Hierarchical Event Estimation for Abnormality Data Transfer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing event estimation systems in manufacturing environments face challenges in real-time data transmission and accurate abnormality detection due to communication speed limitations between lower and upper controllers, leading to inefficient identification of events in control target devices.
Innovation Solution
An event estimation system comprising a lower controller that acquires operation information, estimates abnormalities using a neural network model, and transmits relevant data to an upper controller, which uses a larger neural network model to accurately estimate events based on the received information, allowing for efficient data processing and reduced computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the lower controller performs comprehensive abnormality detection and transmits all operation information to the upper controller, then the accuracy of event estimation is improved, but the communication load and transmission time are increased
Solution Approach 1:
The lower controller extracts only the necessary operation information related to abnormality detection and transmits it to the upper controller, rather than transmitting all operation information. This reduces communication load and transmission time while maintaining detection accuracy.
Solution Approach 2:
The system divides the abnormality detection function between the lower controller (initial detection and information extraction) and the upper controller (comprehensive event estimation). This segmentation allows each controller to perform its specialized function efficiently, reducing overall processing time.
2Measurement precision
If a large neural network model is used for event estimation, then the accuracy of event identification is improved, but the computational load and processing time are increased
Solution Approach 1:
The neural network processing is segmented between two controllers: the lower controller performs initial abnormality detection using a smaller model, and the upper controller performs comprehensive event estimation using a larger model. This distribution allows high-accuracy event identification while managing computational load across the system.
Solution Approach 2:
The lower controller performs preliminary abnormality detection and filters operation information before transmitting it to the upper controller. This preliminary action reduces the amount of data that requires intensive processing by the upper controller, thereby reducing overall computational load while maintaining accuracy.
3Speed
If the lower controller has high computational capabilities for real-time analysis, then the response speed is improved, but the device complexity and cost are increased
Solution Approach 1:
Computational capabilities are segmented between controllers based on their functions: the lower controller handles real-time operation information acquisition and initial abnormality detection with moderate computational requirements, while the upper controller handles comprehensive event estimation. This segmentation achieves responsive abnormality detection without requiring the lower controller to have excessive computational capabilities.
Data Source
AI summary
An event estimation system includes an upper device, and a lower controller device including first circuitry that acquires operation information of a control target device connected to the lower controller device, estimates a presence or absence of an abnormality based on the operation information, holds the operation information for a certain time period, and transmits, based on the presence or absence of an abnormality and to the upper device, the operation information related to the estimation of the presence or absence of the abnormality. The upper device has second circuitry that receives the operation information from the lower controller device, and operates according to the presence or absence of the abnormality, inputs, using an upper neural network model, the operation information, output event information, and estimates an event.