Sigma-Delta Converter Modeling With Recurrent Encoder-Decoder Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sigma-delta converter design methods using deep learning and artificial intelligence are limited in their ability to adapt to specific hardware and performance constraints, requiring tedious redesign for each converter topology.
Innovation Solution
A method involving supervised deep learning with a generic sigma-delta converter model using a recurrent encoder and decoder, constrained by material and functional properties, to optimize a converter design that meets specific hardware specifications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning methods are applied to adapt weights of predefined converter topologies, then the converter can meet specific performance requirements, but the design process requires tedious redesign for each new converter topology
Solution Approach 1:
The patent applies universality by creating a single generic converter model that can be adapted to multiple different converter topologies through weight optimization. Instead of designing separate models for each topology, the same model structure serves multiple functions by learning topology-specific weights through supervised deep learning, thereby reducing redesign time while maintaining performance requirements.
Solution Approach 2:
The patent utilizes parameter changes by optimizing the weight parameters of the generic converter model through supervised deep learning. By changing the weight parameters rather than the model structure itself, the same model can adapt to different converter topologies and performance requirements, eliminating the need for tedious redesign of each topology.
2Loss of time
If a generic converter model is used for multiple topologies, then redesign time is reduced, but the model must be constrained by material and functional properties to ensure accuracy
Solution Approach 1:
The patent applies feedback through supervised deep learning, where the generic converter model's outputs are compared against target values and the weight parameters are iteratively adjusted based on this feedback. This feedback mechanism ensures that the model adheres to material and functional constraints while maintaining accuracy across different converter topologies, balancing the reduction in redesign time with the necessary model complexity.
Data Source
AI summary
The present description concerns a method for designing a sigma-delta type converter comprising a step of supervised deep learning applied to a converter model. The converter model comprises at least one recurrent encoder and at least one recurrent decoder. Each recurrent encoder is based on a generic model comprising a succession of K identical generic cells Cellk, with K an integer parameter greater than or equal to 1 and k an integer index ranging from 1 to K. The sigma-delta converter is obtained by manufacturing an electronic circuit corresponding to the model obtained after the training.