i4Q Infrastructure Monitoring (i4Q IM)

KNOW MORE

i4Q IM is a service for monitoring manufacturing lines, detecting imminent failures and providing alerts for failures in manufacturing processes.

Project: i4Q

Publisher: CERTH

Technology: Infrastructure Monitoring, Machine Learning, AI, Signal Processing

Industry: Industry

INTRODUCTION

i4Q IM Infrastructure Monitoring solution provides an ensemble of monitoring tools for smart manufacturing workload orchestration and predictive failure alerting, including monitoring the health of workloads and productively alerting and taking corrective actions when a predicted problem is detected.

i4Q IM supports industrial companies to reach autonomous operation in manufacturing environments. Specifically, i4QIM solution is a software toolkit which elaborates manufacturing data derived from multiple sources (sensors, other i4Q solutions). These data undergo processing to achieve synchronization, feature extraction, etc., in order to exploit the most critical information.

i4QIM solution predicts problems and failures in manufacturing processes either by applying rule-based techniques or by training/validating/testing machine learning models. If a problem arises alerts are provided through interfaces

FEATURES/BENEFITS

  • Features:
    • License
    • Full customisation
    • Training

 

  • Benefits:
    • Predictive Alerting
    • Failure Detection
    • Defect Elimination

TECHNICAL INFORMATION

Code

i4Q IM

Type

Process

Technology Topics

Artificial Intelligence, Machine Learning, Signal Processing, Time Series Analysis

Development

Python

RESOURCES

How it works

Component

Interact

Identifying technical relationships

Run in parallel

Process data from multiple analytical solutions

This component allows the i4Q IM 

to oversee other analytic solutions 

offering the ability of parallel 

exploitation of multiple analytical 

results.

Process data from multiple analytical solutions

Apply rules to identify harmful events.

Imminent failures in the production 

line can be identified through the 

analysis of various analytical results 

by applying certain rules

Apply rules to identify harmful events.

Trigger alert. 

Upon the detection/prediction of a 

failure using a set of rules, an alert is 

generated to mobilize other 

solutions in taking corrective 

measures

Data split.

Train and validation.

The historical data are spit into training and validation sets to enable the training and validation of the machine learning models

Train and validation. 

Application of the model.

Before the model deployment, the 

training and validation phase in 

crucial to obtain a well-tuned model, 

capable of accurately 

detecting/predicting machine 

component failures and degradation

Application of the model

Trigger alert. 

Alerts are initiated when a machine 

failure is detected/predicted by the 

employed algorithms, to inform 

other solutions about the need to 

take corrective actions.

Apply rules.

Problem detection.

Harmful events can be identified 

with the application of a rule-based 

system utilizing both the sensor data 

and the ML model results.

Problem detection.

Trigger alert.

Machine operators as well as other 

analytical solutions are notified 

about a detected failure via the 

initiation of alerts

USE CASE

SERVICES LINKED

PUBLISHERS

CERTH is one of the leading research centres in Greece & listed among the TOP-15 E.U. institutions with the highest participation in competitive research grants