Feature Value/Machine Learning Function; Function Overview; Frames And Subframes - Omron NX Series User Manual

3 Feature Value/Machine Learning Function

3-1

Function Overview

The Feature Value/Machine Learning Function determines whether equipment events occur in the run-
ning status data (frame variables, variable data, and subframe variables) of the equipment in the Data
Utilization Phase. It consists of the Feature Extraction Function and the Machine Learning Function.
The Feature Extraction Function calculates feature values
ing Function determines whether equipment events occur from the calculated feature values.
The equipment event determination results are output to the Time Series Database Function and to
the system-defined variables.
NX-series AI Controller: Host computer
NY-series AI Controller: Host computer or Windows on the NY-series
AI Controller
PLC Function Module
User program execution
User-defined variables
Frame variables
Variable data
Subframe variables
System-defined
variables
Equipment event
monitoring result
*1. The feature values refer to the quantified features and patterns that are generated from variable data within a
certain time range based on the statistics.
3-1-1

Frames and Subframes

A frame is the range in which feature values are calculated by the Feature Extraction Function. It is
also possible to identify a time range in which values are calculated within a frame, which is referred to
as a subframe. The subframe setting is optional.
When the frame changes, the Feature Extraction Function calculates feature values from the variable
data in the time frame specified with a subframe (or by a frame when subframe is not specified). Fea-
ture values are calculated for each variable data.
3-2
AI Operator
AI machine learning
model
AI Function Module
Feature Value/Machine
Learning Function
Feature
Extraction
Function
Calculation of
feature values
Feature value
Machine Learning
Function
Event
determination
Equipment event
NX/NY-series Artificial Intelligence Machine Automation Controller User's Manual (W594)
*1
from data. Meanwhile, the Machine Learn-
Time Series Storage
Database Function
File name:
TimeSeries
FTR-****.csv
Feature value
File name:
TimeSeries
AIS-****.csv
Equipment event
monitoring result
****:
Indicates the time