A Cost-Effective AIoT System for Real-Time Oil Content Monitoring in Palm Oil Extraction

Applied Computational Intelligence and Soft Computing

卷:2026

Research Article

Atitaya Kham-Ouam¹  Patchanee Laddawong²  Yutthapong Pianroj^3,4  Teerasak Punvichai^4,5  Saysunee Jumrat^3,4  Jirapond Muangprathub^3,4 

[1] Faculty of Humanities and Social Sciences, Prince of Songkla University, Pattani Campus, Pattani, Thailand, psu.ac.th

[2] Faculty of Science and Industrial Technology, Prince of Songkla University, Surat Thani Campus, Surat Thani, Thailand, psu.ac.th

[3] Integrated High-Value Oleochemical Research Center, Prince of Songkla University, Surat Thani Campus, Surat Thani, Thailand, psu.ac.th

[4] College of Digital Science, Prince of Songkla University, Hat Yai Campus, Songkhla, Thailand, psu.ac.th

[5] Faculty of Innovative Agriculture, Fisheries and Food, Prince of Songkla University, Surat Thani Campus, Surat Thani, Thailand, psu.ac.th

关键词: AIoT;    machine learning;    oil content monitoring;    oil palm industry;    reduction of oil;   

DOI  :  10.1155/acis/3785817

 received in 2026-01-22, accepted in 2026-06-18,  发布年份 2026

来源: Hindawi Publishing Corporation

【 摘 要 】

In the palm oil industry, accurate monitoring of residual oil in palm fiber is essential for optimizing extraction efficiency and minimizing production losses. Conventional oil content inspection methods are labor-intensive, time-consuming, and costly, limiting their applicability for real-time process control and smallholder operations. To address these challenges, this study proposes a low-cost Artificial Intelligence of Things (AIoT) system for real-time monitoring and prediction of oil loss in palm fiber. The proposed system integrates moisture, electrical resistance, and capacitance sensors with a NodeMCU ESP32 microcontroller to enable continuous on-site data acquisition. Sensor data are transmitted to a cloud-based platform for analysis and predictive modeling. Multiple machine learning algorithms, including Linear Regression, Support Vector Machine, K-Nearest Neighbors, Neural Networks, and XGBoost, are evaluated for oil loss prediction. Experimental results based on industrial palm oil mill data demonstrate that the XGBoost model achieves the highest predictive performance, with an R2 value of 0.997, substantially outperforming other models. The proposed AIoT system provides actionable decision support for oil extraction process adjustment, offering a scalable and cost-effective solution to reduce oil loss, improve operational efficiency, and enhance sustainability in the palm oil industry.

【 授权许可】

CC BY   
Copyright © 2026 Atitaya Kham-Ouam et al. Applied Computational Intelligence and Soft Computing published by John Wiley & Sons Ltd. 2026

【 图 表 】

Figure 6

Figure 4

Figure 2

Figure 4

Figure 11

Figure 3

Figure 8