Predictive Maintenance is important to many organisations in the mining and manufacturing industries because it results in tangible operations improvement. Predictive Maintenance provides an additional method of monitoring and predicting system behaviour, leading to improved operating efficiency, reduced unplanned downtime, as well as lowered maintenance costs by limiting secondary damage. How one chooses to integrate a Predictive Maintenance model into their organisation is determined by how information about the system is to be communicated to maintenance engineers, operators, and plant managers. Common challenges surrounding how this is done include:
- I have multiple end users – plant managers, operations analysts, maintenance staff, etc.
- I have to allow access through different visual platforms e.g. Power BI, Tableau etc.
- I have to allow access through different data platforms e.g. OSISoft PI, SQL database etc.
- I need to scale to meet production needs
In this Predictive Maintenance series of posts, we considered the workflow depicted in Figure 1, for Predictive Maintenance.
Figure 1: Predictive Maintenance Workflow
In this article we take a look at how these models can be integrated into your organisation.
MATLAB is often considered a production-focused environment where models can be easily deployed to embedded devices, edge, enterprise systems, and the cloud. It provides an environment that is versatile and adaptable to integrate with various data sources, be it flat files, structured and unstructured databases, data historians, or data stored in a variety of cloud platforms. In addition to this it supports cross-platform deployment, which means models do not have to be manually recoded into other languages. Deployment options include C/C++ code, CUDA® code, enterprise IT systems, or the cloud supporting integration with web, database, and enterprise applications.
Figure 2: Integration Options
Figure 2 highlights possible forms of what success could look like for your organisation.
- Desktop Solution – dashboards displaying data which has already been generated and stored and running models on-demand to provide users with up-to-date data and insights.
- Compiled and Shared Solution – web apps or external desktop applications capable of being shared widely within your organisation
- Enterprise Integration Solution – setting up your predictive maintenance model in the cloud to analyse data from any edge device, in real-time, allowing integration with maintenance and service systems. Figure 3 illustrates some of the supported integration platforms.
Figure 3: MATLAB Production Server Architecture and Integration
In this case study we see how Opti-Num was able to help a large energy and chemical company to develop and integrate a predictive maintenance solution into their organisation in 100 hours. Leverage the expertise of Opti-Num consultants so you can get started quickly.
What Can I Do Next?
- Request a trial.
- Find out more from the team.
- Visit our Smart Mining & Manufacturing Focus Area page to learn more.
- Complete Predictive Maintenance Workflow in 100 Hours
- Baker Hughes Develops Predictive Maintenance Software for Gas and Oil Extraction Equipment Using Data Analytics and Machine Learning
- Mondi Implements Statistics-Based Health Monitoring and Predictive Maintenance for Manufacturing Processes with Machine Learning