Knowledge Fusion

Overview

The Knowledge Fusion step allows users to construct, configure and simulate an action model to realise the benefits estimated during the Benefit Estimation step.  Users perform "Knowledge Fusion" in "fusing" existing process knowledge with new knowledge gained from the Knowledge Extraction step in order to realise these benefits.  The underlying model that users configure is called a Fuzzy Logic model and takes the form of a rule-based approach to creating an action model for the process.  Users construct membership functions and draw up a list of rules in order to construct the action object.

The Knowledge Fusion step is accessed from the Troubleshooter project bar by clicking on Knowledge Fusion.  This option is only available after the modeling step has been completed.

The Knowledge Fusion view

Knowledge Fusion toolbar and buttons

  • Add Membership function

    Adds a new membership function to the selected variable.  The user has the option to specify the type, name and value ranges.

  • Reset Membership functions

    Reset the selected membership function to the default values and ranges.

  • Delete Membership function

    Deletes the selected membership function

  • Add rule

    Add a new rule using the rule editor

  • Edit rule

    Edit the selected rule

  • Delete rule(s)

    Delete the selected rule(s)

  • Delete all rules

    Deletes all rules.  (This option cannot be undone)

  • Export to FIS file

    Export the Fuzzy model from the Knowledge Fusion view.  The exported FIS file can be opened in the Fuzzy Editor, or imported into a Action Object blueprint by using the Fuzzy Logic Block in the Architect.

  • Reconfigure

    Reset the entire view and model by re-selecting inputs and outputs.  (Note, this will undo all current changes and configuration)

  • Multi-line rules

    Toggle view multi-line rules on and off.

<< Return to top

Field Selection

Upon accessing the Knowledge Fusion step for the first time in the project lifecycle, the user is presented with a field selection dialog.  From this dialog the user must configure the Input and Output fields that will be used in the action model.

  • Input fields

    These fields are typically the selected process targets for the exercise as defined during the Data Preparation step. These fields will be used to determine the state of the process in order to take the correct action(s) accordingly.

  • Output fields

    These fields are typically the selected process adjustables for the exercise as defined during the Data Preparation step.  These field values will be an output of the resulting action model and will contain the recommended setpoints for optimizing the process in order to obtain the benefits estimated during the Benefit Estimation step.

To configure the field selection dialog:

  1. Select the input fields from the list of available fields by using the add (>), add all (>>), remove (<) and remove all (<<) buttons.

  2. Select the output fields from the list of available fields by using the add (>), add all (>>), remove (<) and remove all (<<) buttons.

  3. Click OK.

<< Return to top

Membership functions

Overview of Membership functions

The Knowledge Fusion view is first displayed with a default set of membership functions for each selected input and output field.  Each field typically has 3 membership functions called Low, Medium and High.  The default ranges for these membership functions are determined by the limits set during the Visualization step of the troubleshooting project:

  • Low

    The default Low membership function is set to Trapesium and uses the ranges (LowLow - x), (LowLow), (Low) and ((Low + High)/2) where x is a small value to put the first and last parameters beyond the range of the LowLow and HighHigh limits.

  • Medium

    The default Medium membership function is set to Triangle and uses the ranges (Low), ((Low + High)/2) and (High)

  • High

    The default High membership function is set to Trapesium and uses the ranges ((Low + High)/2), (High), (HighHigh) and (High + x) where x is a small value to put the first and last parameters beyond the range of the LowLow and HighHigh limits.

Configuring membership functions

  • Names

    Membership function names can be edited directly on the list view by clicking on the membership function name.  The default recommended names are Low, Medium and High.

  • Types

    The membership function type is set on the list view by clicking on the type (Triangle or Trapesium) for the specific field and selecting the desired option from the drop-down.  Options currently include either Triangle or Trapesium.

    • Triangular: Triangular membership functions have three parameters or coordinates

    • Trapesium:  Trapesium membership functions have four parameters or coordinates

  • Ranges

    Membership function ranges can either be set visually or manually.

    • Manually:  

      1. Select the membership function from the list of membership functions.

      2. Click on the parameter cell you wish to edit.  

      3. Enter each parameter value manually.

    • Visually:

      1. Select the membership function on the list of membership functions.  

      2. The membership function is highlighted on the membership function graph.  

      3. Use the dotted vertical lines to set the ranges visually on the graph by clicking and dragging on the dotted vertical line indicating the selected membership function ranges.

<< Return to top


Related topics:

  

CSense 2023- Last updated: June 24,2025