====Creating Live Models==== In some cases, customers have existing models that describe how their system works - but they are usually offline models that are done by hand or in spreadsheets. With the rise of Machine-Learning, they may also have predictive, simulation or analytic models that can be used to improve how people operate equipment. If you want to make these models 'live' - or if you want to perform any other comparisons, calculations and modelling using your live process data - you can use [[model:start|ModelHost]]. ===Why Modelhost?=== ModelHost has a some key benefits... * Connects multiple different models using a single connection, reducing your need for ARDI user licenses, \\ * Allows individual models to share information so you can chain independent models together and perform meta-analytics, \\ * Allows easy goal-seeking and other complex analysis, which allows you to use your existing models in new ways and to answer questions that they weren't specifically designed to answer. \\ * The [[mos:welcome|Modular Output System]] allows you to export any new metrics - such as simulated values, KPIs, totals etc - to other systems or back into ARDI. \\ * Output data is automatically provided via OPC-UA, allowing it to be easily integrated into industrial systems. \\ ===Example Models=== Two models are included in the examples. The **math-based** model does some simple math calculations. The first is a calculation to predict failure probability for a singe turbine, the second uses a dynamic query to build a model output for each asset of a given type. The **ml-based** model uses a machine-learning model to simulate the behaviour of a subsystem. ===Accessing Example Model Data=== Model values are available via the **OPC-UA** protocol. We suggest a free UA-browser tool such as [[https://www.unified-automation.com/products/development-tools/uaexpert.html|UAExpert]] - follow [[Using UA Explorer|these instructions]] to see the data coming from the model. ===Continue Reading=== [[Triggering Code]]