Component-Based Modeling Using Trained Surrogates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for modeling complex physical processes, such as dynamic HVAC systems, are time- and resource-intensive, resulting in simulations that are not commercially viable due to their slow processing times, which are often in real-time or take months to complete, making it difficult to make timely decisions.
Innovation Solution
The use of component-based modeling with trained surrogates that replace complex systems with smaller, pre-trained approximations, allowing for accelerated differential-equation solving and simulation, utilizing neural networks and other learnable functions to recreate system dynamics, enabling faster and more accurate simulations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional neural-network libraries are used for modeling physical processes, then the model can be built using standard tools, but the simulation speed is too slow to be commercially viable
Solution Approach 1:
The patent creates simplified surrogate models that copy the essential dynamics of complex physical systems. These surrogates are trained to reproduce the behavior of full-order models but require significantly fewer computational resources, enabling fast simulation while maintaining acceptable accuracy for decision-making purposes.
Solution Approach 2:
The patent segments the complex physical system into multiple components or states, representing each with simplified dynamics. This segmentation allows the overall system behavior to be reconstructed from simpler parts, reducing computational complexity while preserving essential system characteristics.
2Measurement precision
If detailed dynamic models are used to ensure accuracy, then the model captures system behavior precisely, but the simulation takes months to complete
Solution Approach 1:
The patent changes the parameters of the model by reducing the order of differential equations and simplifying physical relationships. This parameter reduction maintains the essential dynamic behavior needed for accuracy while dramatically decreasing computational requirements, enabling simulations to run in minutes rather than months.
3Speed
If real-time simulation speed is achieved, then decisions can be made quickly, but the model must be significantly simplified
Solution Approach 1:
The patent applies partial action by capturing only the most critical system dynamics and behaviors needed for decision-making, rather than modeling every detail. This selective approach achieves sufficient fidelity for practical applications while enabling real-time simulation speeds.
Data Source
AI summary
Systems and methods of component-based modeling using trained surrogates are disclosed. A component architecture is made up of composable subsystem components. The composable subsystem components are reusable, such that the subsystem components are trained, and a library of trained model-reduced forms is created, which allows for large-scale models to be automatically accelerated for modeling. In this way, complex models are built by stitching together pre-designed, pre-shrunk components consisting of self-contained systems. A novel combination of using surrogates for modeling purposes and accelerated solving is provided to create an architecture that simulates complex physical processes that were previously infeasible to simulate in a commercially reasonable time.