Circuit Trace Design Optimization Using Neural Network Surrogates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The design of circuit traces in printed circuit boards for high-speed communications is an iterative process heavily reliant on designer experience and intuition, requiring extensive processing power and time for accurate simulations, making it inefficient.
Innovation Solution
Training an artificial neural network with design parameter values to determine output formulas for circuit traces, allowing for the fabrication of optimized traces based on these formulas, reducing the need for extensive simulation runs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If accurate simulation models are used for circuit trace design, then design precision is improved, but processing time and computational resources increase
Solution Approach 1:
The patent pre-trains neural network models using extensive simulation data before actual design work. This preliminary training phase captures complex electromagnetic behaviors and material properties, allowing the model to make accurate predictions during actual design without requiring time-consuming simulations at that stage.
Solution Approach 2:
The patent creates simplified neural network surrogate models that copy the input-output behavior of complex electromagnetic simulation models. These surrogate models approximate the results of full-wave simulations but execute much faster, enabling rapid design iteration while maintaining acceptable accuracy.
2Loss of time
If simplified simulation models are used for circuit trace design, then processing time is reduced, but design precision deteriorates
Solution Approach 1:
The patent introduces neural network models as intermediary components between the simplified design inputs and the required output predictions. These neural networks act as mediators that learn complex relationships from training data, providing accurate predictions without requiring complex simulation models during the actual design process.
Solution Approach 2:
The patent transforms the design approach by changing from direct electromagnetic simulation to neural network prediction. This parameter change involves converting continuous simulation problems into discrete training data points that the neural network can process efficiently, fundamentally altering how design predictions are made.
3Manufacturing precision
If iterative design process with extensive simulation is used, then manufacturing precision is improved, but productivity decreases
Solution Approach 1:
The patent performs preliminary training of neural network models with comprehensive simulation data that covers various manufacturing tolerances and material variations. This pre-computed knowledge base enables rapid evaluation of design options without repeating extensive simulations for each iteration, significantly improving design throughput while maintaining precision.
Solution Approach 2:
The patent implements a dynamic design process where the neural network model adapts to different design scenarios and can provide real-time feedback. This dynamic approach replaces static, time-consuming iterative simulations with flexible, rapid predictions that maintain accuracy across varying design conditions.
Data Source
AI summary
A method includes training an artificial neural network with training data that comprises a sets of design parameter values for design parameters for circuit traces in a high speed communication link, determining an output formula that relates a sets of design parameters to a corresponding output parameter for the circuit traces in response to training the artificial neural network, running the output formula using a second set of design parameter values to obtain a corresponding set of output parameters for the circuit traces, determining that the corresponding set of output parameters differ from a set of modeled output parameters by less than a predefined percentage, and fabricating a circuit trace in a printed circuit board based upon the output formula in response to determining that the corresponding set of output parameters differ from the set of modeled output parameters by less than the predefined percentage.