Device-Level Data Modeling for Faster Industrial AI Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Industrial automation systems face challenges in deriving value from large amounts of unstructured and uncorrelated industrial data, leading to inefficiencies in data processing and storage, as well as the identification of spurious correlations, which hinders the extraction of actionable insights.
Innovation Solution
A smart gateway platform that leverages domain expertise to selectively stream relevant industrial data to AI analytics, using smart objects with contextualization metadata to create device-level data models that constrain AI analytics, reducing data space and facilitating quicker derivation of insights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all industrial data is streamed to AI analytics systems for processing, then comprehensive analysis coverage is improved, but data processing time and storage costs increase
Solution Approach 1:
The patent extracts and streams only selected relevant data from industrial devices to AI analytics systems, rather than transmitting all generated data. This selective extraction approach maintains comprehensive analysis coverage for critical parameters while significantly reducing overall data processing time and storage requirements.
Solution Approach 2:
The patent segments industrial data into categories based on relevance to specific analytics objectives. By dividing data streams into relevant and irrelevant portions, the system processes only necessary segments through AI analytics, reducing processing time while maintaining analysis coverage for critical parameters.
2Measurement precision
If domain expertise is integrated into data selection, then identification of spurious correlations is reduced, but system complexity increases
Solution Approach 1:
The patent incorporates domain expertise during the preliminary system configuration phase to establish data selection criteria and analytics models. By pre-configuring relevant data streams and correlation rules based on domain knowledge, the system reduces spurious correlations without requiring complex real-time processing logic during operation.
Solution Approach 2:
The patent introduces an intermediary layer between industrial devices and AI analytics systems that applies domain expertise to filter and contextualize data. This intermediary component manages the complexity of domain knowledge application while presenting simplified, pre-processed data to the analytics engine.
3Measurement precision
If contextualization metadata is created for smart objects, then data relevance is improved, but data preparation time increases
Solution Approach 1:
The patent creates contextualization metadata for smart objects during the initial device configuration and data model establishment phase. By preparing this metadata in advance rather than in real-time, the system achieves high data relevance for analytics while minimizing the time overhead during actual data processing operations.
Data Source
Figure 1
Figure 2a
Figure 2b
AI summary
An industrial device supports device-level data modeling that pre-models data stored in the device with known relationships, correlations, key variable identifiers, and other such metadata to assist higher-level analytic systems to more quickly and accurately converge to actionable insights relative to a defined business or analytic objective. Data at the device level can be modeled according to modeling templates stored on the device that define relationships between items of device data for respective analytic goals (e.g., improvement of product quality, maximizing product throughput, optimizing energy consumption, etc.). This device-level modeling data can be exposed to higher level systems for creation of analytic models that can be used to analyze data from the industrial device relative to desired business objectives.