Quantum-Enhanced Biomanufacturing Anomaly Dataset
Description
The Quantum Biomanufacturing Process Anomaly Dataset (QB-PAD) is a multivariate time-series dataset developed to evaluate anomaly detection models in complex industrial monitoring scenarios. This dataset is derived from the publicly available NASA SMAP/MSL telemetry dataset and is repurposed for research experiments in bioprocess anomaly detection. While the telemetry signals exhibit structural properties similar to industrial multi-sensor systems—such as temporal dependencies, nonlinear correlations, noise, and intermittent anomalies—they are not validated bioreactor measurements. The dataset is used strictly for methodological evaluation of machine learning and quantum machine learning approaches. Purpose Real-world biomanufacturing data is often proprietary and difficult to access. To enable reproducible research, this dataset serves as a controlled experimental benchmark for: Quantum kernel-based classification (QSVM) Deep learning–based reconstruction models (LSTM Autoencoder) Classical machine learning baselines (RBF SVM) Comparative anomaly detection studies The focus is on evaluating model behavior under multivariate temporal conditions rather than validating domain-specific biological phenomena.