or it …