Prototype-based Heterogeneous Federated Learning for Blade Icing Detection in Wind Turbines with Class Imbalanced Data

TL;DR

This paper tackles wind-turbine blade icing detection under data privacy and non-IID heterogeneity. It introduces FedHPb, a federated prototype learning framework that communicates class prototypes and uses a supervised contrastive loss to address class imbalance, enabling robust learning across diverse wind-farm environments. The approach combines a LCNN-based client extractor, per-class prototypes, and global prototype aggregation, achieving substantial improvements over both FL and class-imbalance baselines on real-world data, along with favorable communication efficiency. Ablation and sensitivity analyses confirm the critical roles of the contrastive loss, local-module choices, and window size, suggesting significant practical impact for privacy-preserving, accurate icing detection in distributed wind-energy systems.

Abstract

Wind farms, typically in high-latitude regions, face a high risk of blade icing. Traditional centralized training methods raise serious privacy concerns. To enhance data privacy in detecting wind turbine blade icing, traditional federated learning (FL) is employed. However, data heterogeneity, resulting from collections across wind farms in varying environmental conditions, impacts the model's optimization capabilities. Moreover, imbalances in wind turbine data lead to models that tend to favor recognizing majority classes, thus neglecting critical icing anomalies. To tackle these challenges, we propose a federated prototype learning model for class-imbalanced data in heterogeneous environments to detect wind turbine blade icing. We also propose a contrastive supervised loss function to address the class imbalance problem. Experiments on real data from 20 turbines across two wind farms show our method outperforms five FL models and five class imbalance methods, with an average improvement of 19.64\% in \( mF_β \) and 5.73\% in \( m \)BA compared to the second-best method, BiFL.

Figures (6)

• Figure 1: Overview of FedHPb: In the initial stage, clients update their local prototype sets by reducing classification errors and minimizing the differences between local and global prototypes through supervised contrastive learning. Once the updates are completed, these prototype sets are sent to the central server. The central server creates a global prototype based on the received data and sends this global prototype back to all clients.
• Figure 2: Neural network architecture of the client model.
• Figure 3: Approximate Model Update Size (MB)
• Figure 4: SGD variation curve.
• Figure 5: ADAM variation curve.