Statistical Significance Testing as a Guide to Partial Least-Squares (PLS) Modeling of Nonideal Data Sets for Fuel Property Predictions

Kramer, Kirsten; Morris, Robert E.; Rose-Pehrsson, Susan L.; Cramer, Jeffrey A.; Johnson, Kevin

doi:10.1021/ef700403s

Cited by 19 publications

(17 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this case, being inconsistently informative is defined as having a regression coefficient average/regression coefficient standard deviation ratio lower than that obtained for 85% of the random variables. The value of 85% was used to maintain consistency with the F-test’s pre-existing statistical parameter, which itself was finalized in a previous work . It should be noted that the number of LVs to be used for these UVE-PLS models are, as in the case of standard PLS, calculated from CUMPRESS results using the statistical F-test, as described previously.…”

Section: Methodsmentioning

confidence: 99%

“…The value of 85% was used to maintain consistency with the F-test's pre-existing statistical parameter, which itself was finalized in a previous work. 50 It should be noted that the number of LVs to be used for these UVE-PLS models are, as in the case of standard PLS, calculated from CUMPRESS results using the statistical F-test, as described previously.…”

Section: Energy and Fuelsmentioning

confidence: 99%

See 1 more Smart Citation

Novel Data Abstraction Strategy Utilizing Gas Chromatography–Mass Spectrometry Data for Fuel Property Modeling

et al. 2014

Self Cite

View full text Add to dashboard Cite

The high cost and limited availability of emerging alternative fuels and/or fuel blending stocks with unknown compositions are often major impediments to the certification of these materials as Fit-For-Purpose (FFP) for the U.S. Navy. A method was desired whereby a candidate fuel could be rapidly prescreened to determine if it would be suitable for further, moreextensive FFP testing. The goal of this research was to employ statistical analysis strategies to establish linkages between the chemical constituency of any given fuel or fuel stock, regardless of type or source, and the resultant performance, and/or fuel properties. A chemical profiler developed during the course of this work has previously been used to quantify the constituencies of fuels using gas chromatography−mass spectrometry (GC-MS) data. These constituencies were then correlated to specification properties using partial least-squares regression modeling reconfigured into a multistep, iterative strategy. While this modeling strategy was shown to be successful at predicting the performance properties not only of the training data but also of uncalibrated alternative fuels, the underlying data abstraction strategy was determined to be inherently unsuitable for use with the disparate data from multiple GC-MS instruments due to instrument-based overfitting. The following report details a novel modeling strategy that makes use of normalized total ion chromatography (TIC) peak areas to both streamline the procedural complexity of the previous modeling strategy and more ably quantify chemical constituencies for the purposes of multi-instrument FFP fuel modeling.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Energy and Fuelsmentioning

confidence: 99%

Novel Data Abstraction Strategy Utilizing Gas Chromatography–Mass Spectrometry Data for Fuel Property Modeling

et al. 2014

Self Cite

View full text Add to dashboard Cite

show abstract

“…A thorough analysis of this type of model bias, known as overfitting, can be found in previous work. 20 The present work does not directly perform any calibration transfer operations, as instrument-specific models are constructed for all instruments.…”

Section: ■ Experimental Sectionmentioning

confidence: 99%

“…The F-test, by limiting the number of LVs, protects against models that are too biased toward a specific set of calibration data, which would make the models themselves improperly biased against and less effective when predicting the fuel properties and alternative fuel contents of uncalibrated data. A thorough analysis of this type of model bias, known as overfitting, can be found in previous work . The present work does not directly perform any calibration transfer operations, as instrument-specific models are constructed for all instruments.…”

