Special Control Charts - CUSUM |
|
|
CUSUM control chart and its efficient modification by Lucas (V-mask is not needed) is recommended mainly when a process mean shift needs to be detected fast. The method is based on summing differences from the goal cumulatively (Cumulative SUMs). Imbalance of differences from the goal (more differences of one sign) is quickly detected. Sensitivity of this chart is given by the k parameter, which can be inputted in the CUSUM chart dialog panel. When the data are normal and independent, the technique is much more efficient than the Shewhart X-bar or X-individual chart – number of data points necessary to detect 1 standard deviation departure from the baseline is about 10 times smaller. When a point falls outside the limits, it is taken as an alarm signal. When remedy steps are taken immediately after obtaining the signal, the FIR technique can be applied. The technique is useful as a quick check of whether the corrective action was successful. It is based on moving the point immediately following the remedy action just below/above the limit on the same side where the previous violation occurred. If the action was not successful, the limit is violated again.
Extended control charts - Pdf manual
Example
Data:
Sample |
388.2 |
394.3 |
400.8 |
400.5 |
394.3 |
402.1 |
396.7 |
... |
Output:
CUSUM Control Chart |
|
|
|
Task name : |
Sheet2 |
Data: |
All |
Selected columns : |
|
Sample |
|
|
|
Number of data : |
50 |
Central line : |
412,3969071 |
Target Std. deviation (S) : |
11,56188482 |
Sensitivity to shift (s) : |
1 |
Fast Initial Response: |
No |
Decission bounds +- : |
5 |
No. of excessions (+) : |
0 |
No. of excessions (-) : |
0 |
|