Building Services Control Using Rules and ML Preference Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of building technology systems face challenges in setting and adjusting rules for these systems, particularly in reflecting individual preferences for ambient conditions such as temperature and lighting, which often leads to manual operation of control elements.
Innovation Solution
A computer-implemented method that determines the parameterization of building technology systems by using sensor data to evaluate ambient conditions and either apply predefined parameterization rules if insufficient data is available or utilize a machine learning model when sufficient data exists to learn user preferences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If rule-based control units are used to determine parameterization based on predefined rules, then user comfort is increased and manual operation is reduced, but setting and adjusting rules requires user knowledge and effort which defeats the purpose
Solution Approach 1:
The control unit automatically learns user preferences by monitoring manual adjustments to parameterization, eliminating the need for users to explicitly set rules. The system serves itself by inferring preferences from observed behavior patterns, transforming the rule-setting task from a user responsibility to an automated process.
Solution Approach 2:
The system continuously monitors user interactions with control elements and uses this feedback to refine and update the learned preferences. By analyzing patterns in manual adjustments, the control unit adapts the parameterization rules to better match user preferences over time, creating a closed-loop learning system.
2Adaptability or versatility
If users manually adjust control elements to meet preferences, then individual preferences are satisfied, but user comfort decreases and operation becomes tedious
Solution Approach 1:
The control unit performs preliminary learning during an initial phase by monitoring user adjustments and building a preference model before full automated control is activated. This preliminary action allows the system to prepare personalized parameterization rules in advance, avoiding the need for continuous manual intervention once learning is complete.
Solution Approach 2:
The system automatically adjusts parameterization based on learned preferences without requiring continuous user input. The control unit serves itself by autonomously making parameterization decisions based on the stored preference model, eliminating the need for users to repeatedly manually adjust settings.
3Extent of automation
If machine learning models are used to learn user preferences, then automated parameterization is achieved, but sufficient training data is required which may not be available in newly installed systems
Solution Approach 1:
The system implements a two-phase approach: an initial learning phase where the control unit monitors and stores user adjustments to build training data, followed by an automated phase where the machine learning model is deployed. This preliminary data collection ensures sufficient training data is available before full automation begins.
Solution Approach 2:
The system dynamically transitions from a data-collection mode to an automated machine learning mode as sufficient training data becomes available. The level of automation is adjusted based on the accumulated quantity of training data, allowing the system to adapt its operational mode to the current data availability status.
Data Source
AI summary
A computer-implemented method for determining a parameterization of a building technology system. An ambient condition state is received that is detected by at least one sensor device. A number of ambient condition states stored in a database which satisfy a similarity criterion with regard to the detected ambient condition state is determined. If the number is less than a predetermined threshold, the parameterization is determined of the building technology system based on at least one predefined parameterization rule, or if the number is greater than or equal to the predetermined threshold, the parameterization is determined of the building technology system based on a machine learning model.