====Cognition File Format====

You can hand-make ARDI Cognition files to build your own custom AIs, or modify your existing files to manually adjust the AI.

The core AI configuration is stored in a **json** file in your ARDI site folder - on a Windows system this will often be **c:\Program Files (x86)\Optrix\ARDI\web\sites\default\ai**.

Each AI will also have a sub-folder in that same directory containing the compiled AI, training data and a [[training time file|file containing the training times]]. 

The AI configuration is made up of the following options...

^Option^Meaning^
|name|A human-readable name for the AI|
|aitype|The type of AI - currently MLP or OCSVM|
|resultformat|How to interpret the results - a **value**, a **percentage** or a **boolean**|
|server|The ARDI server host name|
|site|The ARDI server site|
|path|The full path to the folder where your AI is stored|
|inputs|A list of the different AI inputs|
|outputs|A list of the different AI outputs|

For each input and output, you should specify...

^Option^Meaning^
|name|A human-readable name for the input/output|
|ardiasset|The asset id of the point in ARDI|
|ardiproperty|The property id of the point in ARDI|
|min|The minimum value|
|max|The maximum value|

If an output is manually entered, set the //ardiasset// and //ardiproperty// values to 0.

If creating a **One-Class SVM** (OCSVM), you should //not// define any outputs.

<code>
{
        "name": "AI Name",
	"aitype": "MLP",
	"resultformat": "value",
	"server": "localhost",
	"site": "default",
	"inputs": [ {
		"name": "Wind Turbine #20 Voltage",
		"ardiasset": 230,
		"ardiproperty": 28,
		"min": 0,
		"max": 700
	   },{
		"name": "Wind Turbine #20 Temperature",
		"ardiasset": 230,
		"ardiproperty": 29,
		"min": 0,
		"max": 150
	  } ],
          "outputs": [{
		"name": "Wind Turbine #20 Power",
		"ardiasset": 230,
		"ardiproperty": 48,
		"min": 0,
		"max": 3200	}],
	"path": "Path to AI Folder"	
</code>

With the //json// file in place, all you need to do is create the [[training time file|file containing the training times]], and you're ready to [[running your AI|run your AI]].