Protection Relay Decision Logic Generation for Stochastic Grid Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The development of protection logic for industrial automation control systems is increasingly complex due to the introduction of converter-interfaced generation and e-mobility, which introduces stochasticity and challenges the reliability and security of power systems, requiring frequent reevaluation and manual design by expert engineers.
Innovation Solution
A computer-implemented method using machine learning, specifically generative adversarial networks, to autonomously generate and test decision logic for protection relays, simulating various scenarios to enhance performance and robustness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If protection logic is designed manually by human engineers, then the design process can be performed with existing expertise, but the complexity and time required increase significantly due to stochasticity from converter-interfaced generation and e-mobility
Solution Approach 1:
The patent replaces the manual mechanical design process with an automated machine learning system. A GAN-based architecture automatically generates and optimizes protection logic candidates, substituting human engineer expertise with algorithmic processing. The system performs simulations and evaluations autonomously, transforming the manual design methodology into an automated computational process that handles the increased complexity of modern power systems with converter-interfaced generation and e-mobility integration.
2Adaptability or versatility
If protection systems are frequently reevaluated to adapt to changing grid conditions, then system adaptability improves, but the time and resources required for redesign increase
Solution Approach 1:
The patent implements a dynamic protection logic design system that can be frequently reevaluated and updated. The machine learning model accepts new simulation data and grid condition information as inputs, continuously adapting the protection logic to changing system conditions. This dynamic approach allows the protection system to evolve with the grid, incorporating new converter-interfaced generation and e-mobility patterns without requiring complete redesign cycles, thereby reducing adaptation time while maintaining high adaptability.
3Reliability
If comprehensive simulation testing is performed for all anticipated scenarios, then the security and dependability of the protection system are maximized, but the computational time and resources increase
Solution Approach 1:
The patent employs preliminary action by pre-generating diverse simulation scenarios and pre-training the GAN model on comprehensive test cases. The system prepares a library of anticipated grid conditions, faults, and disturbances in advance, allowing the protection logic to be tested against these pre-defined scenarios. This preliminary preparation enables thorough security verification without requiring extensive real-time simulation during the design phase, as the model has already learned from comprehensive pre-simulated conditions.
Solution Approach 2:
The patent uses copying by creating multiple decision logic candidates through the GAN's generative process. Instead of extensively testing a single logic design, the system generates numerous variant candidates that can be evaluated in parallel. This copying approach distributes the simulation testing workload across multiple candidates, allowing comprehensive scenario testing to be performed more efficiently by evaluating several simplified copies simultaneously rather than exhaustively testing one design sequentially.
Data Source
AI summary
To generate a decision logic for a controller of an Industrial Automation Control System, IACS, an iterative process is performed in which a decision logic candidate for the decision logic is generated, and a performance of the decision logic candidate in response to scenarios is computed.