Building Information Model Generation Using Multi-AI Data Merging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of generating an information model (IM) from building management system (MS) data is cumbersome, time-consuming, and prone to errors due to the vast differences among MS providers, leading to incorrect device connections and increased energy consumption.
Innovation Solution
A method utilizing two or more artificial intelligence (AI) processing components to process building management system data, combining their outputs to enhance the accuracy and quality of the IM data, leveraging large language models (LLM) for precise transformation of natural language into standardized IM data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single AI processing component is used to generate IM data from MS data, then the processing speed is maintained, but the accuracy and quality of the IM data deteriorate due to errors in device connections and representation
Solution Approach 1:
The patent combines multiple AI processing components (e.g., multiple large language models) to process the same MS data simultaneously. Each component generates an IM data set, and these results are merged through a combination process that aggregates the outputs. This merging approach improves the overall accuracy and quality of the generated IM data by leveraging the collective processing power of multiple AI components, thereby resolving the contradiction between maintaining processing simplicity and improving data accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the outputs from multiple AI processing components are compared and combined. The combination process analyzes the results from different components, identifies discrepancies, and generates a refined IM data set that incorporates the most accurate information from each component. This feedback loop ensures that errors in individual component outputs are corrected through comparison and aggregation, thereby improving measurement precision while managing system complexity through structured feedback processing.
2Reliability
If multiple AI processing components are used to process MS data, then the quality of IM data improves, but the processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by having multiple AI processing components process the MS data simultaneously and in parallel rather than sequentially. Each component begins processing immediately upon receiving the input data, generating IM data sets concurrently. This parallel processing approach reduces the overall processing time compared to sequential processing, while still benefiting from the improved reliability that comes with multiple independent processing components. The combination process then aggregates these concurrently generated results.
3Measurement precision
If manual analysis of MS data is performed to generate IM data, then the accuracy can be maintained, but the process becomes extremely cumbersome and time-consuming
Solution Approach 1:
The patent replaces the mechanical system of manual analysis with an automated system using multiple AI processing components. Instead of human experts manually analyzing MS data to generate IM data, the system employs large language models and other AI components that automatically process the input data, extract relevant information, and generate accurate IM data sets. This substitution maintains high accuracy levels while dramatically improving productivity and reducing the time required for IM data generation, thereby resolving the contradiction between maintaining precision and improving efficiency.
Data Source
Figure 1~4
Figure 5
AI summary
The disclosure relates to a method (100) for obtaining information model, IM, data for one or more buildings (10), the method (100) comprising: - obtaining building management system, MS, data of a building management system of the one or more buildings (10); - obtaining instruction data comprising one or more instructions for processing the MS data by an Al processing component (44) for obtaining the IM data; - providing the MS data and the instruction data for separate processing by each one of two or more Al processing components (44); - receiving two or more IM data sets from the processing by the Al processing components (44); and - obtaining the IM data based on the two or more IM data sets.