MVBatch: A matlab toolbox for batch process modeling and monitoring

Gonzalez-Martinez; José Camacho; Ferrer

Abstract:

A novel user-friendly graphical interface for process understanding, monitoring and troubleshooting has been developed as a freely available MATLAB toolbox, called the MultiVariate Batch (MVBatch) Toolbox. The main contribution of this software package is the integration of recent developments in Principal Component Analysis (PCA) based Batch Multivariate Statistical Process Monitoring (BMSPM) that overcome modeling problems such as missing data, different speed of process evolution and length of batch trajectories, and multiple stages. An interactive user interface is provided, which aims to guide users in handling batch data through the main BMSPM steps: data alignment, data modeling, and the development of monitoring schemes. In addition, a small-scale non-linear dynamic simulator of the fermentation process of the Saccharomyces cerevisiae cultivation is available to generate realistic batch data under normal and abnormal operating conditions. This generator of synthetic data can be used for teaching purposes or as a benchmark to illustrate and compare the performance of new methods with sound techniques published in the field of BMSPM.

Research areas: