Lighting System Sensor Data Modeling for Building Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern lighting systems face challenges in efficiently processing and modeling environmental sensor data due to large volumes of redundant and irregular data, which affects the accuracy and reliability of lighting control and other building automation systems.
Innovation Solution
A method involving a learning procedure to define a model of environmental conditions based on sensor data, evaluating sensor data strength, selecting a subset of sensors for optimal data usage, and applying the model to control building automation systems, thereby reducing redundant data processing and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data is collected from a large number of sensors over prolonged period of time to enable defining and using a model for predicting environmental conditions, then the model accuracy is improved, but the costs arising from storing and processing the sensor data increase significantly
Solution Approach 1:
The patent extracts and removes redundant sensor data from the dataset before model training. The system identifies and eliminates duplicate or highly correlated sensor readings, keeping only the most informative data points. This extraction process reduces the overall data volume while preserving the essential patterns needed for accurate environmental condition prediction.
Solution Approach 2:
The patent applies different processing quality levels to different portions of sensor data based on their relevance. High-priority sensors providing critical environmental information undergo more rigorous validation and are retained with full detail, while lower-priority redundant sensors are aggregated or summarized. This local quality approach ensures model accuracy is maintained for critical parameters while reducing overall data storage requirements.
2Adaptability or versatility
If sensor data from all sensors is used for modeling environmental conditions, then the comprehensiveness of the model is improved, but irregularities in the available sensor data may compromise performance of the applied model
Solution Approach 1:
The patent performs preliminary data validation and quality assessment before incorporating sensor data into the model. Each sensor's data is pre-screened for irregularities, missing values, and anomalies. Sensors with persistent quality issues are identified and either corrected through interpolation or excluded from model training. This preliminary action ensures that only reliable data contributes to model comprehensiveness, preventing data irregularities from compromising model performance.
Solution Approach 2:
The patent implements a feedback mechanism where the model performance is continuously monitored based on validation data. When irregularities or anomalies are detected in sensor data that affect model predictions, the system automatically adjusts by downweighting or excluding problematic sensor inputs. This feedback loop maintains model reliability while preserving comprehensiveness by adaptively selecting the most trustworthy data sources.
3Reliability
If a large amount of sensor data is processed to define an accurate model, then the reliability of environmental condition prediction is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent segments the sensor data processing into multiple stages: initial data filtering, feature extraction, redundancy removal, and selective model training. Each stage processes only the essential aspects of the data needed for that particular objective. This segmentation allows the system to achieve high prediction reliability through comprehensive analysis while managing computational complexity by breaking down the processing into manageable, optimized stages.
4Measurement precision
If redundant sensor data is retained for model definition, then the accuracy of environmental condition modeling is improved, but the storage costs and processing overhead increase
Solution Approach 1:
The patent transforms redundant sensor data into compressed representations or aggregated features that retain the essential information needed for accurate modeling. Instead of storing and processing every individual redundant reading, the system converts multiple similar readings into summary statistics or representative values. This parameter change maintains modeling accuracy while significantly reducing storage requirements and processing overhead.
Data Source
Figure 1A~2
Figure 3
Figure 4
AI summary
According to an example embodiment, a lighting system (100, 100') is provided, the lighting system (100, 100') comprising a plurality of luminaires (120) arranged for illuminating one or more spaces and a plurality of sensor units (140), each including respective one or more sensors arranged to observe respective environmental characteristics in their respective locations in said one or more spaces and an apparatus (102, 103) comprising means for performing the following: carry out a learning procedure to define a model of one or more aspects of environmental conditions in a portion of said one or more spaces based on respective time portions of a respective plurality of time series of sensor values originating from a first plurality of said sensors; evaluate, for each of said first plurality of said sensors, a respective strength of a match between the time series of sensor values originating from the respective sensor and said model; select a second plurality of said sensors as a subset of the first plurality of said sensors based at least in part on the respective strengths of the matches between the time series of sensor values and the model; and apply said model to control one or more aspects of a building automation system based on the respective plurality of time series of sensor values received from the second plurality of said sensors after said time portion.