Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advancing Machine Learning in Industry 4.0: Benchmark Framework for Rare-event Prediction in Chemical Processes

Vikram Sudarshan, Warren D. Seider

TL;DR

A novel and comprehensive benchmark framework for rare-event prediction is introduced, comparing ML algorithms of varying complexity, including Linear Support-Vector Regressor and k-Nearest Neighbors, to more sophisticated algorithms, such as Random Forests, XGBoost, LightGBM, CatBoost, Dense Neural Networks, and TabNet.

Abstract

Previously, using forward-flux sampling (FFS) and machine learning (ML), we developed multivariate alarm systems to counter rare un-postulated abnormal events. Our alarm systems utilized ML-based predictive models to quantify committer probabilities as functions of key process variables (e.g., temperature, concentrations, and the like), with these data obtained in FFS simulations. Herein, we introduce a novel and comprehensive benchmark framework for rare-event prediction, comparing ML algorithms of varying complexity, including Linear Support-Vector Regressor and k-Nearest Neighbors, to more sophisticated algorithms, such as Random Forests, XGBoost, LightGBM, CatBoost, Dense Neural Networks, and TabNet. This evaluation uses comprehensive performance metrics, such as: $\textit{RMSE}$, model training, testing, hyperparameter tuning and deployment times, and number and efficiency of alarms. These balance model accuracy, computational efficiency, and alarm-system efficiency, identifying optimal ML strategies for predicting abnormal rare events, enabling operators to obtain safer and more reliable plant operations.

Advancing Machine Learning in Industry 4.0: Benchmark Framework for Rare-event Prediction in Chemical Processes

TL;DR

A novel and comprehensive benchmark framework for rare-event prediction is introduced, comparing ML algorithms of varying complexity, including Linear Support-Vector Regressor and k-Nearest Neighbors, to more sophisticated algorithms, such as Random Forests, XGBoost, LightGBM, CatBoost, Dense Neural Networks, and TabNet.

Abstract

Previously, using forward-flux sampling (FFS) and machine learning (ML), we developed multivariate alarm systems to counter rare un-postulated abnormal events. Our alarm systems utilized ML-based predictive models to quantify committer probabilities as functions of key process variables (e.g., temperature, concentrations, and the like), with these data obtained in FFS simulations. Herein, we introduce a novel and comprehensive benchmark framework for rare-event prediction, comparing ML algorithms of varying complexity, including Linear Support-Vector Regressor and k-Nearest Neighbors, to more sophisticated algorithms, such as Random Forests, XGBoost, LightGBM, CatBoost, Dense Neural Networks, and TabNet. This evaluation uses comprehensive performance metrics, such as: , model training, testing, hyperparameter tuning and deployment times, and number and efficiency of alarms. These balance model accuracy, computational efficiency, and alarm-system efficiency, identifying optimal ML strategies for predicting abnormal rare events, enabling operators to obtain safer and more reliable plant operations.
Paper Structure (23 sections, 32 equations, 12 figures, 6 tables)

This paper contains 23 sections, 32 equations, 12 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Progress of Industrial Revolution and Challenge of Rare Un-postulated Abnormal Events
  3. Artificial Intelligence and Machine Learning (AI/ML) for Quantitative Analyses of Rare Events
  4. Benchmark Analyses of ML Algorithms
  5. Prior Research: Developing Improved, Multivariate Alarm Systems Using Forward-Flux Sampling and Machine Learning
  6. Benchmark Analyses of ML Algorithms for Rare Abnormal Events
  7. Materials and Methods
  8. Overview of Key Steps
  9. Step 1: Data Generation via Forward-flux Sampling
  10. Step 2: Data Preprocessing
  11. Step 3: Predictive Modeling via Machine Learning
  12. Step 4: Reporting Key Evaluation Metrics
  13. Step 5: Determine Optimal ML Model
  14. Results and Discussions
  15. Dataset I (PI-Controlled Exothermic CSTR)
  16. ...and 8 more sections

Figures (12)

  • Figure 1: Overview of Key Steps
  • Figure 2: Schematic showing key steps for simulating abnormal trajectories using BG-FFS algorithm (refer to the points in Section \ref{['sec_2.2']})
  • Figure 3: Schematic implementing key steps in RF for regression tasks, showing the: (a) Training phase; (b) Testing phase.
  • Figure 4: (a) Level-wise growth strategy followed by XGBoost; (b) Leaf-wise growth strategy followed by LightGBM
  • Figure 5: For Dataset 1, (a) Heatmap showing the scaled evaluation metrics; (b) Cost computed for all ML models
  • ...and 7 more figures