ModelHost

ModelHost is a platform used to create or deploy models using live data.

The type of model is up to you, but they can range from the incredibly simple (such as re-scaling a number) to the extremely complex.

This makes it easy to not only get live model values, but use those values the way you want to.

• Performing simple on-line calculations,
• Calculating maximum possible speeds given current constraints,
• Finding the ideal speed for maximum efficiency,
• Predicting current values using indirect data*,
• Alerting of potential/upcoming failures*,
• Detecting unusual operating conditions and divergence,

* These solutions often use AI/ML models

Each model takes live data from ARDI (along with any fixed values you need to use) and ‘runs’ your model using up-to-the-moment data. Each model can then provide live results that are sent via a number of methods, such being readable via OPC-UA, Modbus/TCP, Prometheus, MQTT etc.

The model is only run when relevant changes in your live data are seen, which helps avoid excessive use of your computing resources.

Models can also be connected together, so that the output of one model can feed into another, to add multiple simple elements together to form very complex predictive and analytic systems.

Models can also be run in goal seeking mode, which lets you try to automatically find ideal configurations to achieve specific outcomes (ie. the ideal speed to run a production line).

ModelHost can also be deployed on remote servers, allowing you to run AI-heavy workloads on other hardware across the site, enterprise or cloud network.