Hybrid Power Device Control for Real-Time Global Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing power device controllers struggle to find a global optimum in real-time operation due to high numerical effort, which is necessary for certified safe operation, especially in sensitive areas where clear assignment of manipulated variables to operating points is required.
Innovation Solution
A controller combines a non-evolutionary algorithm for accurate local optimization with an evolutionary algorithm to expand the search for a global optimum, allowing real-time operation and efficient optimization of power devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a non-evolutionary algorithm (e.g., gradient method) is used for optimization, then calculation accuracy for local optimum is improved, but the ability to find global optimum deteriorates
Solution Approach 1:
The patent combines a non-evolutionary algorithm (first optimization module) with an evolutionary algorithm (second optimization module) into a hybrid optimization system. The non-evolutionary algorithm provides accurate local optimization, while the evolutionary algorithm performs global search to escape local optima, thereby resolving the contradiction between calculation accuracy and global optimum finding capability.
2Reliability
If an evolutionary algorithm is used to search for global optimum, then reliability of operation is improved, but numerical effort and computational complexity increase
Solution Approach 1:
The optimization process is segmented into two distinct phases: a first optimization phase using a non-evolutionary algorithm for quick local optimization, and a second optimization phase using an evolutionary algorithm for global search. This segmentation allows the computationally intensive evolutionary algorithm to be applied only when necessary, reducing overall numerical effort while maintaining reliability.
Solution Approach 2:
The non-evolutionary algorithm performs preliminary optimization to find a local optimum before the evolutionary algorithm is activated. This preliminary action provides a good starting point for the evolutionary algorithm, reducing the search space and computational effort required for global optimization while ensuring certified safe operation.
3Productivity
If a hybrid optimization algorithm is implemented, then real-time optimization capability is improved, but controller complexity increases
Solution Approach 1:
The controller dynamically switches between the non-evolutionary and evolutionary algorithms based on operational requirements. The system can adaptively choose the appropriate optimization method for different operating conditions, enabling real-time optimization while managing controller complexity through conditional execution rather than always-active complex algorithms.
Data Source
AI summary
A controller for a power device, the controller including: a first optimization module, which is configured to determine a first optimized manipulated variable vector by way of a first non-evolutionary algorithm, the controller being configured to determine an actuation manipulated variable vector which includes a plurality of manipulated variables for actuating the power device; a second optimization module, which is configured to receive the first optimized manipulated variable vector from the first optimization module and to determine a second optimized manipulated variable vector by way of a second evolutionary algorithm using the first optimized manipulated variable vector; and a control module, which is configured to determine the actuation manipulated variable vector based on the second optimized manipulated variable vector.