Quality prediction and diagnosis of refined palm oil using partial correlation analysis

Abstract

Regression technique such as partial correlation analysis has been widely used as tool of prediction in business, finance and biomedical field. However, the application of predictive analysis in chemical process, specifically palm oil refinery process has rarely been done. Therefore, the objective of this paper is to present a quality prediction and diagnosis tool using partial correlation analysis, with the aim to predict the quality of refined palm oil and to diagnose the crude palm oil and process variables. Several statistical analysis are applied in data pre-process to obtain statistical sample size, optimum sampling and processing time of the process. The predictor coefficient is developed using partial correlation analysis while control chart is used to monitor the process behavior of both predicted and actual output value. The monitored out-of-control behavior is then diagnosed using SPE-contribution plot to identify the faulty input variables, thus pre-treatment can be executed before the refining process. The predicted model is successfully developed with MSE value less than 0.01 and three faulty variables are identified.

A Review of an Artificial Intelligence Framework for Identifying the Most Effective Palm Oil Prediction

“…Palm oil yield prediction is only indirectly addressed in a few ML-based studies conducted in the industry. So far, there have been five studies examining how well scientists can predict phenotype palm oil yields [56,87,141,150,151]. In light of these studies, it is difficult to determine which algorithm is best for predicting palm oil yields.…”