Modular Converter Control via Model Predictive Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modular converters face challenges in reducing switching losses, balancing capacitor voltages, and minimizing harmonics in input and output currents, particularly during both steady-state and transient operating conditions.
Innovation Solution
A method for controlling modular converters that involves selecting future switching sequences, predicting current trajectories, and evaluating a cost function to optimize switching states, thereby directly controlling load currents without the need for a modulation stage, and balancing capacitor voltages by minimizing switching frequency and harmonic distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If conventional hierarchical control with vector control and carrier-based PWM is used, then closed-loop control of load currents is achieved, but switching losses are high and capacitor voltage balancing is complex
Solution Approach 1:
The patent changes the control parameters from continuous PWM duty cycles to discrete switching states (0 or 1 for each converter module). This discretization reduces switching frequency and losses while simplifying the control structure. The cost function evaluates multiple candidate switching sequences and selects the optimal one, replacing the complex hierarchical vector control with a direct model predictive approach that inherently balances capacitor voltages through the switching state selection.
2Reliability
If capacitor voltages are balanced by sorting and prioritizing charging/discharging, then voltage balancing is achieved, but control complexity increases and response during transients is slow
Solution Approach 1:
The patent performs preliminary evaluation of future capacitor voltages for all candidate switching sequences before selecting the optimal sequence. The cost function includes terms that predict future module voltages and penalize sequences that lead to voltage imbalance. This preliminary assessment ensures capacitor voltage balancing is achieved proactively rather than through reactive sorting and prioritization, improving transient response while maintaining balancing reliability.
3Loss of energy
If switching frequency is reduced to lower switching losses, then energy efficiency improves, but current control precision and harmonic reduction become more difficult
Solution Approach 1:
The patent implements a feedback mechanism where the cost function evaluates the predicted impact of each candidate switching sequence on load currents, capacitor voltages, and circulating currents. The actual system state is fed back into the prediction model to determine the next optimal switching state. This closed-loop model predictive control maintains current control precision even at reduced switching frequencies by continuously adapting switching decisions based on actual system behavior and predicted outcomes.
4Adaptability or versatility
If hierarchical control layers are used to address circulating currents and energy balance, then comprehensive control is achieved, but overall control complexity and computational burden increase
Solution Approach 1:
The patent merges the functions of multiple hierarchical control layers into a single unified model predictive control framework. The cost function simultaneously considers load current tracking, capacitor voltage balancing, and circulating current minimization in one optimization step. This consolidation eliminates the need for separate control loops and hierarchical layers, reducing overall control complexity while maintaining comprehensive adaptability across all control objectives through the integrated cost function evaluation.
Data Source
AI summary
A method for controlling a modular converter with a plurality of converter modules includes: selecting possible future switching sequences of the converter based on an actual converter switching state; predicting a future current trajectory for each switching sequence based on actual internal currents and on actual internal voltages; and determining candidate sequences from the switching sequences, wherein a candidate sequence is a switching sequence with a current trajectory that respects predefined bounds with respect to a reference current or, when a predefined bound is violated, moves the current closer to such a predefined bound. The method includes predicting future module voltages for each candidate sequence; evaluating a cost function for each candidate sequence; and selecting the next converter switching state as a first converter switching state of a candidate sequence with minimal costs.