주성분분석과 다중선형회귀분석을 적용한 혐기성소화 시스템 평가

초록

This paper demonstrates application of the principal components analysis-multiple linear regression (PCA-MLR) technique on an anaerobic digestion system. The in-house developed software (developed using MATLAB pre-built functions and App Designer, available with MATLAB Runtime 2021b) performed statistical analysis and data-driven modelling, using a user-friendly graphical interface to avoid the programming required for complex calculations. The application of the PCA-MLR technique was demonstrated using data collected from the operation of a lab-scale anaerobic digester using food waste. By processing complex data froｍ the anaerobic digestion process, the PCA technique identified correlated parameters to select valid variables for MLR model development. As a result, a confident MLR model that predicted effluent concentrations of volatile fatty acids froｍ the change in the organic influent concentrations was constructed and its performance was also validated using statistical indices, e.g., R2 = 0.86, F0-value = 301, and p-value = 2.52 × 10-42. The PCA-MLR technique facilitated data processing and interpretation for the optimization and inspection of the anaerobic digestion system, thus supporting effective decision-making for reactor operation.

키워드