Cause Analysis

Cause analysis can be done through interpreting the effect that specific inputs have on the model output.

This is represented in the form of bar graphs, with each bar depicting the relative effect of the input on the model output. The cause indicators pinpoint the major causes of a process deviation and provide it in a form that is easy to understand and interpret. The figure below shows the knowledge extraction dialog with the causal analysis graph in the top right-hand corner.

If two rulers are selected the causal analysis is shown of over that area while with only one ruler will show the causal analysis on that point.

The Cause Analysis view

NOTE: The cause analysis view can also be launched from the main toolbar by clicking on the Causal Analysis button.

The cause graph will appear as shown in the Cause analysis graph above. The cause graph will change corresponding to the position of the slide rule on the model output graph, this can be done manually or automatically - when in playback mode the graph will update automatically as the data progresses past the slide rule. In the Cause analysis graph above the relative effect of all the inputs on the model output are shown. At this particular instant it is evident that the slag depth and feed power product significantly to the model deviation.

Auto Scaling

Auto scaling is by default enabled on the causal analysis graph. This can be disabled by unchecking the Auto Scale check box. When auto scaling is disabled the minimum and maximum values can be edited in the Min and Max fields correspondingly.