Predictive maintenance is used to prevent unplanned downtime by leveraging advanced data capturing and analysis techniques.

The many benefits that arise from being able to predict failures and quality issues make predictive maintenance a leading use case of Industry 4.0.

Until now, operators and maintenance personnel have performed maintenance according to a preset schedule, also known as “preventive maintenance”. This method can consume unnecessary labor and material resources, and results in maintenance activity that only prevents about 50% of a factory’s failure events.

Join us in this 2-part article series covering predictive maintenance tools and use cases, including insight into PdM in both setup and deployment phases.


How does Predictive Maintenance work?

Part of the Industry 4.0 revolution, predictive maintenance makes use of a set of tools such as process modeling, data integration, machine learning, and insight visualization – to provide manufacturers with actionable insights about the health of their production assets and processes.

For a general idea of how this is achieved, this post will include a list of predictive maintenance tools, including the components of a typical predictive maintenance system.


Predictive Maintenance Tools – Setup

A predictive maintenance system is set up using a number of tools, each with its own defined functionality:


Function: Capture data in real time

The market for industrial sensors is constantly growing in variety and quality with a vast number of vendors offering sophisticated products at relatively low costs.

Some common parameters captured by industrial sensors include:

  • Vibration
  • Temperature
  • Pressure
  • Light
  • Water/lubricant quality
  • Chemical content
  • Liquid/solid levels

In manufacturing, data sent from these sensors can be used for predictive maintenance. Through the application of advanced analysis techniques such as machine learning and artificial neural networks, predictions are formed regarding Remaining Useful Life (RUL) and other asset health and performance-related KPIs.

The data captured by sensors can also be utilized in other manufacturing use cases such as:

Track & Trace – following raw materials, components, mobile equipment, and finished products throughout the factory/process.

Quality Assurance – testing using optical sensors and identifying equipment anomalies that will likely result in quality issues.

Inventory Management – monitoring and controlling the supply of spare parts and raw materials, preventing inventory surplus or shortages.


Data Communication and Gateway Devices

Function: Facilitate communication in the Industrial IoT network

Gateway Devices are intermediate connection points that connect controllers (PLCs, wireless devices etc.), and sensors to a computing platform – whether on-premise or on-cloud. Gateway devices also protect the IoT network and the data being transported, providing an additional layer of security.

IoT Communication Protocols form the language of the Industrial IoT network. Common protocols such as MQTT, AMQP and CoAP, make sure that devices across the network speak the same language (interoperability) while being light on resources such as power and memory consumption.


The Edge & Cloud Nodes

Function: Data aggregation and device management

Edge computing in the smart factory setting.
Edge nodes maintain the security of the industrial IoT network while reducing latency in communication.

Even a fairly simple production line can generate massive amounts of data which need to be captured, aggregated, normalized, and analyzed.

To secure the network, and to maintain a low level of data communication latency, edge nodes are deployed in close proximity to the production line. The edge nodes handle part of the analytical and processing workload. This also makes scaling the network much easier.


Industrial IoT Platform

Function: Modeling, simulation, data integration, and machine learning.

Building in-house solutions for predictive maintenance is expensive and heavy on resources, while many companies lack the relevant skills.

Industrial IoT platforms vary in the scope and quality of their features, but the more comprehensive platforms will usually offer the following capabilities:

  1. Solution Modeling – Creating a visual blueprint of the PdM solution that includes the various components within the context of your operation: physical equipment, sensors, communication protocols, data analytics, and dashboards.
  2. Simulator – A test environment to validate the functionality and cost of your PdM system before development.
  3. Data Integration – Gateways and connectors enabling streamlined IIoT development with connectivity to your OT and IT data sources.
  4. Predictive Analytics and Machine Learning – Insights from the platform are used by operational teams on the ground to uncover the root cause of manufacturing issues, maximize overall equipment effectiveness (OEE), and reduce unplanned downtime.

BONUS: Digital Twin – Leading IIoT platforms provide intuitive visualization of the condition of production lines in real time. Alerts and automated root cause analysis, powered by machine learning and AI, deliver deep insight into upcoming downtime and quality disruptions.


Read on to learn more in Part II
In Part II of this article, learn how predictive maintenance is deployed in the manufacturing setting, and how an industrial IoT platform is used as a Human-Machine Interface for maintaining control over asset and process health.