Process monitoring and fault detection on a hot-melt extrusion process using in-line Raman spectroscopy and a hybrid soft sensor

Furqan Tahir, Muhammad T. Islam, John Mack, John Robertson, David Lovett

Research output: Contribution to journalArticle

Abstract

We propose a real-time process monitoring and fault detection scheme for a pharmaceutical hot-melt extrusion process producing Paracetamol-Affinisol extrudate. The scheme involves prediction of Paracetamol concentration from two independent sources: a hybrid soft sensor and a Raman-based Partial Least Squares (PLS) calibration model. Both these predictions are used by the developed PCA (Principal Component Analysis) and SPC (Statistical Process Control) monitors to detect process faults and raise alarms. Through real-time extrusion results, it is shown that this two-sensor approach enables the detection of various common process faults which would otherwise remain undetected with a single-sensor monitoring scheme.
LanguageEnglish
Pages400-414
Number of pages15
JournalComputers and Chemical Engineering
Volume125
Early online date25 Mar 2019
DOIs
Publication statusPublished - 9 Jun 2019

Fingerprint

Process monitoring
Fault detection
Extrusion
Raman spectroscopy
Acetaminophen
Sensors
Statistical process control
Principal component analysis
Drug products
Calibration
Monitoring
Pharmaceutical Preparations

Keywords

  • process monitoring
  • fault detection
  • hot melt extrusion
  • PCA
  • PLS
  • calibration model
  • hybrid soft sensor
  • Affinisol 15LV
  • HPMC

Cite this

@article{31303d1363d54ae296031f245f9691c1,
title = "Process monitoring and fault detection on a hot-melt extrusion process using in-line Raman spectroscopy and a hybrid soft sensor",
abstract = "We propose a real-time process monitoring and fault detection scheme for a pharmaceutical hot-melt extrusion process producing Paracetamol-Affinisol extrudate. The scheme involves prediction of Paracetamol concentration from two independent sources: a hybrid soft sensor and a Raman-based Partial Least Squares (PLS) calibration model. Both these predictions are used by the developed PCA (Principal Component Analysis) and SPC (Statistical Process Control) monitors to detect process faults and raise alarms. Through real-time extrusion results, it is shown that this two-sensor approach enables the detection of various common process faults which would otherwise remain undetected with a single-sensor monitoring scheme.",
keywords = "process monitoring, fault detection, hot melt extrusion, PCA, PLS, calibration model, hybrid soft sensor, Affinisol 15LV, HPMC",
author = "Furqan Tahir and Islam, {Muhammad T.} and John Mack and John Robertson and David Lovett",
year = "2019",
month = "6",
day = "9",
doi = "10.1016/j.compchemeng.2019.03.019",
language = "English",
volume = "125",
pages = "400--414",
journal = "Computers and Chemical Engineering",
issn = "0098-1354",

}

Process monitoring and fault detection on a hot-melt extrusion process using in-line Raman spectroscopy and a hybrid soft sensor. / Tahir, Furqan ; Islam, Muhammad T.; Mack, John; Robertson, John; Lovett, David .

In: Computers and Chemical Engineering, Vol. 125, 09.06.2019, p. 400-414.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Process monitoring and fault detection on a hot-melt extrusion process using in-line Raman spectroscopy and a hybrid soft sensor

AU - Tahir, Furqan

AU - Islam, Muhammad T.

AU - Mack, John

AU - Robertson, John

AU - Lovett, David

PY - 2019/6/9

Y1 - 2019/6/9

N2 - We propose a real-time process monitoring and fault detection scheme for a pharmaceutical hot-melt extrusion process producing Paracetamol-Affinisol extrudate. The scheme involves prediction of Paracetamol concentration from two independent sources: a hybrid soft sensor and a Raman-based Partial Least Squares (PLS) calibration model. Both these predictions are used by the developed PCA (Principal Component Analysis) and SPC (Statistical Process Control) monitors to detect process faults and raise alarms. Through real-time extrusion results, it is shown that this two-sensor approach enables the detection of various common process faults which would otherwise remain undetected with a single-sensor monitoring scheme.

AB - We propose a real-time process monitoring and fault detection scheme for a pharmaceutical hot-melt extrusion process producing Paracetamol-Affinisol extrudate. The scheme involves prediction of Paracetamol concentration from two independent sources: a hybrid soft sensor and a Raman-based Partial Least Squares (PLS) calibration model. Both these predictions are used by the developed PCA (Principal Component Analysis) and SPC (Statistical Process Control) monitors to detect process faults and raise alarms. Through real-time extrusion results, it is shown that this two-sensor approach enables the detection of various common process faults which would otherwise remain undetected with a single-sensor monitoring scheme.

KW - process monitoring

KW - fault detection

KW - hot melt extrusion

KW - PCA

KW - PLS

KW - calibration model

KW - hybrid soft sensor

KW - Affinisol 15LV

KW - HPMC

UR - https://www.sciencedirect.com/journal/computers-and-chemical-engineering

U2 - 10.1016/j.compchemeng.2019.03.019

DO - 10.1016/j.compchemeng.2019.03.019

M3 - Article

VL - 125

SP - 400

EP - 414

JO - Computers and Chemical Engineering

T2 - Computers and Chemical Engineering

JF - Computers and Chemical Engineering

SN - 0098-1354

ER -