Anomaly detect
The Anomaly detect channel generates an output marking possible anomalies in transducer or computed channel data. The channel continuously tracks the selected parameters and outputs a status for each specified window. Use with a Bitmap trigger computed channel to generate triggers based on the anomaly detection.
NOTE
The Description is initialized on selection of the input channel.
- Window samples: Specify the number of input samples used to generate one output sample. This sets the size of the analysis window and also the associated output sample rate (i.e., the output sample rate is the input sample rate divided by this parameter). This can be any positive integer greater than one. The first output sample will have a time stamp of 0 in regard to elapsed run time. Because of this, the output data samples will appear to lead the input channel data samples when both are plotted against elapsed run time. A Time base shifter computed channel defined with a Lag of 1 sample can be used to change this if desired.
- Flat line detect
- On: Select to enable flat line detection.
- Range gate: Specify the gate for detecting a flat line anomaly. If the difference between the maximum and minimum data samples in the analysis window is less than the specified gate, the channel adds one (1) to the output.
- Drift detect
- On: Select to enable drift detection.
- Mean gate: Specify the gate for detecting a drift anomaly. Drift is measured against the reference mean value of the first analysis window of each test run. If the difference between current window mean and reference mean exceeds the specified gate, the channel adds two (2) to the output.
- Limit detect
- On: Select to enable limit detection.
- Minimum and Maximum: If any data sample in the analysis window is less than the specified minimum or greater than the specified maximum, the channel adds four (4) to the output.
- Kurtosis detect
- On: Select to enable kurtosis detection.
- Maximum: Specify the maximum kurtosis value. If the kurtosis coefficient for the data in the analysis window is greater than the specified maximum value, the channel adds eight (8) to the output. The following equation is used to calculate the kurtosis coefficient, K:
where, n is the number of data samples in the analysis window and
is the mean of the data samples in the analysis window.