It is often a single connector that turns sensor readings into valuable insights

Boost IoT Platforms with Machine Intelligence: The PredictiveWorks. CrateDB Connector

Machine Data

Track & Collect marks the Beginning

Industry 4.0 is shaping the digital transformation in manufacturing. And data platforms get challenged by millions of machine generated data points. From industrial internet to connected cars and wearables, sensor readings will be everywhere and data platforms have to be prepared to make sense of large scale machine data.

Machine data originate from thousands of high frequency sensor readings and ship with individual message formats.

Message interoperability and the collection into one place for real-time visibility into power, temperature, pressure, speed and torque is of high priority. No doubt. But, for many IoT-scale data platforms, this is the only priority. Companies are left alone to gain higher valuable insights from their large scale machine data.

How do they extract trends and forecasts from sensor readings? What is the demand for electric power within the next 24 hours? Do we have to initiate an unscheduled maintenance? Are there indicators of abnormal machine behavior?

Answering this questions is vital. What, if one data connector is often enough to empower existing IoT-scale platforms with machine intelligence?

CrateDB

SQL Database for IoT-scale Workloads.

Connected factories, energy networks, smart city infrastructures, and vehicle fleets:

CrateDB is the SQL database of choice for this scenario.

CrateDB supports aggregation, filtering and grouping of massive amounts of data in real-time with ease.

CrateDB seamlessly integrates with Grafana for time series visualization and Tableau for more general analysis.

With an MQTT ingestion and a Postgres access, it can be put at the heart of an end-to-end IoT solution with ease.

With this in mind, companies hesitate to change horses although this IoT setup is not enough to respond to advanced Industry 4.0 demands.

CrateDB beyond SQL

Machine Intelligence for Machine Data

Google’s Cloud Data Fusion charged with 200+ plugins, with connectors to modern data sources and plugins for full-spectrum machine intelligence, offers a code-free approach to empower existing IoT data platforms with artificial intelligence fast:

For CrateDB setups, it is nothing more than selecting the right data connector and time series of huge amounts of machine data can be used to train accurate forecast models without writing a line of code.

Model building and deployment are based on the same technical foundation:

A data model that has been accepted for demand prediction, forecasting, trend analysis and other use cases is directly available and lo longer has to be reimplemented for production.

This architecture sketch illustrates how a popular IoT-scale setup with CrateDB can be boosted with code-free machine intelligence.

Caution: Time series analysis is an example and no limitation.

Leveraging the CrateDB data connector enables model building from huge amounts of historical sensor readings, persisted in CrateDB.

Machine intelligence results retrieved from applying these trained models to real-time sensor readings are sent back to CrateDB to benefit from existing visualization setups with Grafana or Tableau.

Conclusion

In Industry 4.0 and the networking and linking of many machines and other devices, large amounts of data are being created faster and faster. And this data is crucial to answering questions like: Will my machine break down tomorrow? Or are there irregularities in the data that might even be dangerous?

How can these and many other questions be answered? With machine intelligence.

PredictiveWorks. offers another awesome connector to the SQL database CrateDB. We charged Google’s Cloud Data Fusion besides 200+ plugins also with a connector to modern data sources and plugins to let our customers benefit from full-spectrum of machine intelligence.

It is nothing more than selecting the right data connector and time series of huge amounts of machine data can be used to train accurate forecast models. Without the need to write code.