AI-Based Medium-Voltage Drive Control Using Reduced-Order Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medium voltage variable frequency drives face challenges in efficiently controlling and optimizing the operation of multilevel power converters due to the complexity of high-voltage handling and the need for precise control of semiconductor switches, which is not adequately addressed by existing digital control systems.
Innovation Solution
A control system that utilizes a reduced order model accessed by a processor to analyze sensor values and determine operating modes for power converter cells, incorporating artificial intelligence for self-learning and predictive maintenance, enabling adaptive responses and efficient operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a traditional digital control system is used for multilevel power converters, then the system structure is well-defined with separate control components, but the control precision and optimization capability are insufficient for high-voltage handling
Solution Approach 1:
The patent transforms the control approach by changing from traditional parameter-based control to model-based control using reduced order models. This allows the system to handle high-voltage conditions more precisely by using simplified dynamic models that capture essential behavior without the complexity of full-order models, thereby improving control precision while managing system complexity.
Solution Approach 2:
The patent introduces reduced order models as intermediary elements between the complex power converter system and the control processor. These models act as mediators that simplify the representation of system dynamics, enabling the control system to make accurate decisions without directly processing the full complexity of the multilevel power converter equations.
2Measurement precision
If full-order models are used for control and analysis, then accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent applies parameter changes by reducing the order of the system models used for control. The reduced order models retain the essential dynamic characteristics needed for accurate control decisions while eliminating unnecessary complexity, thereby achieving a balance between model accuracy and computational efficiency that enables real-time control of medium voltage drives.
Solution Approach 2:
The patent uses partial action by implementing control based on reduced order models that capture only the most critical system dynamics rather than attempting to model every aspect of the power converter. This partial modeling approach provides sufficient accuracy for control purposes while dramatically reducing computational requirements compared to full-order models.
3Adaptability or versatility
If AI and reduced order models are integrated into the control system, then adaptive control and predictive maintenance are enabled, but device complexity increases
Solution Approach 1:
The patent implements multi-functionality by integrating AI algorithms with the reduced order model-based control system. This unified control architecture performs multiple functions including real-time adaptive control, predictive maintenance, and system optimization within a single integrated framework, thereby enhancing adaptability while managing complexity through functional consolidation.
Solution Approach 2:
The patent enables self-service capabilities through the integration of AI and reduced order models, allowing the system to automatically adapt its control parameters and predict maintenance needs without external intervention. The reduced order models provide real-time system state estimation that feeds into AI algorithms for autonomous decision-making, reducing the need for complex external control infrastructure.
Data Source
AI summary
A variable frequency drive system includes a power converter with a plurality of power cells supplying power to one or more output phases, each power cell having multiple switching devices incorporating semiconductor switches; a plurality of sensors monitoring values of the power converter; and a control system in communication with the power converter and controlling operation of the plurality of power cells, the control system comprising a processor configured via executable instructions to access a first reduced order model of the power converter; receive the values provided by the plurality of sensors; analyze the values in connection with the first reduced order model to determine one or more operating modes; and output one or more determined operating modes of the power converter.