High-Fidelity Synthetic Data Generation Framework for Industrial 3-Phase Induction Motors

Synthetic data enhances automated fault detection in industrial motor diagnostics by simulating diverse electrical and mechanical failure scenarios and compensating for limited real-world measurements. This paper proposes a method to generate synthetic vibration and current signals for three-phase induction motors using statistical modeling, clustering, and probability density functions. The generated dataset includes healthy motor operation and five different fault conditions with noise injection, phase jittering, and transient anomalies to enhance the realism of the data. The results demonstrate that synthetic data can serve as a reliable foundation for AI-based motor fault analysis, providing a scalable alternative to costly real-world data collection.