Control sequence generation system and methods
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods face challenges in efficiently managing and maintaining the optimal state of buildings by controlling temperature and humidity levels across different zones, due to complex interactions between the building, its occupants, and external factors like uneven heating.
Innovation Solution
A control sequence generation system using a heterogenous physics network neural network that iteratively adjusts control sequences based on a cost function, comparing simulated and target demand curves, to optimize equipment operation and energy usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If traditional building state control methods are used, then equipment can maintain basic operation, but energy consumption is high and comfort optimization is limited
Solution Approach 1:
The system performs preliminary action by using a thermodynamic model to predict future building state requirements and pre-optimize control sequences before they are executed. The iterative optimization process anticipates energy consumption patterns and adjusts equipment operation in advance, allowing the system to reduce energy loss without requiring complex real-time control mechanisms during actual operation.
Solution Approach 2:
The patent replaces traditional mechanical control systems with a computational approach using a thermodynamic model and iterative optimization algorithm. Instead of relying on complex mechanical sensors and actuators to continuously adjust building states, the system uses software-based simulation and cost function minimization to determine optimal control sequences, thereby reducing energy consumption without proportionally increasing physical device complexity.
2Productivity
If iterative optimization is applied to minimize cost function, then energy efficiency improves, but computational time and iterations required increase
Solution Approach 1:
The system applies preliminary action by pre-defining the thermodynamic model structure and cost function parameters before optimization begins. This allows the iterative process to focus only on adjusting control sequence variables rather than exploring the entire solution space, significantly reducing computational iteration time while maintaining high energy efficiency results.
Solution Approach 2:
The optimization process uses dynamics by adaptively adjusting the control sequences based on feedback from each iteration. The system dynamically modifies equipment operation schedules and intensity levels in response to simulated building state outcomes, allowing it to converge on optimal energy efficiency solutions faster through iterative learning rather than static pre-programmed controls.
3Manufacturing precision
If thermodynamic model simulation is used to optimize control sequences, then building comfort and state accuracy improve, but computational complexity increases
Solution Approach 1:
The patent substitutes complex physical measurement and adjustment mechanisms with a computational thermodynamic model. Instead of using numerous sensors and manual adjustment devices to achieve precise state control, the system uses software-based thermal and humidity simulations to predict building responses, thereby improving state control accuracy while avoiding proportional increases in physical device complexity.
Solution Approach 2:
The system uses copying by creating a virtual thermodynamic model that replicates building physics behavior. This digital twin allows the optimization process to simulate and evaluate control sequences without physically implementing them, enabling high state control accuracy through repeated virtual testing while keeping the actual building equipment simple and unchanged.
Data Source
AI summary
A model receives a target demand curve as an input and outputs an optimized control sequence that allows equipment within a physical space to be run optimally. A thermodynamic model is created that represents equipment within the physical space, with the equipment being laid out as nodes within the model according to the equipment flow in the physical space. The equipment activation functions comprise equations that mimic equipment operation. Values flow between the nodes similarly to how states flow between the actual equipment. The model is run such that a control sequence is used as input into the neural network; the neural network outputs a demand curve which is then checked against the target demand curve. Machine learning methods are then used to determine a new control sequence. The model is run until a goal state is reached.