WINDWISE · Scalable and Private Multi-Level Fleet-Wide Monitoring for Distributed Wind Energy Assets

Линиите свързват координатора с партньорите.

Накратко на български

Цел на проекта

WINDWISE targets scalable and private wind-fleet condition monitoring by uniting intra-farm centralized federated learning (CFL) with an inter-farm decentralized federated learning (DFL) layer in a hierarchical CFL-DFL (HFL) architecture. Current practice is limited by fragmented data ownership, uneven computing and bandwidth, and privacy risks that hinder cross-operator/company collaboration. WINDWISE addresses these bottlenecks by keeping raw operational data on wind farms while enabling collaborative model updates within and across farms. The project will (i) formalize and evaluate CFL for intra-farm coordination under realistic constraints, (ii) develop system-aware client (turbine) selection to manage straggler clients that lag in computation or communication and improve learning efficiency, and (iii) establish a ring-based DFL layer for cross-silo collaboration without central coordination. Methods will be validated through controlled benchmarks against siloed and CFL-only baselines, using open and partner datasets with virtualized clients to test bandwidth, latency, and dropout conditions. Expected outcomes include privacy-compliant diagnostics with fewer false alarms, higher availability, and lower operations and maintenance (O&M) costs, plus practical policies for scheduling and aggregation and an open toolkit to support adoption. This work directly supports wind-asset management by improving reliability and availability, enhancing condition awareness, and safeguarding data sovereignty and privacy.

Оригинален текст от CORDIS (на английски).

Участници

• TAMPEREEN KORKEAKOULUSAATIO SR · TampereКоординаторФинландия
• BERNER FACHHOCHSCHULE · BernШвейцария

Връзки

• Виж в CORDIS
• DOI: 10.3030/101276753