Microalgal Density and Mass Estimation Using Low-Cost Spectrometer: NIR-VIS Modeling With Evolutionary Optimization

Abstract

Estimating microalgal concentration can be a nontrivial endeavor due to their nonlinearity at high cell densities. The conventional estimation method is cell counting, which is time consuming and leads to inaccurate readings. Alternatively, spectral reflectance provides a more precise measurement by using specific wavelengths that correspond directly to pigment absorption in microalgae, allowing for faster determination of cell density and biomass. Unfortunately, the experiment is usually conducted in the laboratory with expensive and high-resolution devices. In this letter, we build a low-cost, real-time Internet-of-Things-based spectral prototype sensor for estimating density and mass of microalgae. The device uses wavelengths in the range of 400–1000 nm, making it low resolution. Multi expression programming is employed to model the measured data. Results show that nonlinear models perform better with $R^{2}$ values spanning from 0.93 to 0.99 for two species of microalgae.