Thermostat Time-to-Temperature Prediction Using Global and Local Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Intelligent controllers lack the capability to continuously and accurately calculate and display the time remaining until a control task is completed, particularly in dynamic environments where multiple factors influence the controlled parameters, such as temperature in HVAC systems.
Innovation Solution
The intelligent controller employs multiple models for predicting the time remaining until specified parameters are reached, using data collected over time to generate both global and local models, which are used to estimate and display the remaining time for completion of control tasks, with a specific example demonstrated in intelligent thermostats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple models are used to predict time remaining, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent divides the prediction system into multiple separate models (global model and local models) that can be independently constructed and maintained. Each model focuses on specific aspects of time behavior, allowing them to be developed, validated, and updated independently while working together to provide comprehensive predictions.
Solution Approach 2:
The system dynamically selects and switches between different models based on current operating conditions. Local models are used when specific conditions are met, otherwise the global model is applied. This dynamic approach allows the system to adapt to changing environments while maintaining prediction accuracy.
2Loss of information
If continuous calculation and display of time remaining is implemented, then information completeness is improved, but loss of time for processing increases
Solution Approach 1:
The system pre-calculates and stores time behavior models based on historical data before they are needed for prediction. By having these models ready in advance, the system can quickly query and apply them during operation without performing complex real-time calculations, thus reducing processing time while maintaining information completeness.
Solution Approach 2:
The patent creates simplified representations (models) of complex time behavior patterns from historical data. These models serve as copies that capture essential time behavior characteristics without requiring the full complexity of original data, enabling fast predictions while preserving accurate information about time remaining.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method of calculating a time-to-temperature estimate for a thermostat controlling an HVAC system comprises receiving a first temperature representing a current temperature of an enclosure and receiving a second temperature representing a target temperature of the enclosure. A first component of an estimated time representing an estimated time to transition the temperature in the enclosure from the first temperature to the second temperature by the HVAC system is computed. The first component is computed based on: a plurality of historical time and temperature values recorded from one or more previous temperature change cycles of the enclosure; and at least one external factor selected from a group consisting of: an outside temperature, an outside sunlight amount, an outside wind velocity, and a time of day. A second component of the estimated time based on a current trajectory of time and temperature values as the enclosure transitions from the first temperature to the second temperature is computed. The method also includes: computing the estimated time based on the first component of the estimated time and the second component of the estimated time; and causing the estimated time to be displayed on a display device. A thermostat for calculating a time-to-temperature estimate is also disclosed and claimed.