Predictive maintenance platform for Gearboxes
We often hear the term Industry 4.0 as a necessary process for a more automated, efficient and sustainable future. In this post I would like to present a real example of an Industry 4.0 project we are working on.
At Magnet we help traditional industries digitize their activities so they can gain insights from data and automation. Recently we have built a gearbox monitoring platform called SmartG that is able to manage large amounts of data in real-time and my idea in this post is to summarize some key technical features that I find interesting.
The functional scope of SmartG covers monitoring, smart analytics, customized reports, alerts and warnings. SmartG basically gets data from the physical world via sensors which are sent to the software platform and is capable of handling large amounts of data to produce insights.
In summary, SmartG brings business value to many shareholders: high availability, better asset utilization, higher product quality and so on. My background is on computer programming so I will focus more on software-related aspects rather than mechanical ones, but if you are interested in reading more you can find more information about this project here and here.
Data transmission
We have chosen a publish/subscribe protocol called MQTT for bidirectional communication and for faster data transmission.
The MQTT server of our choice has been RabbitMQ for being lightweight and deployable on-premise and cloud, which is a critical requirement for this project.
The hardware device is sending us three JSON files, with different frequencies (1 second and 10 second) and 1000 records in each message.
Time-series database
We at Magnet love PostgreSQL, it’s our preferred database, so we felt we struck it lucky when we found Timescale, built on PostgreSQL and focused on time-series data; just what we needed. As I said, Timescale is PostgreSQL for time-series, so we have the best of both worlds: relational and time-series databases for the price of one. It’s super fast, scales horizontally and we can continue to write SQL queries! What’s not to like?
On premise vs cloud deployment
SmartG has been deployed in a company in the automotive sector and the final client’s preference was to deploy the solution on their on-premise environment.
Our HOP platform, which is the underlying technology stack behind SmartG, is ready for the challenge: we have created the Test environment on the Cloud (AWS) to accelerate development process, and the production environment on-premise, as requested by the client.
Dynamic dashboards
To fully leverage the value of data, clients need to experiment with it, they need to play with data creating customized panels by themselves without the help of programmers. Each user has different roles and interests so we need to be able to personalize the dashboard for them. In order to do this, we have seamlessly integrated Grafana and Highcharts in SmartG.
Reactive web user interfaces
The look and feel of the user interface is fast and interactive thanks to the underlying React-based Clojurescript library called re-frame.
Re-frame offers great productivity to produce sophisticated webapps with less lines of code and simpler models. Many companies have been using it successfully since 2015 and it allows us to use the same programming language (Clojure) in both the server and the client.
According to Hermann, Pentek, Otto (2016), there are 4 design principles and goals that are integral to Industry 4.0, namely Interconnection, Information transparency, Technical assistance and Decentralized decisions.
These goals are tackled with different technological approaches in each project, depending on clients’ needs and resources. In fact, we have used different tools in other Industry 4.0 projects we are building, to achieve similar concepts.
After an intense 6-month work and collaboration, SmartG is up and running with an on-premise deployment; another small step for the digitalization of traditional manufacturing and industry.