Predictive moisture modelling and closed-loop process recommendations integrated with SCADA and PLC, combining process data, predictive models, and controlled PLC/SCADA workflows to maximize edible oil extraction.
FMCG / Edible Oil Processing
Predictive Machine Learning
Process Optimization & SCADA Control
Python ML, Edge AI IPC, SCADA/PLC APIs, Historian DB
In edible oil manufacturing, copra (dried coconut kernel) is conditioned in a cooker before expelling. Achieving the exact, precise moisture level during conditioning is absolutely critical for maximizing plant productivity and oil yield.
Historically, this process was manually controlled through SCADA by operators adjusting steam pressure based on intuition and periodic manual sampling. This resulted in inconsistent batches, suboptimal oil extraction, and wasted thermal energy.
Operators needed repeatable, data-driven recommendations that could be monitored and validated in real-time against output quality, eventually closing the loop for automated PLC control.
A structured process-data workflow combining prediction, optimization, control integration, and continuous monitoring.
The Edge AI system seamlessly ingests and preprocesses massive streams of real-time process data directly from the plant's PLC/SCADA architecture and various field instruments, including temperature and humidity sensors.
Supervised regression models dynamically analyze inputs including raw copra moisture, particle size, bed height, relative humidity, load amps, and expeller HF speed to accurately predict the cooked-copra output moisture in real-time.
The AI software integrates directly with existing PLCs through robust APIs and middleware. It sends automated, micro-adjusted steam pressure and feed-rate recommendations to maintain optimal moisture levels continuously.
Verified results, predictions, and automated actions are stored in a Historian database and synchronized to real-time dashboards, giving shift managers complete visibility over the optimization loop.
A modular industrial-AI architecture connects plant data, predictive models, control integration and monitoring.
| Process Inputs | SCADA and PLC programme server |
|---|---|
| Data Platform | AI industrial PC |
| ML Model | Python ML and optimization services |
| Control | API/middleware for PLC communication |
| Monitoring | Historian/process database and monitoring application |
Transforming intuition-based manual operations into a highly optimized, data-driven science.
Supports radically more consistent copra conditioning, applying repeatable, optimal thermal criteria across all products and shifts.
Reduces heavy dependence on operator-only intuition and manual parameters, substantially strengthening overall operational control.
Creates a rigorous, data-driven basis for continuous process improvement, measurability, and maximum oil extraction yield.
Dramatically improves visibility of complex inputs and creates structured, searchable historian records for deep reporting and audit.
