Some screen dumps from our latest version Pythia 2.x (Sneak preview)
The entry screen:
So far we are able to read sensor data from .txt and .csv type files and present the results of the analysis in a variety of combinations.
Pythia calculates the values histogram and stability pie on the fly when the timeseries for a sensor is read from the input file.
Below is an example of the overview graphs for a particular set of sensors with data from an experimental cooling installation; in this case the “CO2HydrateSlurry Pressure/Flow/Mass Correlation” which consist of
the combination for Mass, Frost, Temperature(Tank) and Pressure(Tank). The data is logged by a SCADA system and from there exported as a CSV file. This file is then imported into Pythia.
Zooming in on the individual sensor readings will display, for example, the analysis for the sensor “Pressure Tank”:
It’s all about statistics!
Pythia uses various statistical algorithms to calculate and display the graph of a sensor’s measurements in a timeline.
Below Pythia shows a timeline when zoomed out:
And now when zoomed in for the timeframe around the hand-pointer above:
The statistical algorithm in Pythia ensures that stretching or shrinking the timeline will show the correct shape of the measurements, as are present for the respective selected time frame. These are recognizable in the date/time notations below the X-axis below the graph.
This feature will enable an analyst to examine the process in the finest detail at his/hers proposal, i.e. the smallest time slice that a SCADA system can provide for any given system.
Patterns.
Pythia is developed to find event patterns in a timeline.
To do so, it must know on what event this pattern is based, so an analyst has to point out the moment in time that an event of interest happened, and the amount of time that precedes this moment. Pythia will then create this pattern and stores it in its database.
In this example the analyst chooses for the “Paired Z-Test” algorithm to examine the timeline.
The “Match Event Threshold” variable (just to the left) will hold the dimensionless value for the likelihood that the analyst needs to know whether Pythia needs to show this as a viable pattern found when it searches the entire time line for it. This value can be anything between 0 (zero), which means 0% chance and 1 (one) which means 100% chance. Here it is set to 0.4
This is the graphical representation of the selected timeframe and the sensor’s measurements.
There is another algorithm attached to this sensor:
The “Value Interval” algorithm.
This algorithm does not require the selection of a particular time frame, it uses simply a high value and a low value as tolerances to examine the timeline for the occurrences of sensor measurements that over- or under shoot for these tolerances. This can then be examined in conjunction with the process stability as presented earlier.
In this example the tolerances are set to 0.5 °C and -0.4 °C, the temperature range in which the process will produce the wanted product.
Because any measurement can only between tolerances or not, setting the treshold has to be set either 0 (zero) or 1 (one) in order to show the analysis results.
Now the analyst can run the test and the PatternMatcher will find all patterns in this timeline that are connected to a specific event for both the algorithms and as a result of the value in the treshold that is set in the algorithm definition earlier:
And in more detail, zoom in:
(notice the gray and red pattern recognition results are displayed together)
These are the combined graphics in which the above one “E40 Frost” is a part:
By switching off some of the process parameters in the legenda at the right, the overall view can be showed in the graphs of interest in more detail:
The result of the “Pattern Matcher”.
The green stars represent the found matches of a previous pattern that is set by the analyst. In this example the pattern of interest is of a temperature nature, so these are chosen by checking the boxes in the legenda at the right.
A word about the pattern recognition algorithm:
As can be seen in the legenda, there are two kinds of results: “Result” and “Event” and the number of event-occurences. And in this example is there is the notation of the algorithm that is used to perform the matcher test.
Pythia has a dictionary of statistical algorithms from which an analyst can choose in order to get the desired kind of pattern test, some are discussed earlier in this document.
The needed algorithm can be chosen in the dropdown box called: “Match Algorithm”, in this case the “Welch T-Test”. The treshold is set to 0.95, hence the pattern name: “TemperatureTank TT61.07 Welch T-Test 95%”.
(The “TemperatureTank”is the sensor’s name and “TT61.07” is the sensor’s identification, usually used in an installation design and/or in a SCADA environment)
The “Match Event Treshold” contains the value on which the “Pattern Matcher” will discriminate on deciding whether a match is to be considered valid. It is the PROBABILITY factor for a pattern recognition.
