System and method for controlling an operation of a vapor compression cycle based on a hybrid model of dynamics of the vapor compression cycle
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vapor compression cycle models face challenges in achieving accurate predictions due to mismatches between physics-based and data-driven approaches, with physics-based models requiring unjustifiable complexity and data-driven models lacking interpretability and neglecting fundamental physical laws, while both require large datasets and sensor data limitations.
Innovation Solution
A hybrid model combining physics-based and data-driven models, using a constrained Kalman smoother to estimate physics-based parameters and a neural network to correct residual errors, optimizing the joint optimization problem into two separate unconstrained problems for efficient parameter estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a physics-based model is used for predicting vapor compression cycle behavior, then the model is derived from first principles of physics, but simplifying assumptions lead to a mismatch between model predictions and data collected from the system
Solution Approach 1:
The patent combines a physics-based model with a data-driven model to create a hybrid model. The physics-based model provides the fundamental structure and relationships, while the data-driven model compensates for the inaccuracies introduced by simplifying assumptions, thereby improving prediction accuracy without requiring excessive complexity
Solution Approach 2:
The patent introduces correction parameters through the data-driven model that adjust the predictions of the physics-based model. These parameters are learned from system data and modify the model outputs to better match actual system behavior, effectively changing the model parameters to improve accuracy
2Reliability
If a data-driven model is used for predicting vapor compression cycle behavior, then the model can capture complex system behavior, but large datasets including full-state trajectories are required which are unavailable due to limited sensor data
Solution Approach 1:
The hybrid model merges the strengths of both physics-based and data-driven approaches. The physics-based model provides a structured framework that can operate with limited data, while the data-driven component learns from available sensor data to correct systematic errors, achieving good accuracy without requiring large datasets
Solution Approach 2:
The physics-based model acts as an intermediary that structures the relationship between inputs and outputs, enabling the data-driven model to learn more efficiently from limited data. The physics model provides priors that guide the learning process, reducing the amount of data needed compared to a purely data-driven approach
3Adaptability or versatility
If a data-driven model is used for predicting vapor compression cycle behavior, then the model can be flexible, but the model suffers from non-interpretability and does not explicitly account for fundamental physical laws that govern the system
Solution Approach 1:
The hybrid model combines the interpretability and physical law adherence of physics-based models with the flexibility and adaptability of data-driven models. The physics-based component ensures fundamental physical laws are respected, while the data-driven component adds flexibility to capture complex behaviors not explicitly modeled in the physics equations
Data Source
AI summary
The present disclosure discloses a system and a method for controlling an operation of a vapor compression cycle based on a hybrid model of dynamics of the vapor compression cycle including a physics-based model and a data-driven model. The method comprises executing a constrained Kalman smoother over the observed variables collected over multiple instances of time to jointly estimate the parameters of the physics-based model and states of the vapor compression cycle, and updating the data-driven model to minimize a difference between the states estimated by executing the constrained Kalman smoother and the states predicted by the physics-based model. The method further comprises updating the hybrid model with the estimated parameters of the physics-based model and the updated data-driven model, and controlling the operation of the vapor compression cycle using the updated hybrid model.