Aggregation Layer Analytics for Low-Connectivity Edge Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current industrial systems lack the capability for real-time predictive and prescriptive analytics close to the point of operation, especially in settings with limited network connectivity and noisy sensor data, and are not effective in handling complex systems with multiple data sources.
Innovation Solution
A distributed analytics system with an aggregation layer subsystem and edge processing devices that perform predictive and prescriptive analytics using cause and effect modeling, capable of operating with limited connectivity and handling noisy data, providing control signals and estimates such as remaining operational life of components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is collected and analyzed through centralized network or cloud-based servers, then comprehensive data analysis capability is achieved, but network connectivity and computational resources are required continuously, increasing system complexity and latency
Solution Approach 1:
The patent segments the centralized data analysis system into distributed edge computing nodes deployed at multiple locations within the industrial system. Each edge node performs local data processing and analysis, eliminating the need for continuous centralized network connectivity while maintaining comprehensive data analysis capability across the distributed system.
Solution Approach 2:
The patent implements local data processing capabilities at edge computing nodes positioned close to data sources. This local quality approach enables real-time analytics at the source without requiring continuous network connectivity to centralized servers, reducing latency and network dependency while maintaining analysis effectiveness.
2Measurement precision
If big data models are trained and deployed at the data center level, then accurate predictive analytics are achieved, but the training and deployment process is detached from real-time system operation, increasing latency
Solution Approach 1:
The patent performs preliminary data processing, feature extraction, and model training preparation at edge computing nodes before data is transmitted to the data center. This preliminary action reduces the amount of data requiring centralized processing and enables faster model deployment that remains synchronized with real-time system operations.
Solution Approach 2:
The patent transitions from a single centralized dimension for model training to a multi-dimensional distributed architecture where preliminary processing occurs at the edge dimension and final model training/refinement occurs at the data center dimension. This dimensional change enables both high accuracy and real-time responsiveness by distributing computational tasks across different spatial and temporal dimensions.
3Productivity
If PLC-based real-time control systems are used, then automated control response is achieved, but predictive and prescriptive analytics capability is lost, reducing operational insight
Solution Approach 1:
The patent merges PLC-based real-time control functionality with edge computing capabilities that provide predictive and prescriptive analytics. The edge nodes execute control logic similar to PLCs while simultaneously performing advanced analytics, combining the advantages of automated real-time control with predictive insights without sacrificing either capability.
Data Source
AI summary
An aggregation layer subsystem, and method of operation thereof, for use with an architect subsystem and a plurality of edge processing devices in a distributed analytics system, wherein each edge processing device is adapted to monitor and control the operation of at least one monitored system according to a first analytic model, the aggregation layer subsystem comprising: a processor and memory, the memory containing instructions which, when executed by the processor, enables the aggregation layer subsystem to: receive a second analytic model from the architect subsystem, the second analytic model based on characteristics of at least one monitored system associated with at least one of the plurality of edge processing devices; receive monitored system information from each of the plurality of edge processing devices; and, provide control signals to the at least one monitored system, via one of the edge processing devices, according to the second analytic model in response to the monitored system information.