Section: Experimental Sectionmentioning

confidence: 99%

Expanded Framework for the Prediction of Alternative Fuel Content and Alternative Fuel Blend Performance Properties Using Near-Infrared Spectroscopic Data

et al. 2015

Self Cite

View full text Add to dashboard Cite

PCA score space representing the NFPM jet fuel calibration data set. Rays have been included to approximately indicate the manner in which various alternative fuel types deviate from the petroleum fuel cloud. 197x147mm (96 x 96 DPI) . PCA score space representing the NFPM diesel fuel calibration data set. Rays have been included to approximately indicate the manner in which various alternative fuel types deviate from the petroleum fuel cloud. 196x150mm (96 x 96 DPI) ABSTRACTPartial least squares (PLS) regression models can be constructed from near-infrared (NIR) spectroscopic data to identify and predict critical specification properties of jet and diesel fuels for quality surveillance pre-screening. This same approach has also been used previously to identify Fischer-Tropsch synthetic fuels and fatty acid methyl ester fuels, predict their quantities in blends with jet and diesel petrochemical fuels, and even correct fuel property predictions when alternative fuel contents in blends would affect the predictions of the properties in question. The present work expands upon these previous results by incorporating several additional alternative fuel types into a more generalized alternative fuel content and property modeling framework than was developed previously. The framework consists of a single generalized PLS modeling solution to simultaneously accommodate multiple alternative fuels considered isoparaffinic in nature, as well as smaller-scale modeling solutions to accommodate individual alternative fuels that are not similarly isoparaffinic in nature. This expanded framework provides the means to allow NIR PLS models to predict and quantify alternative fuel contents in blends, and accurately predict affected fuel properties, in a robust fashion that, due to the use of more generalized modeling than has been seen in previous work, better accommodates a future of unknown and unknowable alternative fuel types.

show abstract

“…Because of the correlation of IR overtone bands with the chemical and physical properties of fuels, near-IR (NIR) spectroscopy has gained widespread acceptance as a secondary analytical method in fuel analysis [1][2][3][4][5][6][7][8][9][10][11][12][13][14]. Typically, using the partial least squares (PLS) method, NIR spectra of fuels are correlated with their known American Society for Testing and Materials (ASTM)-determined properties.…”

Section: Introductionmentioning

confidence: 99%

Calibration transfer of near‐IR partial least squares property models of fuels using virtual standards

Cooper

Larkin

Abdelkader

2011

Journal of Chemometrics

View full text Add to dashboard Cite

a Partial least squares (PLS) models of 10 important jet and diesel fuel properties were built using spectra from a master near-IR dispersive instrument and then subsequently transferred to a secondary dispersive instrument via a novel calibration transfer method using virtual standards and a slope-bias correction. Implementation of the transfer requires that only seven spectra of neat solvents be acquired on the master and secondary instruments. The spectra of the neat solvents are then used to digitally replicate spectra from the calibration set to generate virtual standards. Comparison of PLS predictions for the master and secondary instrument virtual standards provides a simple but effective slope-bias correction for transfer. The transferred fuel properties include American Petroleum Institute gravity, % aromatics, cetane index, flashpoint, hydrogen content, % saturates, and distillation temperatures at 10%, 20%, 50%, and 90% volume recovered. Transfer error was lower than using either the pure solvents with a slope-bias correction or than using a piecewise direct standardization calibration transfer using fuel spectra. Transfer error was higher than when using actual fuels to transfer the calibration. The use of virtual standards eliminates the need to maintain either complex fuel standards or the master instrument for future instrument calibration transfers.

show abstract

Statistical Significance Testing as a Guide to Partial Least-Squares (PLS) Modeling of Nonideal Data Sets for Fuel Property Predictions

Cited by 19 publications

References 29 publications

Novel Data Abstraction Strategy Utilizing Gas Chromatography–Mass Spectrometry Data for Fuel Property Modeling

Novel Data Abstraction Strategy Utilizing Gas Chromatography–Mass Spectrometry Data for Fuel Property Modeling

Expanded Framework for the Prediction of Alternative Fuel Content and Alternative Fuel Blend Performance Properties Using Near-Infrared Spectroscopic Data

Calibration transfer of near‐IR partial least squares property models of fuels using virtual standards

Contact Info

Product

Resources

About