Model hierarchy overview

The Hierarchy View shows the model structure of the NN block the application is currently connected to.

You can navigate between nodes of the Hierarchy tree. After you select a node, the other views are updated with corresponding information. The nodes that can appear in the Hierarchy View are:

• NN Block: The NN block being accessed. This is the root node for the tree.
• Models: This container node holds the child Model nodes. This node is a single instance child of the “NN block” node.
• Model: A model node can be either a:
  • Sensitivity Analysis Model Node indicated by a bar and line icon. The node was created by a Sensitivity Analysis. The model has not been trained yet.
  • Trained Model Node indicated by a neural network icon. The model has been trained. (The node was a Sensitivity Analysis model node that was transformed to a Trained Model node during training.) A Training Model node contains all the information to define an NN model and create an intelligent sensor.

    While only one or zero Sensitivity Analysis Model Nodes can exist at a time, any number of Trained Model Nodes can exist at one time. After selection of a Sensitivity Analysis Model node the sensitivity trends of all the inputs to the output appear in the Configuration View, and the status of the network is shown as Untrained. When a Trained Model Node is selected the sensitivity trends of all the inputs to the output appear in the Configuration View, the status of the network is shown as Trained, and basic information of the selected NN model, such as number of Inputs used, number of Hidden Neurons, Train and Test errors are displayed.

    The Model name is a string with the block’s path (NN module name, NN function block name), followed by the date of the model creation in yyyymmdd format, and the time of the model creation in 24-hour format. The Model nodes are children of the Models node.

• Sensitivity Analysis node for a specific input: Each sensitivity node contains the information to define the Sensitivity of the output to a specific input parameter. Sensitivity nodes are children of a Model node. If you are an Expert user, you can edit the input.
• Trained Model node for a specific input: Each sensitivity node contains the information to define the Sensitivity of the output to a specific input parameter. Sensitivity nodes are children of a Model node.