Wind Turbine Control Cyberattack Detection With Model Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in effectively detecting and isolating cyberattacks on wind turbine control systems due to their complexity and stochastic operation conditions, making it difficult to develop a reliable detection model that can cover a wide range of operational scenarios.
Innovation Solution
A method and system utilizing an ensemble of classification models, including machine learning algorithms like decision stumps and convolutional neural networks, to receive and analyze time series data from monitoring nodes, extract features, and determine the probability of a cyberattack through a decision fusion module, implementing control actions when the probability exceeds a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a single detection model is used for the entire turbine unit, then the model covers the whole system, but the model becomes overwhelmingly complex and difficult to train
Solution Approach 1:
The patent divides the wind turbine control system into multiple subsystems (pitch control, torque control, yaw control, generator control) and develops separate detection models for each subsystem. This segmentation reduces the complexity of individual models while maintaining comprehensive coverage across the entire turbine unit, as each model focuses on specific subsystem characteristics and operational parameters.
2Reliability
If a complex model is trained to cover wide range of stochastic operation conditions, then detection accuracy improves, but training becomes practically impossible
Solution Approach 1:
The patent segments the training process into subsystem-specific training tasks, where each detection model is trained on data relevant to its specific subsystem and operational conditions. This approach makes training practically feasible by breaking down the overwhelming task of training a single comprehensive model into multiple manageable training tasks, while still achieving high detection accuracy across diverse stochastic operation conditions through the collective expertise of multiple specialized models.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for detecting a cyberattack on a control system of a wind turbine includes providing a plurality of classification models of the control system. The method also includes receiving, via each of the plurality of classification models, a time series of operating data from one or more monitoring nodes of the wind turbine. The method further includes extracting, via the plurality of classification models, a plurality of features using the time series of operating data. Each of the plurality of features is a mathematical characterization of the time series of operating data. Moreover, the method includes generating an output from each of the plurality of classification models and determining, using a decision fusion module, a probability of the cyberattack occurring on the control system based on a combination of the outputs. Thus, the method includes implementing a control action when the probability exceeds a probability threshold.