Intelligent Asset Data Linking with Expert-Validated Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing software systems for industrial facilities lack a common identifier for linking asset data across different phases of a plant's lifecycle, requiring manual and ad-hoc linking that is cumbersome and not generally applicable.
Innovation Solution
A method and system for intelligent linking of asset data using a computer-implemented approach that generates linking recommendations based on asset data aspect similarities, allowing for semi-automatic or fully automated integration of asset data aspects into a combined representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual linking of asset data from different source systems is performed, then linking accuracy can be maintained through expert knowledge, but the effort and time required increases significantly
Solution Approach 1:
The patent introduces an intermediary system that acts as a mediator between manual expert linking and automated processing. This system uses machine learning models trained on expert-linked data to generate linking recommendations, which can then be reviewed and approved by experts. The intermediary captures expert knowledge in a reusable form that automates the linking process while maintaining accuracy through expert validation.
Solution Approach 2:
The system performs preliminary action by pre-training machine learning models on expert-linked asset data before actual linking tasks. This preliminary training phase captures expert knowledge and patterns in asset data relationships, enabling the system to generate accurate linking recommendations without requiring experts to manually link each new set of assets from scratch.
2Adaptability or versatility
If ad-hoc linking is performed for single tasks, then specific linking needs are met, but the linking relation is not generally applicable for different data tasks
Solution Approach 1:
The patent implements universality by creating a general-purpose asset data linking system that can handle multiple different data tasks and source systems. The machine learning models are trained on diverse asset data from various sources and are designed to generate linking recommendations that are generally applicable across different scenarios, tasks, and time periods, rather than being tailored to single specific use cases.
Solution Approach 2:
The system enables self-service by allowing the linking process to be automatically performed without requiring expert intervention for each linking task. The machine learning models autonomously generate linking recommendations by analyzing asset data properties and comparing them against learned patterns from training data, making the system self-sufficient for routine linking operations.
3Ease of operation
If common identifiers are shared across all software systems, then linking becomes trivial, but the flexibility to maintain system-specific data representations is lost
Solution Approach 1:
The patent introduces an intermediary layer that sits between source systems with their own identifiers and the need for integrated asset representation. This intermediary system uses machine learning to map and link identifiers from different source systems without requiring changes to the source systems themselves, preserving system specificity while enabling easy linking through the intermediary's intelligent matching capabilities.
Solution Approach 2:
The system segments the linking function into a separate, dedicated component that operates independently from source systems. This segmentation allows each source system to maintain its own data representations and identifiers while the intermediary segmentation handles the linking logic, enabling both system specificity and linking ease without requiring common identifiers across all systems.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
What is proposed is a method and respective system for intelligent linking of data associated with an asset. After receiving unlinked asset data sets of different asset data aspects from different source systems, linking recommendations are generated, which are based on asset data aspect similarities, which are calculated after an analysis of properties of the asset data aspects. Similar asset data aspects are grouped, and each asset data aspect of a group is assigned to a unique linked asset identifier thus producing a recommendation that maps each linked asset data aspect to all related asset data aspects. The generated linking recommendations are output for inspection based on expert knowledge, and for approval or modification of the linking recommendation by erasing or adding asset data aspects. In a last step a final linking result with linked asset identifiers is released as basis for generation of a combined asset representation for integration into a sink system.