Identification of ichthyotoxic red tide algae based on three-dimensional fluorescence spectra and particle swarm optimization support vector machine

Abstract

Acccurate identification whether red tide has ithyotoxicity is very significant for microalgae monitoring. In order to realize the rapid and non-destructive detection of ichthyotoxic red tide algae, a detection method combining three-dimensional (3D) fluorescence spectrum and particle swarm optimization support vector machine (PSO-SVM) was developed to monitor the ichthyotoxic red tide algae with cell concentrations from 104 cells/mL to 106 cells/mL. The contour maps contracted form three-dimensional fluorescence spectra of six common species of ichthyotoxic algae and eight common species of non-ichthyotoxic algae,which are analyzed to select the optimal emission and excitation wavelength span. The new feature data are acquired by using the emission spectrum data at 480 nm and 510 nm excitation wavelengths. The new feature data are used as the input of particle swarm optimization support vector machine to establish the optimal classification model of ichthyotoxic algae, which achieves an classification accuracy of 100% for the test set. The optimal classification model is successfully applied to identify the ichthyotoxicity of different algae including Heterosigma akashiwo, Chattonella marina, Phaeocystis globosa, Prorocentrum donghaiense, Karenia dunnii, Isoscelina galbana, Isosceles globosa and Skeletonema costatum.