An intelligent SCADA-integrated deep learning framework for bird-safe offshore wind farm operation.

Area of Science:

• Environmental Science
• Artificial Intelligence
• Renewable Energy Engineering

Background:

• Offshore wind energy expansion raises concerns about avian mortality from turbine collisions.
• Current mitigation methods (radar, manual intervention, deterrents) suffer from inaccuracies and delays.
• Lack of species-level identification hinders effective, targeted mitigation strategies.

Purpose of the Study:

• To develop an intelligent framework for automated avian detection and classification to mitigate collision risks.
• To integrate a Deep Convolutional Neural Network (DCNN)-based Bird Detection and Classification (BDC) model with Supervisory Control and Data Acquisition (SCADA) systems.
• To enable real-time, SCADA-driven turbine control actions based on bird proximity and species identification.

Main Methods:

• Developed a DCNN-based BDC model trained on over 90,000 images across 525 avian species.
• Integrated the BDC model with a SCADA system using a multi-zone proximity assessment strategy.
• Evaluated performance through comparative analysis with traditional classifiers and simulation of system latency.

Main Results:

• The proposed BDC model achieved high accuracy (99.62%), precision (99.92%), recall (100%), and F1-score (99.93%).
• System demonstrated low inference latency (<30 ms) and SCADA response execution (<40 ms).
• Simulation results indicate a robust proof-of-concept for AI-driven avian collision mitigation.

Conclusions:

• The AI-integrated framework shows significant potential for automated, risk-aware turbine control to protect avian wildlife.
• This approach offers a novel solution for environmentally sustainable offshore wind farm operations.
• Further validation in real-world offshore conditions is necessary to confirm deployment robustness.