KNIME logo
Contact usDownload
Read time: 2 min

Anomaly detection for predictive maintenance

Predict when critical equipment parts go bad and prevent failures and downtime

January 11, 2024
Put simplyML 201 & AI
Anomaly-detection-predictive-maintenance
Stacked TrianglesPanel BG

Producers must be increasingly competitive to protect their market share. Strong competition means machinery must operate at peak performance for as long as possible and without interruption. Manufacturing plants are a complex array of components and automated systems finely tuned to work together. Critical parts are monitored for proper functioning, with sensors providing data at regular intervals.

In order for companies to ensure that maintenance occurs at exactly the right time, they have to know about impending issues far enough in advance in order to take action.

Anomaly detection prevents impending breakdowns

Manufacturing plants are a complex array of components and automated systems finely tuned to work together. Critical parts are monitored for proper functioning, with sensors providing data at regular intervals.

A predictive model can be trained on readings, taken while parts are functioning correctly. The model will be able to detect anomalous data and predict impending breakdowns.

The sensor data is read into a KNIME workflow, which is automatically executed daily on KNIME Business Hub. The model is applied and the deployment workflow calls PMML models which, in the case of an anomaly, assess whether a first or second level alert should be activated.

The first level alert is generated by applying prelearned PMML models. Next, those first level alerts are combined and analyzed to report the second level alert.

In the case of a first level alert the REST service enables the action - an email to be sent or even a siren to start.

If a critical anomaly is detected in the form of a second level alert, a system shutdown can be invoked.

Breakdowns predicted 10 weeks in advance

Production managers can more efficiently manage their manufacturing plant with anomaly detection thanks to:

  • Accurately predicted breakdowns up to ten weeks in advance
  • Escalation of maintenance actions e.g. from email to system shutdown
  • Creation of an accurate, customized preventive maintenance program
  • Gradations in alert severity which are recognized and acted upon, based on how critical they are

A scalable anomaly detection solution for manufacturers

The anomaly detection solution deployed on KNIME Business Hub gives you the scalability you need thanks to KNIME Business Hub's cloud-native architecture. Vast computational resources are available to deploy predictive analytics on the huge amounts of sensor data typical in manufacturing.

Find out more about KNIME for manufacturing analytics.