Industrial Multivariate Explorer

A solution with Matlab for time series mining

Abstract

The design and features of a Matlab’s application to support applied researches for serial time computing is presented. The input data can be from historical record coming from chemical and thermal processes and also it can be generated by simulation. Up to 8 signals, linearly normalized and distributed can be visualized on an axis Matlab’s object. By means of two cursors, the user can choose short windows of recorded signals. On this serial time sections, in this version, statisticians are computed and they facilitate the static modeling. They can be saved into an Excel file. It is an opened software application permitting to include new features. The Windows between 2 cursors command facilities the dynamic modeling. Its applicability is exemplified by times series from industry (from a 250 MW thermal power plant) and generated by simulation