Chemometric Methods in Process Analysis

Abstract

Chemometrics may be loosely defined as the development and application of mathematical and statistical techniques to aid in the analysis of chemical data. The goal of chemometrics is to guide the chemist in the optimum extraction of chemically relevant information from a series of complex chemically oriented observations. To this end, this article addresses a wide array of chemometric tools that are applicable to industrial process analysis. The three main sections present statistical and mathematical tools in the context of univariate, multivariate, and multiway analyses. Each of the three sections discusses the benefits and limitations of qualitative analysis, quantitative analysis, and process monitoring with the class of data in question. Additional sections address the appropriate pretreatment of collected data to aid in efficient analysis and validation of the derived chemometric models. In these sections methods for background correction and instrument standardization are discussed.

Advances in Mid-Infrared Spectroscopy for Chemical Analysis

“…Consequently, either more or less sophisticated sample pretreatment techniques reducing the complexity of the matrix and/or multivariate data processing algorithms have emerged as an integral strategy of modern MIR spectroscopy. Although a discussion on multivariate calibration, data evaluation, and data mining schemes is beyond the scope of this review, the interested reader is referred to several exemplary papers outlining the importance of chemometrics in MIR spectroscopy (3,6,(14)(15)(16)(17)(18)(19).…”

Frequency/voltage wide‐area measurements over transmission control protocol/internet protocol communication network for generator trip identification concerning missed data

“…These values of these components are used as the signature of Kuraymat station trip. As given in (13), each row represents the PC values for the signature of the corresponding FDR for training case-I. The first matrix corresponds to the frequency measurements and the other corresponds to the voltage measurements.…”

Monitoring of water quality in South Paris district by clustering and SIMCA classification

“…It reduces the significant dimensionality of a data matrix, allowing to retain most of the original information content. The original variables (X) are transformed by linear combination in a limited number of new variables, called principal components (PCs) [26]:…”