https://www.mdu.se/

Anomaly detection plays a crucial role in industrial settings, particularly in maintaining the reliability and optimal performance. Traditional anomaly detection methods typically rely on the use of single models with labelled data and often face challenges in handling diverse data characteristics and variations in noise levels, resulting in limited effectiveness. Self-supervised learning (SSL) allows models to learn from unlabeled data by creating their own supervisory signals through tasks like reconstruction (as in autoencoders), making it a powerful technique for tasks of anomaly detection. This work proposes a novel approach named as Self-Supervised Learning Framework with Dual Ensemble Voting Fusion for Maximizing Anomaly Prediction in Timeseries (S2DEVFMAP). Wherein the use of five heterogeneous independent models combined with a dual ensemble fusion of voting techniques is demonstrated. Diverse models capture various system behaviors, while the fusion strategy maximizes detection effectiveness and minimizes false alarms. Each base autoencoder model learns a unique representation of the data, leveraging their complementary strengths to improve anomaly detection. The use of dual ensemble technique is proven to maximize the identification of anomalies. Experimentation is done on a real-world dataset of an industrial cooling system in a power station to demonstrate the effectiveness of the proposed approach.