A Process-Specific Custom AI Platform for Rapid Implementation using Raw Data Alone

CEPiML™

A process-specific custom AI platform that automatically performs training, analysis, and optimization using simple collected process data.

Technological Edge

Cloud Service ↗

Process-Specific Custom AI Platform
CEPiML™

CEPiML™ is a solution that creates process-specific AI models through machine learning by identifying effective variables within a process using raw data collected from actual plants without preprocessing.

It features automatic hyperparameter tuning via AutoML, BO, and GA, leading to the automatic derivation of optimal operation guidelines.

Simple Use, Rapid Analysis

One-Click Processing from Upload to Prediction

Seamlessly handle analysis, model generation, and forecasting just by uploading data.

Perform analyses in minutes that previously took days or weeks by eliminating the need for additional data preprocessing.

Instantly verify results across various operating conditions.

Easy Custom AI Generation

Fully Automated Workflow without Complex Configurations

Automatically selects the optimal model based on AutoML (equipped with 10+ AI algorithms).

Execution is possible immediately without manual parameter settings.

Designed with an intuitive UI, making it easy even for beginners to use.

High-Accuracy Optimal Setpoints

AI-Driven Support Beyond Prediction to Decision-Making

Provides optimal conditions and decision-making support based on operational process-based AI.

Automated search for optimal operation setpoints based on target values.

Identification of key factors through impact analysis of operational variables.

Strategic operation planning enabled through scenario-based forecasting.

CEPiML™ Custom AI Implementation Workflow

Data Collection & Automated Preprocessing

Collection of diverse process data (sensor, experimental, production data, etc.)

Handling missing values and removing outliers

Feature engineering and data cleansing

Machine Learning Model Development

AutoML-based model exploration and training

Construction of regression / classification / optimization models

Automated selection of algorithms tailored to process characteristics

Model Validation & Interpretation

Performance evaluation based on Cross-validation

SHAP / Feature Importance analysis

Model interpretation and derivation of key variable impacts

CEPiML™ can be provided as an on-premise (installed) or cloud service depending on the user environment.

CEPiML™ Service Deliverables

The deliverables shown are representative examples; final deliverables may vary depending on the specific project.

Prediction & Decision-making Tools

Optimal Conditions based on Target Variables

Generative AI Operation Guides

CEPiML™ Interface Showcase
Implementation in Utility Systems

1,575× Faster Inference
Compared to CFD
Trained on 1,284 cases using PINN + LSTM + openFOAM

Display of AI-Recommended
Optimal Setpoints
Current operating conditions vs. ML-optimized conditions: View potential savings on a single screen

Screenshots are from a system where CEPiDTS™ has been implemented.

Digital Twin Solution

An integrated process optimization solution that combines process modeling, custom AI development, and real-time monitoring to simultaneously deliver operational, financial, and ESG excellence.

A solution that mathematically models processes to provide a precise understanding of process structures and behaviors.

A service that visualizes real-time data on a dashboard, enabling instant status assessment and immediate response to process conditions.