HVAC Site Controller with Cloud-Trained AI Model for Adaptive Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current HVAC control systems lack efficient and adaptive methods for optimizing heating, ventilation, and air-conditioning operations, particularly in varying occupancy conditions and seasonal changes, while also ensuring privacy and handling diverse field devices and potential hazards.
Innovation Solution
A cloud-assisted AI system that trains a neural network model using data from sensors and actuators, differentiating between fast and slow temperature changes, and employing encryption for privacy, capable of operating with limited resources and handling a wide range of field devices, including drones, and switching to hazard mode for emergencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If cloud-based AI training is used to optimize HVAC control, then adaptability and control precision are improved, but device complexity and communication requirements increase
Solution Approach 1:
The patent introduces a cloud-based AI training system as an intermediary between the HVAC control system and optimization algorithms. The cloud server receives sensor data, trains neural network models, and returns trained models to the HVAC controller, thereby providing advanced adaptability without increasing local device complexity
Solution Approach 2:
The patent replaces traditional rule-based HVAC control mechanisms with AI-based neural network control. The neural network learns optimal control strategies from historical data and sensor inputs, providing superior adaptability compared to fixed mechanical control logic
2Measurement precision
If comprehensive sensor data collection is implemented to improve control accuracy, then measurement precision is improved, but energy consumption and data processing requirements increase
Solution Approach 1:
The patent implements selective data collection where the system gathers comprehensive sensor data during the AI training phase in the cloud, but only transmits essential processed features to the HVAC controller for real-time operation. This partial action approach maintains measurement precision while reducing ongoing energy consumption
Solution Approach 2:
The AI model performs self-learning and self-optimization using historical sensor data stored in the system. The neural network automatically identifies patterns and optimizes control parameters without requiring continuous external intervention or excessive real-time data processing, thereby reducing energy consumption
Data Source
Figure 1
Figure 2
Figure 3
AI summary
System for heating, ventilation, air-conditioning. A site controller (6) comprising a processor, a memory storing a parametrized model (9, 26) for control of a heating, ventilation and/or air-conditioning installation (3a - 3e, 4a - 4e, 5a - 5d) associated with the site (1), the processor being configured to: read a first set of sensor signals from a first sensor; process the first set of sensor signals into a first set of measured values; transmit the first set of measured values to a remote controller (8); receive model parameters for the parametrized model from the remote controller (8); read the parameterized model from the memory; read an additional sensor signal from the first sensor; compute an actuator setting signal using the parametrized model and using the received model parameters based on the additional sensor signal; and transmit the actuator setting signal to a first actuator (3a - 3e, 4a - 4e).