AI DC-DC Converter Control for Dynamic Load and Ripple Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional DC-DC converters face challenges in achieving stable voltage, efficient power transfer, and rapid response to dynamically changing load conditions due to limitations in controller algorithms and fixed control flow, which restricts simultaneous consideration of multiple output parameters like efficiency, settling time, and output ripple.
Innovation Solution
A DC-DC converter system utilizing a convolutional neural network to process two-dimensional state information from sensors, generating control signals that optimize power delivery, efficiency, and ripple response, allowing for dynamic adjustment of operation modes and parameters to satisfy varying load demands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional controller algorithms are sequentially executed to control multiple output conditions, then each condition can be controlled individually, but interference between output conditions occurs and target characteristics cannot be achieved
Solution Approach 1:
The patent segments the control problem by dividing multiple output conditions into separate virtual machines, where each virtual machine independently controls a specific output condition without interfering with others. This segmentation eliminates the interference problem while maintaining individual condition control.
Solution Approach 2:
The patent introduces a virtual machine as an intermediary layer between the controller and multiple output conditions. This virtual machine abstraction manages and coordinates control algorithms, allowing multiple conditions to be controlled simultaneously without direct interference.
2Productivity
If a fixed control flow is used in conventional DC-DC converters, then the control process is simple, but the processing speed is limited and fixed amount of time is required
Solution Approach 1:
The patent implements dynamic control flow where virtual machines can be created, suspended, and resumed based on real-time operating conditions. This dynamic approach allows the system to adapt processing resources to actual needs, improving processing speed while reducing unnecessary time consumption.
Solution Approach 2:
The patent changes the control parameter from fixed sequential execution to dynamic virtual machine scheduling. By adjusting which virtual machines are active and their execution priorities based on real-time conditions, the system optimizes processing speed and reduces control loop time.
3Loss of energy
If maximum power point tracking technique is used to improve energy transfer efficiency, then efficiency is optimized for light or heavy loads, but existing accumulated information cannot be considered and response to dynamic changes is limited
Solution Approach 1:
The patent creates multiple virtual machines that can dynamically switch between different power management modes (such as maximum power point tracking and other efficiency optimization techniques) based on real-time load conditions. This allows the system to adapt to dynamic changes while maintaining optimal efficiency.
Solution Approach 2:
The patent implements feedback mechanisms where virtual machines monitor operating conditions and adjust their control strategies accordingly. Accumulated information from previous operations is considered through this feedback loop, enabling the system to respond effectively to dynamic load changes while maintaining efficiency.
4Reliability
If numerous output parameters are sequentially considered in control loops, then each parameter can be controlled, but the time required for one control loop is fixedly increased in proportion to the number of parameters
Solution Approach 1:
The patent segments the control of numerous output parameters by assigning each parameter or group of parameters to dedicated virtual machines. This allows parallel processing of multiple parameters simultaneously, eliminating the sequential bottleneck and reducing overall control loop time while maintaining comprehensive parameter control.
Solution Approach 2:
The patent transitions from one-dimensional sequential parameter control to multi-dimensional parallel control through virtual machines. By adding the dimension of concurrent execution, the system can control multiple output parameters simultaneously without time proportionally increasing with the number of parameters.
Data Source
AI summary
As inputs of a controller of a direct current (DC)-DC converter are sampled for a predetermined time and thus two-dimensional state information in which one axis is an input physical quantity and the other axis is a time is generated, the two-dimensional state information is processed by a convolutional neural network to determine and output one of a plurality of control signals. An artificial intelligence control part may operate in accordance with a plurality of operation conditions or dynamically determined operation conditions by applying different artificial intelligence engines according to operation modes.