Except the “Value Interval” algorithm, which only returns a “0”or a “1”, all algorithms return a value BETWEEN “0” and “1”. This means that Pythia can calculate the probability factor that an analyst sets in his examination. This allows to incorporate risk management in a process investigation.
Below is the screenshot of the pattern that is defined in this example:
Output from an analysis:
Below is an example of (possible) output that Pythia creates on the fly when a pattern matcher test is performed:
==================================================================================================================
==================== Conjuncture Analysis Report at 9/20/2017 11:40:13 AM ======================================
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 4 Event(s): CoolPressure Z_TestMatcherPT11.09 95%; Overall Probability: 99.98%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 9 Event(s): TemperatureTank TT61.07 Welch-T 95%; Overall Probability: 97.91%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 34 Event(s): E40 Frost Z_TestMatcher 75%; Overall Probability: 92.12%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 80 Event(s): E40 Frost Z_TestMatcher 50%; Overall Probability: 73.42%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 18 Event(s): E40 Frost Z_TestMatcher 95%; Overall Probability: 99.14%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 2 Event(s): CoolTemp TT11.08 Z_TestMatcher .01% Copy; Overall Probability: 55.00%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 2 Event(s): CoolTemp TT11.08 Z_TestMatcher 5% Copy; Overall Probability: 55.00%
==================================================================================================================
------------------------------------------------------------------------------------------------------------------
Summary: 1 Event(s): CoolTemp TT11.08 Z_TestMatcher 95%; Overall Probability: 100.00%
==================================================================================================================
Number of Sure Probability (= 100%) on CoolTemp TT11.08 Z_TestMatcher 95%: 8
------------------------------------------------------------------------------------------------------------------
Number of High Probability (= between 98% and 100%) on CoolTemp TT11.08 Z_TestMatcher 95%: 41
------------------------------------------------------------------------------------------------------------------
Number of Medium Probability (= between 96% and 98%) on CoolTemp TT11.08 Z_TestMatcher 95%: 5
------------------------------------------------------------------------------------------------------------------
Number of Low Probability (= under 96%) on CoolTemp TT11.08 Z_TestMatcher 95%: 2
==================================================================================================================
##01; SelectedObject: CoolPressure Z_TestMatcherPT11.09 95%
##02; SelectedObject: TemperatureTank TT61.07 Welch-T 95%
##03; SelectedObject: E40 Frost Z_TestMatcher 75%
##04; SelectedObject: E40 Frost Z_TestMatcher 50%
##05; SelectedObject: E40 Frost Z_TestMatcher 95%
##06; SelectedObject: CoolTemp TT11.08 Z_TestMatcher .01% Copy
##07; SelectedObject: CoolTemp TT11.08 Z_TestMatcher 5% Copy
##08; SelectedObject: CoolTemp TT11.08 Z_TestMatcher 95%
==================================================================================================================
##01; SelectedObject: @: 6/10/2017 4:52:00 PM +02:00; #2; ; High EventProbability: 098.3% on: TemperatureTank TT61.07 Welch-T 95%
##02; SelectedObject: @: 6/10/2017 4:46:36 PM +02:00; #2; ; High EventProbability: 098.7% on: TemperatureTank TT61.07 Welch-T 95%
##03; SelectedObject: @: 6/10/2017 4:43:56 PM +02:00; #2; Medium EventProbability: 096.0% on: TemperatureTank TT61.07 Welch-T 95%
