TEP: Fault Type Descriptions for Industrial Process Diagnosis

93.6% multiclass diagnosis accuracy was achieved by a proposed model using this benchmark data. The dataset describes fault types for the Tennessee Eastman Process, a standard benchmark for industrial fault detection and diagnosis. It was authored by Babar Hayat and last updated on June 2, 2026.

Use Cases

• Benchmarking fault diagnosis models based on the described Tennessee Eastman Process framework.
• Training sensor-aware attention mechanisms based on the described process centrality and sensor reliability concepts.
• Evaluating model calibration metrics like Brier score and expected calibration error based on the described performance results.

Strengths

• Model performance on the data is quantified with specific metrics: 93.6% multiclass diagnosis accuracy and a 94.0% F1 score.
• Calibration improvements are documented with specific reductions in negative log-likelihood (0.197 to 0.182) and Brier score (0.101 to 0.095).
• The data is associated with a well-known industrial process benchmark, the Tennessee Eastman Process.

Limitations

• The dataset is very small at 5.5 KB, suggesting limited scope or a high-level description rather than raw process data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for model training.

Provenance

Source

figshare

Collection Method

Likely derived from or describes the standard Tennessee Eastman Process benchmark.

Freshness

Last updated 2026-06-02 17:36:39; freshness should be verified.