
Probability and Statistics Basic Probability and Statistics Math Forum 
 LinkBack  Thread Tools  Display Modes 
October 6th, 2017, 12:20 PM  #1 
Newbie Joined: Oct 2017 From: sweden Posts: 3 Thanks: 0  Real life process industry weighing "issue"(Statistics)
I'm doing some programming in process industry. Imagine sending powder to "mixing station". Powder has 38 ingredients that are weighed up in pretty simple linear process of ingredient 1+2..etc. Principle: [Compressor] (compressed air)> [sending station] (air)/(airpowder mix) > [mixer] > [filter] > (evacuation air) Problem is mostly getting good signal out of weighing the product [usually mixer, placed on 34 weighing cells]. Main factors are: 1. Mixer is throwing powder around. 2. Evacuating filters to avoid "powder dust" create underpressure that decreases overall weight. 3. That is balanced slightly by compressor pushing powder into mixer (changes with powder resistance). 4. Signal disturbances. [minimal] Let's ignore factors 24 & concentrate to mixer throwing powder around to simplify issue. My idea was to continuously record weighing values & use statistics to get as accurate result as possible. Let's say I record 10 values continuously. I thought to combine mean of these 10 numbers & median of most recent 3 to tilt the result towards "more updated recent" numbers. What is best way to get accurate & updated result when adding weight(product)? Last edited by skipjack; October 6th, 2017 at 01:15 PM. 
October 6th, 2017, 03:52 PM  #2 
Senior Member Joined: Sep 2015 From: USA Posts: 1,785 Thanks: 920 
I'd be happy to offer you my services as a systems engineer to help your company solve this problem. My rate is 200USD/hr plus all travel and hotel expenses and an agreed per diem which we can discuss. If you are interested you can contact me here to start the process. I'd love to visit Sweden! 
October 6th, 2017, 10:36 PM  #3 
Newbie Joined: Oct 2017 From: sweden Posts: 3 Thanks: 0 
I'm sorry, but you are missing my point here. The system is perfectly functional & running. Also client is more then happy with 15 kilo variation that is within error or margin. I am interested of purely mathematical solution to this real life problem. 
October 6th, 2017, 10:49 PM  #4  
Senior Member Joined: Sep 2015 From: USA Posts: 1,785 Thanks: 920  Quote:
the model that you choose is going to heavily depend on the real life characteristics of your powder thrower. It's impossible to say from here what those characteristics are. In other words it's a problem that only someone very familiar with your specific setup could begin to address. At best we can refer you to basic parameter estimation and it is quite likely that that's as far as you'll need to go. 2 questions 1) is the actual amount of powder expected to change while mixing? 2) is the mixing intensity constant?  
October 6th, 2017, 11:40 PM  #5  
Newbie Joined: Oct 2017 From: sweden Posts: 3 Thanks: 0 
Actually, really good & valid questions. Quote:
Yes, but amount of powder is changing. That will likely make it difficult to guess exactly how mixing motor influences measuring values. Last edited by skipjack; October 7th, 2017 at 12:45 AM.  

Tags 
industry, issuestatistics, life, process, real, weighing 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
A "tree of life" of mathematical knowledge?  questioner1  Math  2  May 28th, 2014 03:15 AM 
Issue with "stress" problem  Girn  Physics  2  October 12th, 2012 12:14 AM 
Book/article describing the process of "creating" the proof  honzik  Math Books  3  June 11th, 2012 12:49 PM 
"Half Life" of an object equation question.  FoScho  Algebra  3  November 30th, 2008 04:54 PM 
Issue with "stress" problem  Girn  Algebra  0  December 31st, 1969 04:00 PM 