Lactic Fermentation Spectral Analysis of Target Substrates and Food and Feed Wastes for Energy Applications

The article deals with the creation of a calibration model of lactic acid content in an aqueous solution. The research concept included the preparation of a control tool for the process of modifying the properties of the food fraction for methane fermentation bacteria. The thesis was formulated that it is possible to prepare a systemic solution for real-time observation and monitoring of lactic acid secretion during the digestion of a hydrated mixture of food fractions. The scientific aim of the work was to develop and verify a calibration model of lactic acid content in an aqueous mixture with limited transparency for visible light waves. The research methodology was based on near-infrared spectroscopy with multivariate analysis. Stochastic modeling with noise reduction based on orthogonal decomposition was used. A calibration model was created using Gaussian processes (GP) to predict the lactic acid concentration in an aqueous solution or mixture using an NIRVis spectrophotometer. The design of the calibration model was based on absorbance spectra and computational data from selected wavelength ranges from 450 nm to 1900 nm.
The measurement data in the form of spectra were limited from the initial wider range (400–2250 nm) to reduce interference. The generated calibration model achieved a mean error level not exceeding 2.47 g·dm−3 of the identified lactic acid fraction. The coefficient of determination R2 was 0.996. The effect of absorbing the emitter waves was achieved despite the limited transparency of the mixture.