System for heating, ventilation, air-conditioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current HVAC systems lack efficient and adaptive control mechanisms that can optimize energy usage based on real-time occupancy and environmental conditions, leading to suboptimal energy consumption and comfort levels.
Innovation Solution
A system utilizing a site controller with a cloud interface that processes sensor data from various sources, including occupancy sensors, to adjust actuator settings via a parametrized AI model, allowing for real-time optimization of heating, ventilation, and air-conditioning operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If traditional HVAC control systems are used, then the system structure is simple, but energy efficiency and adaptability to occupancy changes are poor
Solution Approach 1:
The system divides control into two segments: a cloud-based AI model training system that processes historical data and generates optimized control parameters, and a local controller that executes real-time control based on received parameters. This segmentation allows complex AI processing to occur in the cloud while keeping the local system relatively simple.
Solution Approach 2:
A communication interface serves as an intermediary between the cloud-based AI system and the local HVAC controller. This intermediary transmits optimized control parameters from the cloud to the local controller, enabling the local system to benefit from complex AI processing without requiring substantial local computational resources.
2Measurement precision
If real-time sensor data processing is implemented, then occupancy detection accuracy improves, but computational load increases
Solution Approach 1:
The AI model is trained in advance using historical sensor data and occupancy information to learn optimal control strategies. This preliminary training phase allows the system to process real-time sensor data more efficiently by applying pre-learned patterns rather than performing complex computations in real-time.
Solution Approach 2:
The patent replaces complex real-time computational processing with a pre-trained AI model that has already learned optimal control patterns. Instead of performing intensive computations locally in real-time, the system uses the pre-trained model's knowledge to quickly determine appropriate control actions based on current sensor readings.
3Productivity
If cloud-based AI processing is used, then control optimization improves, but system reliability may decrease due to network dependency
Solution Approach 1:
The system performs preliminary training of the AI model in the cloud to generate optimized control parameters and models. These pre-computed models are then transmitted to the local controller, which can execute control decisions autonomously without requiring continuous cloud connectivity, thus maintaining reliability during network outages.
Solution Approach 2:
The system dynamically adapts its operation mode based on network availability. When the cloud is accessible, it receives updated optimized parameters for improved control. When the cloud is inaccessible, the local controller seamlessly transitions to using previously received parameters, ensuring continuous operation without interruption.
Data Source
AI summary
Various embodiments include a site controller comprising: a processor; a memory storing a parametrized model for control of a HVAC system; a sensor interface connected to a first sensor; an actuator interface connected to a first actuator; and a cloud interface connected to a remote controller. The processor is configured to: read a first set of sensor signals from the first sensor; process the first set of signals into a set of measured values; transmit the measured values to the remote controller; in response, receive a set of model parameters from the remote controller; read the parametrized model from the memory; read an additional sensor signal from the first sensor; compute an actuator setting signal from the additional signal using the parametrized model and the set of model parameters; and transmit the actuator setting signal to the first actuator.