Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Feature Wavebands Selection Using Immune Genetic Algorithm for Prediction of Soluble Solids Content in Citrus Fruit

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2012 Dallas, Texas, July 29 - August 1, 2012  121340825.(doi:10.13031/2013.42202)
Authors:   Qibing Zhu, Xin Zhao, Min Huang
Keywords:   Near-infrared Spectra; Waveband Selection; Immune Genetic Algorithm; Partial Least Squares Regression

FT-NIR spectroscopy is a promising technique for rapid and noninvasive measurement of multiple quality attributes of citrus fruit. But conventional methods (such as multi-linear regression or MLR) for feature wavebands selection are time-consuming and may not be optimal. An immune genetic algorithm (IGA) approach was proposed to select the feature wavebands for the FT-NIR spectroscopy data of Gong Chuan citrus as a precursor to the development of calibration models for predicting fruit soluble solids content (SSC). All wavebands were average divided into 25 intervals and 11 optimal intervals including 662 wavebands were achieved using IGA. Partial least squares (PLS) regression and cross-validation methods were used to develop models predicting SSC coupled with 662 feature wavebands. Compared with standard genetic algorithm (SGA), IGA yielded the optimal regions with faster convergence and preventing premature. The prediction results of IGA model (RP = 0.915 and RMSEP = 0.71 %) was better than full wavebands model (RP = 0.886 and RMSEP = 0.81 %). IGA approach would provide an effective means for selecting feature wavebands to predict SSC in citrus fruits, it is possible to optimize data selection, improve the precision of prediction and reduce the number of variables of calibration.

(Download PDF)    (Export to EndNotes)