Improved Incremental Orthogonal Centroid Algorithm for Visualising Pipeline Sensor Datasets

摘要

Each year, millions of people suffer from after-effects of pipeline leakages, spills, and eruptions. Leakages Detection Systems (LDS) are often used to understand and analyse these phenomena but unfortunately could not offer complete solution to reducing the scale of the problem. One recent approach was to collect datasets from these pipeline sensors and analyse offline the approach yielded questionable results due to vast nature of the datasets. These datasets together with the necessity for powerful exploration tools made most pipelines operating companies "data rich but information poor". Researchers have therefore identified problem of dimensional reduction for pipeline sensor datasets as a major research issue. Hence, systematic gap filling data mining development approaches are required to transform data "tombs" into "golden nuggets" of knowledge. This paper proposes an algorithm for this purpose based on the Incremental Orthogonal Centroid (IOC). Search time for specific data patterns may be enhanced using this algorithm.