Zone-Level Occupancy Prediction for HVAC Energy Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
HVAC systems in large buildings face challenges in efficiently managing energy consumption due to difficulty in predicting when different zones will be occupied, leading to unnecessary energy use in unoccupied areas.
Innovation Solution
A method and system using machine learning models trained on occupancy data from sensors like Wi-Fi Access Points, Bluetooth low energy sensors, and CO2 sensors to predict zone-level occupancy and adjust temperature set points accordingly, optimizing heating and cooling cycles based on real-time occupancy trends and booking status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the HVAC system is activated early to ensure comfortable temperature in occupied zones, then occupancy comfort is improved, but energy consumption increases due to heating/cooling unoccupied zones
Solution Approach 1:
The building is divided into multiple zones with independent HVAC control. Each zone's temperature is adjusted based on its specific occupancy status rather than treating the entire building as a single unit, allowing unoccupied zones to be excluded from heating/cooling operations
Solution Approach 2:
The system predicts future occupancy patterns using machine learning models trained on historical data. Temperature adjustments are made in advance based on predicted occupancy, ensuring comfortable conditions are ready before occupants arrive while avoiding energy waste in zones predicted to remain unoccupied
Solution Approach 3:
The system continuously monitors actual occupancy using sensors (motion detectors, access control systems, CO2 sensors) and compares it with predicted occupancy. This feedback loop allows the system to learn from prediction accuracy and adjust future predictions, while also enabling real-time HVAC adjustments when actual occupancy differs from predictions
2Stability of the object's composition
If the HVAC system operates continuously to maintain comfort in all zones, then temperature stability is improved, but energy efficiency deteriorates
Solution Approach 1:
The HVAC system transitions from static continuous operation to dynamic operation that adapts to changing conditions. The system adjusts heating/cooling operations in real-time based on predicted and actual occupancy patterns, maintaining temperature stability only in zones where occupants are present or expected
Solution Approach 2:
The system changes operational parameters (temperature setpoints, HVAC equipment runtime) based on occupancy predictions. When zones are predicted to be unoccupied, the system adjusts parameters to reduce or eliminate heating/cooling operations, thereby improving energy efficiency while maintaining comfort in occupied zones
3Loss of energy
If zone-level occupancy prediction is implemented, then energy optimization is improved, but system complexity increases due to multiple sensors and machine learning models
Solution Approach 1:
The system uses multi-functional components that serve multiple purposes. For example, access control sensors not only track building entry/exit but also provide occupancy data for HVAC control. The machine learning model serves both occupancy prediction and anomaly detection functions, reducing the need for separate dedicated systems
Solution Approach 2:
The system introduces an intermediary layer (the machine learning prediction system) that processes sensor data and translates it into actionable HVAC control decisions. This intermediary simplifies the overall system architecture by decoupling the sensing layer from the control layer, allowing each to be optimized independently
Data Source
AI summary
A method and system to predict occupancy status is disclosed. The method comprises receiving, via at least one processor, occupancy data of one or more zones via a plurality of sensors for a first period of time; determining occupancy trends for each zone for the first period of time using a trained machine learning (ML) model; mapping the determined occupancy trends for each zone with fluctuations in occupancy of each zone in real-time and booking status of each zone; and predicting occupancy of each zone for a second period of time and a threshold time for each zone to heat or cool at one or more temperature set points using the trained ML model based at least on the mapping. Thereafter, the method comprises adjusting the one or more temperature set points for each zone at the threshold time based at least on the prediction.