##04; SelectedObject: @: 6/10/2017 4:11:16 PM +02:00; #5; ; High EventProbability: 098.6% on: E40 Frost Z_TestMatcher 95%
##05; SelectedObject: @: 6/10/2017 4:11:16 PM +02:00; #4; ; High EventProbability: 098.6% on: E40 Frost Z_TestMatcher 50%
##06; SelectedObject: @: 6/10/2017 4:11:16 PM +02:00; #3; ; High EventProbability: 098.6% on: E40 Frost Z_TestMatcher 75%
##07; SelectedObject: @: 6/10/2017 4:10:24 PM +02:00; #5; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 95%
##08; SelectedObject: @: 6/10/2017 4:10:24 PM +02:00; #4; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 50%
##09; SelectedObject: @: 6/10/2017 4:10:24 PM +02:00; #3; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 75%
##10; SelectedObject: @: 6/10/2017 4:08:38 PM +02:00; #5; ; High EventProbability: 099.7% on: E40 Frost Z_TestMatcher 95%
##11; SelectedObject: @: 6/10/2017 4:08:38 PM +02:00; #4; ; High EventProbability: 099.7% on: E40 Frost Z_TestMatcher 50%
##12; SelectedObject: @: 6/10/2017 4:08:38 PM +02:00; #3; ; High EventProbability: 099.7% on: E40 Frost Z_TestMatcher 75%
##13; SelectedObject: @: 6/10/2017 3:52:28 PM +02:00; #5; Medium EventProbability: 096.6% on: E40 Frost Z_TestMatcher 95%
##14; SelectedObject: @: 6/10/2017 3:52:28 PM +02:00; #4; Medium EventProbability: 096.6% on: E40 Frost Z_TestMatcher 50%
##15; SelectedObject: @: 6/10/2017 3:52:28 PM +02:00; #3; Medium EventProbability: 096.6% on: E40 Frost Z_TestMatcher 75%
##16; SelectedObject: @: 6/10/2017 3:45:56 PM +02:00; #5; ; High EventProbability: 099.0% on: E40 Frost Z_TestMatcher 95%
##17; SelectedObject: @: 6/10/2017 3:45:56 PM +02:00; #4; ; High EventProbability: 099.0% on: E40 Frost Z_TestMatcher 50%
##18; SelectedObject: @: 6/10/2017 3:45:56 PM +02:00; #3; ; High EventProbability: 099.0% on: E40 Frost Z_TestMatcher 75%
##19; SelectedObject: @: 6/10/2017 3:45:28 PM +02:00; #2;Low EventProbability: 095.8% on: TemperatureTank TT61.07 Welch-T 95%
##20; SelectedObject: @: 6/10/2017 3:44:30 PM +02:00; #5; ; High EventProbability: 099.2% on: E40 Frost Z_TestMatcher 95%
##21; SelectedObject: @: 6/10/2017 3:44:30 PM +02:00; #4; ; High EventProbability: 099.2% on: E40 Frost Z_TestMatcher 50%
##22; SelectedObject: @: 6/10/2017 3:44:30 PM +02:00; #3; ; High EventProbability: 099.2% on: E40 Frost Z_TestMatcher 75%
##23; SelectedObject: @: 6/10/2017 3:34:26 PM +02:00; #5; ; High EventProbability: 098.8% on: E40 Frost Z_TestMatcher 95%
##24; SelectedObject: @: 6/10/2017 3:34:26 PM +02:00; #4; ; High EventProbability: 098.8% on: E40 Frost Z_TestMatcher 50%
##25; SelectedObject: @: 6/10/2017 3:34:26 PM +02:00; #3; ; High EventProbability: 098.8% on: E40 Frost Z_TestMatcher 75%
##26; SelectedObject: @: 6/10/2017 3:33:56 PM +02:00; #2;Low EventProbability: 095.1% on: TemperatureTank TT61.07 Welch-T 95%
##27; SelectedObject: @: 6/10/2017 3:25:26 PM +02:00; #5; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 95%
##28; SelectedObject: @: 6/10/2017 3:25:26 PM +02:00; #4; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 50%
##29; SelectedObject: @: 6/10/2017 3:25:26 PM +02:00; #3; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 75%
##30; SelectedObject: @: 6/10/2017 3:25:16 PM +02:00; #2;Sure EventProbability: 100.0% on: TemperatureTank TT61.07 Welch-T 95%
##31; SelectedObject: @: 6/10/2017 3:24:26 PM +02:00; #5; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 95%
##32; SelectedObject: @: 6/10/2017 3:24:26 PM +02:00; #4; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 50%
##33; SelectedObject: @: 6/10/2017 3:24:26 PM +02:00; #3; ; High EventProbability: 099.8% on: E40 Frost Z_TestMatcher 75%
##34; SelectedObject: @: 6/10/2017 3:22:32 PM +02:00; #5;Sure EventProbability: 100.0% on: E40 Frost Z_TestMatcher 95%
