Validation of a Machine Learning-Based IDS Design Framework Using ORNL Datasets for Power System With SCADA
Bibliographic record
Abstract
Supervisory Control and Data Acquisition (SCADA) systems are widely used for remote monitoring and control of industrial processes, such as oil and gas production, power generation, transmission and distribution, and water treatment. Despite the enhanced accessibility, control, and data availability afforded by recent advances in communication technologies, the utilization of these technologies exposes critical infrastructures such as power systems to potential cyber threats. A Machine Learning (ML)-based Intrusion Detection System (IDS) seems promising; however, the development of ML models often requires custom methodologies for data preprocessing and training. This strategic approach is necessary for creating high-performance models that can be robustly evaluated and seamlessly integrated into real-time systems. As a result, we propose an ML-based IDS design framework for a SCADA-based power system incorporating effective modeling aspects, such as dataset preprocessing to ensure accurate representation, data augmentation for achieving a balanced dataset, automated feature selection to reduce dimensionality, and rigorous model training and testing procedures. To substantiate our proposed design framework, we conducted a series of experiments using a publicly available ORNL (Oak Ridge National Laboratory) dataset for . The evaluation process encompasses efficient validation techniques with unseen data. Furthermore, the augmented dataset emerged through the aggregation of readings from four Phasor Measurement Units (PMUs) collected over a specific time span into a unified dataset. Among the assessed classifiers, the Random Forest (RF) model, trained on an augmented and balanced dataset, outperformed others, yielding an F1 score of 94.09% during testing with unseen data.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".