Unsupervised anomaly detection on 3-year multivariate sensor data from a cyclone preheater using Python. Detected 437 abnormal timestamps using Isolation Forest, One-Class SVM, and Z-score filtering. Includes visual timeline, anomaly impact summary, and Excel highlights.
visualization python scikit-learn manufacturing matplotlib unsupervised-learning data-cleaning anomaly-detection process-monitoring fault-detection multivariate-time-series scikit-learn-pipelines industrial-ml cyclone-preheater
-
Updated
Aug 1, 2025 - Jupyter Notebook