##35; SelectedObject: @: 6/10/2017 3:22:32 PM +02:00; #4;Sure EventProbability: 100.0% on: E40 Frost Z_TestMatcher 50%
##36; SelectedObject: @: 6/10/2017 3:22:32 PM +02:00; #3;Sure EventProbability: 100.0% on: E40 Frost Z_TestMatcher 75%
##37; SelectedObject: @: 6/10/2017 3:21:10 PM +02:00; #5; ; High EventProbability: 098.1% on: E40 Frost Z_TestMatcher 95%
##38; SelectedObject: @: 6/10/2017 3:21:10 PM +02:00; #4; ; High EventProbability: 098.1% on: E40 Frost Z_TestMatcher 50%
##39; SelectedObject: @: 6/10/2017 3:21:10 PM +02:00; #3; ; High EventProbability: 098.1% on: E40 Frost Z_TestMatcher 75%
##40; SelectedObject: @: 6/10/2017 3:16:06 PM +02:00; #5; ; High EventProbability: 099.5% on: E40 Frost Z_TestMatcher 95%
##41; SelectedObject: @: 6/10/2017 3:16:06 PM +02:00; #4; ; High EventProbability: 099.5% on: E40 Frost Z_TestMatcher 50%
##42; SelectedObject: @: 6/10/2017 3:16:06 PM +02:00; #3; ; High EventProbability: 099.5% on: E40 Frost Z_TestMatcher 75%
##43; SelectedObject: @: 6/10/2017 2:20:14 PM +02:00; #5; ; High EventProbability: 098.5% on: E40 Frost Z_TestMatcher 95%
##44; SelectedObject: @: 6/10/2017 2:20:14 PM +02:00; #4; ; High EventProbability: 098.5% on: E40 Frost Z_TestMatcher 50%
##45; SelectedObject: @: 6/10/2017 2:20:14 PM +02:00; #3; ; High EventProbability: 098.5% on: E40 Frost Z_TestMatcher 75%
##46; SelectedObject: @: 6/10/2017 2:16:54 PM +02:00; #5; ; High EventProbability: 098.9% on: E40 Frost Z_TestMatcher 95%
##47; SelectedObject: @: 6/10/2017 2:16:54 PM +02:00; #4; ; High EventProbability: 098.9% on: E40 Frost Z_TestMatcher 50%
##48; SelectedObject: @: 6/10/2017 2:16:54 PM +02:00; #3; ; High EventProbability: 098.9% on: E40 Frost Z_TestMatcher 75%
##49; SelectedObject: @: 6/10/2017 12:58:46 PM +02:00; #2; Medium EventProbability: 097.3% on: TemperatureTank TT61.07 Welch-T 95%
##50; SelectedObject: @: 6/10/2017 1:43:56 PM +02:00; #5; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 95%
##51; SelectedObject: @: 6/10/2017 1:43:56 PM +02:00; #4; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 50%
##52; SelectedObject: @: 6/10/2017 1:43:56 PM +02:00; #3; ; High EventProbability: 099.1% on: E40 Frost Z_TestMatcher 75%
##53; SelectedObject: @: 6/10/2017 1:36:52 PM +02:00; #8;Sure EventProbability: 100.0% on: CoolTemp TT11.08 Z_TestMatcher 95%
##54; SelectedObject: @: 6/10/2017 1:36:52 PM +02:00; #7;Sure EventProbability: 100.0% on: CoolTemp TT11.08 Z_TestMatcher 5% Copy
##55; SelectedObject: @: 6/10/2017 1:36:52 PM +02:00; #6;Sure EventProbability: 100.0% on: CoolTemp TT11.08 Z_TestMatcher .01% Copy
##56; SelectedObject: @: 6/10/2017 1:36:26 PM +02:00; #1;Sure EventProbability: 100.0% on: CoolPressure Z_TestMatcherPT11.09 95%
==================================================================================================================
The process took 230946 milliseconds. DataPointsCounter: 100899
==================================================================================================================
Pythia can submit any kind of output to any kind of channel like SMS, e-mail, or whatever an organization may need.
External parameters.
Pythia can use any kind of external parameters that can, will or may influence any process, like atmospheric phenomena.
As an example, below is a graph with a zoomed part of the weather in the town Eelde (province Groningen). The data is derived directly from the Dutch weather agency KNMI and the data is directly converted an feeded into Pythia.
It is clear that in some circumstances, like in the water or energy branche, weather is an important part of process investigation.
As with the weather, Pythia is able to import any kind of external parameter(s) that can contribute to any investigation in any type of process.
In case of any questions or more explanation, please contact us. (See the contact page)