1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
|
/**
\page tutorial-spc Tutorial: Using Statistical Process Control to monitor your signal
\tableofcontents
\section tuto-spc-intro Introduction
Statistical Process Control (SPC) is defined as the use of statistical methods to monitor
if a signal is "in control".
In this tutorial, we will use a Statistical Process Control method to monitor if a
random signal following a normal distribution is "in control".
\subsection tuto-spc-intro-methods Available methods
The different methods available in ViSP aim at detecting if the mean of a signal is
changing, either due to an abrupt jump or due to a slow drift.
The different methods that are available are the following:
- *Exponentially Weighted Moving Average* (EWMA), implementent in the vpStatisticalTestEWMA class.
- *Hinkley's test*, implemented in the vpStatisticalTestHinkley class.
- *Mean Adjusted Cumulative Sum* (mean adjusted CUSUM), implemented in the vpStatisticalTestMeanAdjustedCUSUM class.
- *Shewhart's test*, implemented in the vpStatisticalTestShewhart class.
- *Sigma test*, implemented in the vpStatisticalTestSigma class.
We refer the reader to the documentation of each class to have more detailed information on
each method.
\section tuto-spc-tutorial Explanations about the tutorial
\subsection tuto-spc-tutorial-howtorun How to run the tutorial
To see the different options of the tutorial, please run the following commands:
```
$ cd $VISP_WS/visp-build/tutorial/mean-drift
$ ./tutorial-meandrift -h
```
If you run the program without argument, you should see something similar to the following image:
\image html img-tutorial-spc-run.jpg
A Gaussian signal of mean equal to 6 and of standard deviation equal to 2 is generated,
without any mean drift. The program tells you which method has been chosen in the
console, and which are its parameters. A monitoring loop stops once an alarm
is raised. When the alarm is raised, some information about the alarm and the
test signal(s) + limits of the SPC method are given. Press `Return` to leave the program.
\subsection tuto-spc-tutorial-explained Detailed explanations about the SPC tutorial
For this tutorial, we use the main program tutorial-meandrift.cpp .
It uses the following enumeration to let the user choose which SPC method to use to monitor
the signal:
\snippet tutorial-meandrift.cpp Enum_For_Test_Choice
The program arguments are parsed and fill the following structure that stores the SPC methods
parameters:
\snippet tutorial-meandrift.cpp Structure_Parameters
It is possible to choose to monitor only upward mean drifts, only downward mean drifts or both.
To do so, use the `--alarms` option with the name of the alarm(s) you want to monitor.
First, the plot that will show the data is created in the following section of code:
\snippet tutorial-meandrift.cpp Plot_Init
Then, the desired method is created in the following section of code, with the desired parameters:
\snippet tutorial-meandrift.cpp Test_Creat
Then, the method is filled with "in control" signals to initialize the expected mean and the standard deviation:
\snippet tutorial-meandrift.cpp Test_Init
Then, the monitoring loop is run, where the signal is randomly generated with a potential mean drift
(if desired). This random signal is then tested, and the loop is stopped if an alarm we
want to monitor is raised:
\snippet tutorial-meandrift.cpp Loop_Monitor
Finally, some information about why the loop was stopped is displayed in the console:
\snippet tutorial-meandrift.cpp Failure_Debrief
The program stops once the `Return` key is pressed.
*/
|
