Compression factor

Modified on 2020/12/17 14:03 by JEG Categorized as Definitions, How to
===How the compression factor works===
The compressionfactor for non-digital points is using a Swinging Door Trending(SDT) algorithm to determine which data points are saved. This means that instead of saving all data points to a more compact file format, only select data points are saved based on the compression factor.


The model works by taking the initial data point and making an additional two points(see pictures below). These two constructed points will be placed directly above and below the initial data point at a delta y distance equal to the compression factor. The angles(a and b in pictures below) are then calculated and added together between the initial data point, the two constructed points and the new data point. The lower constructed point will always make a line through the lowest value data point, and the upper constructed point will always go for the highest value data point. If the combined angles of a and b are at or above 180 degrees, then the last recorded data point will be saved. This can be recognized as when the constructed points form lines that will never meet no matter how far out they go(i.e. parallel or actively moving away from each other). For geometric reasons, the first new data point will always be rejected since it is simultaneously the highest and lowest value data point. After the first data point has been rejected, the next data point is evaluated using the same constructed points, but letting them split up to go through the highest and lowest data points recorded so far. This continues until a data point eventually causes the lines to never meet. This data point is then saved, and the process starts anew with this data point as the new initial data point, and new constructed points being made above and below it.


This process has been exemplified in the five pictures below, where the yellow diamond points are saved data points, the red crosses are discarded data points and the blue squares are the constructed points. In the first picture, the first data point is rejected due to angles a and b not being at or above a combined 180 degrees.


The second picture has a slightly lower data point than the first, so the lower constructed point makes a line through that, while the upper constructed continues its line through the now discarded first data point. Since a and b are still not above 180 degrees, the second data point is also discarded, and the third data point in the third picture is evaluated. This data point has a higher value than the first point, so the upper constructed point makes a line through that, while the lower constructed point retains the line through the second point. since a and b are still not above 180 degrees, the third data point is discarded as well, and a fourth data point is evaluated. This data point makes it so that a and b are above 180 degrees, so it is saved, and the process starts anew in picture five.


[image:{UP(Compression-factor)}Edit1.png|medium|left][image:{UP(Compression-factor)}Edit2.png|medium|left][image:{UP(Compression-factor)}Edit3.png|medium|left][image:{UP(Compression-factor)}Edit4.png|medium|left][image:{UP(Compression-factor)}Edit5.png|medium|left]


























===How to set your compression factor===
What compression factor(CF) to give points is entirely dependent upon expected point behaviour. If a point only has very small variations, then it will need a CF significantly lower than another point in the same system, which varies by a lot. The CF should therefore be decided on a point by point basis. Below is a table of some suggested CF values for certain levels of variation, examples of which can be seen in the pictures below the table.


An excel example sheet has also been included in the attachments, which allows the user to easily model different levels of variation and CF values to better reflect the system that they are sitting at. The input values to be changed are marked as red(initial value), yellow(Random number generator range), and blue(the compression factor). Multiple scenarios for each set of conditions can be checked by simply pressing F9 to update the excel sheet.


Note: if further clarifications are needed, JEG can be contacted.


{| border="1" cellpadding="1" cellspacing="1" style="width:500px;" 
|- 
| Typcial range of data variation from data point to data point
| Compression Factor
|- 
| 0.005
| 0.001
|- 
| 0.01
| 0.002
|- 
| 0.05
| 0.01
|- 
| 0.1
| 0.02
|- 
| 0.2
| 0.05
|- 
| 0.5
| 0.1
|- 
| 1.0
| 0.2
|}


[image:{UP(Compression-factor)}RNG1.png|medium|left|Random variation range 1 with CF 0.2][image:{UP(Compression-factor)}RNG02.png|medium|left|Random variation range 0.2 with CF 0.05][image:{UP(Compression-factor)}RNG01.png|medium|left|Random variation range 0.1 with CF 0.02][image:{UP(Compression-factor)}RNG001.png|medium|left|Random variation range 0.01 with CF 0.002]
Meta Keywords:
Meta Description:
Change Comment:
FLSmidth QCX wiki

Compression factor
