Physics-Based Control Sequence Generation to Reduce Training Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing approaches to controlling building states are cumbersome and laborious, requiring extensive training and computational resources due to the complex interactions between buildings, occupants, and equipment, making it difficult to efficiently maintain desired conditions.
Innovation Solution
A control sequence generation system using a heterogenous physics network with nodes representing equipment behavior, which iteratively adjusts control sequences through machine learning to minimize a cost function, optimizing energy use and comfort without the need for extensive training.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional control methods are used to manage building states, then comprehensive control of building conditions can be achieved, but the process becomes cumbersome and laborious requiring extensive training and computational resources
Solution Approach 1:
The patent replaces traditional machine learning training processes with a physics-based thermodynamic model. Instead of using complex neural networks that require extensive training data and computational resources, the system uses physics equations that inherently encode the relationships between building components, eliminating the need for laborious training while maintaining control accuracy.
Solution Approach 2:
The patent changes the fundamental approach from data-driven parameter optimization to physics-driven parameter relationships. By using thermodynamic equations that naturally represent the physical relationships in building systems, the model achieves reliability without requiring extensive training to learn these relationships from data.
2Productivity
If traditional machine learning models are used, then control sequences can be generated, but extensive training and computational resources are required
Solution Approach 1:
The patent substitutes traditional machine learning training computations with direct physics-based simulations. The thermodynamic model uses fundamental energy balance equations that require minimal computational resources compared to training deep neural networks, while still generating effective control sequences through iterative optimization.
Solution Approach 2:
The physics-based model inherently contains the knowledge needed to solve control problems through its structure and equations. Instead of requiring external training data and computational training processes, the model uses its built-in physics relationships to directly generate control sequences, making the system self-sufficient and computationally efficient.
3Reliability
If complex interactions between building, occupants, and equipment are modeled, then accurate control can be achieved, but the modeling process becomes daunting
Solution Approach 1:
The patent segments the building system into discrete thermodynamic zones and components, each represented by physics equations. This segmentation allows the complex interactions between building, occupants, and equipment to be modeled through manageable modular equations rather than a monolithic complex model, maintaining accuracy while reducing modeling difficulty.
Solution Approach 2:
The thermodynamic model uses universal physics equations that can represent multiple building components and interactions through a unified framework. The same energy balance equations apply to different zones, equipment, and environmental conditions, providing accurate modeling of complex interactions without requiring separate complex models for each component.
Data Source
AI summary
Various embodiments described herein relate to a method, device, and non-transitory machine-readable storage medium for controlling a system having a plurality of devices including one or more of the following: calculating a demand curve for a controlled space representing at least one target amount of state for delivery to the controlled space; creating a cost function that compares the demand curve to simulated state curves produced by a model of the controlled system based on respective sets of control actions issuable to the plurality of controlled devices; performing an optimization method that tunes a candidate set of control actions to reduce a cost output by the cost function based on the candidate set of control actions; and issuing control actions of the candidate set of control actions to respective ones of the plurality of controlled devices to cause the controlled system to deliver state to the controlled space.