PLS

by

SciX 2025 Conference – On Machine Learning, PLS, and Local Weighting

Venue: Northern Kentucky Convention Center, Covington, KY

Date: Oct 5 – 10, 2025

Brian Rohrback will be presenting in the session on Artificial Intelligence and Machine Learning in Process Analytical Technology (PAT). See abstract below.

On Machine Learning, PLS, and Local Weighting

AUTHOR: Brian Rohrback
ABSTRACT:
A dozen years ago, Infometrix embraced the target of completely automating the installation and maintenance of any optical spectrometer and for any application. Ultimately, the goal is to identify a generic machine learning approach that can be taken to mimic the results that an experienced chemometrician would achieve if charged with producing an optimized model.  The idea has been presented in numerous publications and this specific work was triggered initially by Workman et al. (1995). Clearly, the chemometrics tasks can be broken down, assigning best practices procedures for each. One part of this process is choosing the algorithmic approach. Partial Least Squares (PLS) is the workhorse and is incorporated into nearly every spectroscopy system.   In cases of non-linearity, a locally weighted application of PLS will avoid the failures of non-linear methods.  Local models can also simplify model maintenance as conditions (spectrometer, ingredients, unit operations) change.

Automation of Local Regression Model Building for Spectroscopic Data – Journal of Chemometrics

by

ISA 2020 – Rethinking Calibration for Process Spectrometers II

The Long Beach Convention Center
Long Beach, CA
1:30pm, April 27th

 

Brian Rohrback
Infometrix, Inc.
Will Warkentin
Chevron Richmond Refinery

 

KEYWORDS
Best Practices, Calibration, Cloud Computing, Database, Gasoline Blending, Optical Spectroscopy, PLS, Process Control

ABSTRACT
Optical spectroscopy is a favored technology to measure chemistry and is ubiquitous in the hydrocarbon processing industry. In a previous paper, we focused on a generic, machine-learning approach that addressed the primary bottlenecks of mustering data, automating analyzer calibration, and tracking data and model performance over time. The gain in efficiency has been considerable, and the fact that the approach does not disturb any of the legacy (i.e., no changes or alterations to any analyzer or software in place) made deployment simple.

We also standardized a procedure for doing calibrations that, adheres to best practices, archives all data and models, provides ease of access, and delivers the models in any format. What remains is to assess the speed of processing and the quality of the models. To that end a series of calibration experts were tasked with model optimization, restricting the work to selecting the proper samples to include in the computation and setting the number of factors in PLS.  The amount of time and the quality of the models were then compared.  The automated system performed the work in minutes rather than hours and the quality of the predictions at least matched the best experts and performed significantly better than the average expert.  The conclusion is that there is a large amount of recoverable giveaway that can be avoided through automation of this process and the consistency it brings to the PLS model construction.

INTRODUCTION
There is a lot of mundane work tied to the assembly of spectra and laboratory reference values to enable quality calibration work.  There is also insufficient guidance when it comes to the model construction task.  How much time should be spent on this task?  How to best assess whether a spectrum-reference pair is an outlier or not? How many cycles of regression-sample elimination make sense? Where do we switch over from improving the model by adding PLS factors to overfitting and incorporating destabilizing noise?

For more information or the full paper, contact us.

by

Rethinking Calibration for Process Spectrometers

Click on image to view the full paper.

 

 

Title:
Rethinking Calibration for Process Spectrometers

Authors:
Will Warkentin, Chevron
Brian Rohrback, Infometrix

Abstract:
Optical spectroscopy is a great source of process chemistry knowledge. It has the advantage of speed, sensitivity, and simple safety requirements. As one of very few analyzer technologies that can measure chemistry, it has become a workhorse in the hydrocarbon processing industry. What if we could put a spectroscopy system in place and have it handle the application and communicate results as soon as it is turned on? Then, if predictions do not match legacy standards, the system dials itself in or calls for help. And, we are not constrained on either the hardware or the software front. In this paper, we address the primary bottleneck of mustering data, automating analyzer calibration, and tracking data and model performance over time.

Keywords:
Best Practices, Calibration, Cloud Computing, Database, Optical Spectroscopy, PLS, Process Control