Cancel
Showing results for 
Search instead for 
Did you mean: 

Simcenter Tecware: Spike Filter

Siemens Phenom Siemens Phenom
Siemens Phenom

(view in My Videos)

 

Have data with spikes or dropouts?  Facing the prospect of re-instrumenting and acquiring again?

 

Sometimes there can be dropouts or spikes in the signals measured by sensors due to intermittent cable connections, telemetry issues, or other problems. Rather than throwing out this data, it might be possible to make the data useable by removing these spikes.

 

Removing spikes and correcting dropouts is easy with Simcenter Tecware TimeEdit tool!

 

Getting Started

 

If you have TecWare, or have or Testlab tokens, it is possible to run the Tecware TimeEdit tool. Click here for instructions on accessing Tecware TimeEdit with Testlab tokens..

 

1. From Windows explorer, doubleclick on the time history to open it in Tecware TimeEdit.

 

doubleclick.pngDoubleclick on the Simcenter Testlab LDSF file from Windows explorer to start TimeEdit.

By default, a small part of the beginning of the time history is shown.

 

2. To see all the time data, zoom out with either the slider bar in the lower right corner OR use the Double Arrow button on the left side of the Time Edit display (CTRL-F).

 

time_edit_slider.pngTo view the entire time history, either use the slider bar on the bottom right, or click on the overview icon on the left.

There are two modes to remove spikes: manual editing of data points OR the spike filter.  These are descibed in the next sections.

 

Manual Editing

 

1. Choose the magnifying glass icon (left side of screen) to zoom into the area with spikes.

 

magnify_zoom.pngClick on the Freehand Zoom icon on the left side and then select the segment to view.

In the display, click once one each side of the event or events to be zoomed in upon.  This can be done multiple times if needed. To zoom back out use CTRL-F.

 

2. If zoomed in enough, the data points are marked by a ‘+’ sign indicating that they can be edited. At the top of the screen, click on the ‘Move Points’ icon.

 

edit_points3.pngUsing the Move Points icon, individual data points can be edited.

Click in the display where the new amplitude of a time history data point should be.  The old data trace will be in gray, the new data trace will be in black.

 

This process is manual, and requires each drop out or spike be addressed individually.  After modifying the data, and satisfied with the result, go to the end of the article to learn how to save the modified data.

 

The manual method is time consuming.  An alternative is to use the automated spike filter method, which is described in the next section.

 

Spike Filter

 

Tecware TimeEdit has the ability to automatically detect and remove spikes from a time data trace.

 

1. In the TimeEdit Menu, select “Methods -> Signal Modifications Methods -> Spike Removal”.

 

spike_removal2.pngChoose "Methods -> Signal Modification Methods -> Spike Removal"

2. Parameters for how the spikes are determined are set in the Spike Filter menu. Statistics are performed on small data blocks – both the mean and variance of these blocks are calculated and compared.

 

spike_parameters.pngIn the spike filter menu, the data points used in the mean and variance calculations are defined.

When the variance exceed a specific threshold, a peak is identified, and the data is replaced.  Press the Start button.

 

3. After the calculation is done, the original trace is shown in gray, while the new trace is shown in black.

 

zoom.pngThe original signal is in gray, while the modified signal is in black.

Parameters can be adjusted as needed to properly identify the spikes.

 

Saving Modified Signal

 

The modified data can be saved to a new LDSF file or other file format.

 

1. In the TimeEdit Menu, choose “File -> Save Signals As” to store the modified signal.

 

save4.pngChoose "File -> Save Signals As" and press continue to store the modified data.

Press the Continue button.

 

2. Give the file a new name. Select the File Type for the modified data (LDSF, RPC, etc).

 

save5.pngStore the data to a new name, with LDSF selected as the file type.Finished!

 

Questions?  Email peter.schaldenbrand@siemens.com or contact Siemens PLM GTAC support.

 

More Fatigue and Durability links: 

Simcenter Tecware Links:

Comments
Experimenter
Experimenter

interesting feature the spike filter.

 

How to tell if the spikes in the data are real or errors?  For example, the spike could be a dropout in the signal, but maybe it was due to driving over a sharp bump?

Siemens Phenom Siemens Phenom
Siemens Phenom

Telling the difference between a real spike (due to bumps in the road, severe usage) versus a "fake" spike (due to transducer intermittent cable connection, telemetry dropout) requires some engineering judgement.  It can be especially difficult to tell the difference on severe durability and fatigue tests, where the part being measured is subjected to severe impacts, etc.

 

Some considerations:

 

1.  All Channels? Do the spikes occur at the same time on all channels, or just on one or two channels?  For example, if a circuit box has several accelerometers on it, and has a drop test done, all the accelerometers should show a spike or impact at the same time on all channels.  Usually, if there is an intermittent cable connection, the odds that the connectors on accelerometers fail at the same time is small.  An intermittent connection usually will show up on one or two transducers, and the spikes will not be aligned with impacts or other events in the other channels.

 

2.  Repeats? If doing a durability test recording of the same course or road surface several times in a row, the bumps should appear at the same spot each time.  Three repeats of the same test could be overlaid to see if the events occur at the same spot in all three recordings.  If they do occur, there is a higher chance that the events are real, and not fake. Having some knowledge of the physical test and where/when real events occur helps.  Also taking care in aligning the start times of the test repeats is important. 

 

3.  Physically impossible?  Knowing the test limits, for example that a displacement of 5 meters is not physically possible, helps in identifying fake spikes.  If the test object moves only 5 millimeters (which can be seen with your eye), but the transducer moves 5 meters, the spike is not real.  This could be a calibration issue, but could also indicate some sort of dropout or other issue.

 

Hope that helps!

Contributors