Building Automation Control Using ML for Adaptive User Behavior
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing building automation systems require significant user effort and time for programming and adjustment, lacking flexibility to adapt to changing user behaviors without manual programming.
Innovation Solution
A building automation system that incorporates a database for storing past data, a communication unit for sensor data, an actuator control unit, and a machine learning model to generate action recommendations and control actuators, reducing the need for manual programming and enhancing flexibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional building automation systems are used with manual programming and adjustment, then the system can be controlled according to user needs, but the user effort and time required for programming and adjustment is significant
Solution Approach 1:
The system automatically learns user behaviors and preferences through observation and machine learning algorithms, enabling self-configuration without manual programming. The automation system performs self-adjustment based on learned patterns, eliminating the need for extensive user programming effort while maintaining high automation capability.
Solution Approach 2:
The system pre-learns and stores user behavior patterns during initial observation periods, preparing action recommendations in advance. This preliminary learning phase enables the system to provide automated control recommendations immediately when needed, reducing both programming effort and response time.
2Adaptability or versatility
If traditional building automation systems are used with fixed programming, then the system structure is stable, but the flexibility to adapt to changing user behaviors is limited
Solution Approach 1:
The system dynamically adapts its control strategies by continuously learning new user behaviors and updating its knowledge base. The machine learning models are designed to evolve with changing user preferences while maintaining stable core control functions, achieving both adaptability and stability through hierarchical architecture.
Solution Approach 2:
The system incorporates feedback loops where user responses to automated recommendations are continuously monitored and used to refine future recommendations. This feedback mechanism allows the system to adapt to changing behaviors while maintaining stable operational patterns through iterative improvement.
3Measurement precision
If machine learning models are used to generate action recommendations, then the accuracy and correctness of recommendations is improved, but the complexity of the system increases
Solution Approach 1:
The machine learning system is divided into multiple specialized models, each handling specific aspects of building control (e.g., temperature control, lighting, shading). This segmentation allows each model to focus on specific prediction tasks, improving accuracy while managing overall system complexity through modular architecture.
Solution Approach 2:
An intermediary layer is introduced between the machine learning models and the building actuators, which translates complex model predictions into actionable control commands. This intermediary simplifies the interface between the complex AI system and the physical building systems, reducing perceived complexity while maintaining high recommendation accuracy.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The present invention relates to a building automation system (10) for automatically controlling the facility of a building. The building automation system (10) comprises a data base (14), for storing past data, comprising event data, sensor data, user context data (30) and actions manually created and/or predefined by the user (34) in relation to this data. Further, a communication unit (22) for getting sensor data of at least one sensor (26a, 26b, 26c) of the building and user context data (30), and an actuator control unit (42) for controlling at least one actuator (46a, 46b, 46c) of the building are provided. The building automation system (10) further comprises a proposal unit (38) for outputting recommended actions to the user (34) and receiving feedback of the user (34) thereto In addition a computer implemented machine learning model (18) and/or process model (50) created via data mining is described, for generating action recommendations for the user (34) and control commands for the actuator control unit (42).