Jump to content

Recommended Posts

  • Seeq Team
Posted

Hi Sam-

Doing this is possible using a combination of capsules and signal from conditions. Use the following steps to perform this calculation:
1. Create a condition for when your line is running. (If you have a tag that specifies Running vs. Not running, just use Value Search). 
Running.png

Screen Shot 2018-11-09 at 9.09.33 PM.png

2. Next, create Daily Capsules using Periodic Condition.
Days.png

Screen Shot 2018-11-09 at 9.09.59 PM.png

3. Now use Signal from Condition to calculate the total duration Running per day.
Running duration.png

Screen Shot 2018-11-09 at 9.10.21 PM.png
This will result in a new signal that reports the total duration Running per day.

Please let me know if you have any additional questions.

Posted

Trying to follow this but when I click on the screenshots It asks me to sign into Altassian and I have no account?   

BTW What's the link between Altassian and SeeQ - I see a few things like the SeeQ KB have altasian URL's? 

 

 

Guest Jon Peterson-Test
Posted

@carl.abz.uk Thanks for pointing the bad links out. The pngs @Lindsey.Wilcox pasted into her response were actually references to online content (as opposed to copying and up loading the png to the forum). Problem is the online content cannot be accessed from Seeq.org unless you had a Seeq login to Atlassian; I suspect to avoid cross site scripting (XSS). Images in Atlassian KB are actually links to the original image. 

I'll be posting an internal note for proper use. 

Thanks,

Jon 

Posted

So back to the topic at hand  - unlike Sam / Lyndsay  I'm trying to do the same thing but would like hours running per month over a year.  Often the system can run for many days with no problem so with value search I use a max capsule duration of 40 days - the end result is SLOW!   So i invert the logic and look for the number of hours the system is NOT flowing.  I create monthly capsules and do the same trick.  And I do indeed see a new (brown) signal of non running hour per month.  

End result looks good, I set the analysis range to 1 year - the trace follows up - but when I export to Excel SeeQ just sits there with the export wheel spinning, I gave up after 10 minutes

All I want is a table with somthign like Jan. X hours Feb. Y hours Mar. X hours etc etc.flow_yr.thumb.PNG.e793f3c75abeb482a5d4d01b73a4b546.PNG

But I suspect the problem is Seeq is trying to export everything?  My signal is generally captured 1/min so over a year I have 1440 * 365 = 525,600 points so it's quitea lot of data?  I get wanrings over the interpolation insterval, but then i can't change this - and indeed asI an showing step data I'm not sure  why there is interpolation anyway?flow_yr_err.PNG.609109edd3049b97e032c0a1439ae59c.PNG

During the import of my original signal is there a way to resample the data to say 5 min resolution rather than 1 to keep the volume manageable (if indded this is the problem)

Is there a way to only export one trace (i.e. the 12 values I want)?

Guest Jon Peterson
Posted

On the export, we export everything on the screen. Also, there's the option of the gridding of data:

chrome_2018-11-12_14-10-48.thumb.png.3a3692ba2751ddb70e9cd75357984ec8.png

Gridding is often useful (likely not in your case) to get many signals all on the same timestamp. The default is automatic where we choose a gridding for all tags that's roughly the sample spacing of the signal with the most data. I suspect that is what is taking so long. We always export all the data that is on the worksheet. In your case we are generating lots of samples; no doubt that's the performance issue. 

As I show above, just put the trace to export on the worksheet and then choose Ungridded Original time stamps; we know ungridded will only have one sample per month. I created the following chart from the export:

EXCEL_2018-11-12_14-13-18.thumb.png.df75c22a1913375ae00fb13313a44bea.png

Seeq Organizer Topics are a great way to produce the chart. I created the following and set it up to execute on the current year:

chrome_2018-11-12_14-24-39.thumb.png.4051e746527d4495d1028c6364cbe007.png

The max interpolation, as you noticed, is set automatically to be ~2x a month. That because any math we do on the month capsule requires us to look to the right and left of any capsule. We give the warning, because often math on that big of time period can take a lot of time. In the future we will calculate that warning and only bring it up if it is an issue. 

Posted

OK I got there in the end, turning off gridding and the export works fine.  Many thanks.  Now the follow up questions....

You say you export "everything on the screen" - by experimentation I think the fact a trace is visible or greyed  out (i.e..  not selected on the details pane) makes no difference to the export.    The only way I can find to export just the bar graph (12 points per year = quick!) without including the other data (~500k points/year = SLOW) is to delete the other traces?  

seeq_export_screens.thumb.JPG.b87e9147f37285caac7f6736146fee4c.JPG

I duplicated the page for this - see below - then deleted the original signal, capsules and conditions to leave just the bar graph, but this seems a long winded way of doing it.  Is there no way in Export  on my first page to select what I want to export?

Also  it's a minor point but when moving the cursor the width of the bar chart seems to change the time the value is reported over - in the example there is only one value written on 31-Aug for the result at end of August - yet the cursor returns the resuls as "far to the right" as  4th September  but not form 5th September as by this point we are not on the width of the bar chart.

seeq_bar_chart_width.thumb.JPG.a4d2bd020196bf8632662671686faab4.JPG

 

 

 

Guest Jon Peterson
Posted

Hi Carl,

Yes, you got it correct. If it is on the screen in any way we export it. 

The bar chart is a visualization of the continuous signal--we interpolate between the samples; that being said, I consider this a bug. The interpolation can be a flat line (step), or interpolated, so, there can be "values" at the timestamps between samples. I experimented with setting the signal to discrete--the cursor should only show values a the time of the sample, but it extrapolates on the width of the column, which is arbitrary. I'd argue that moving to a bar chart could imply discrete; but certainly if discrete, it should do no interpolation. I'm going to file a bug on this. 

  • 11 months later...
Posted

Hi Atul,

as far as I know there is currently no way to import the data out of an Excel file in a Seeq native way. I would suggest contacting Seeq support regarding this topic.

However you can evaluate the following options:

- If you are able to transform the data into CSV format you can use the CSV import.

- Another possibility would be using the way that Brian described here (with using Excel instead of Access):

- If you have access to Seeq Data Labs you could use Python and Pandas to read the data from Excel and write it back to Seeq (https://datatofish.com/read_excel/).

- Using Seeq Connector SDK to implement your own connector or using Seeq Server SDK/REST API to push data to Seeq

Regards,

Thorsten

  • Amanda Chng featured and unfeatured this topic

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...