how to compare the test result with standards?

[Thiyagu]

 Hi,

I am doing my  Post Graduation Final Year Project. 

Project Abstract:

Collect the Power generator output(Volatge and Current) and find out the different electrical Problems such like, Stator inturn failure, Rotor Failure.

I have collected the Data in time domain and converted in to Frequency domain. I am trying to compare the frequency domain data  with standard data values.

But  i am not able to compare the both the data and find out the Problem. I have tried this using Limit testing. but i am not get the exact result.

So please support to solve the issue.

Thanks

Thiyagi..

[billko]

---

@Thiyagu wrote:

 Hi,

I am doing my  Post Graduation Final Year Project. 

Project Abstract:

Collect the Power generator output(Volatge and Current) and find out the different electrical Problems such like, Stator inturn failure, Rotor Failure.

I have collected the Data in time domain and converted in to Frequency domain. I am trying to compare the frequency domain data  with standard data values.

But  i am not able to compare the both the data and find out the Problem. I have tried this using Limit testing. but i am not get the exact result.

So please support to solve the issue.

Thanks

Thiyagi..

---

What are "standard data values"?  And what is the "exact result" you are looking for?  What is the "inexact result"?  What does your code look like?

[Thiyagu]

Hi,

Thanks for your immediate reply.

What are "standard data values"?

Standard Data values will vary depending upon my Device Rating. In the attahced file refer the OK graph.

what is the "exact result" you are looking for?

I need to compare the faulty device output  graph with OK Device graph output and identitfy the faults. 

Requirement:

i am in plan to identify the different electrical power transformer faults such like coil to coil, phase to phase and inter turn problems using good transformer result through graph comparsion. But currently i am calculating the average of the signal and identitfy the probelms. This method is not consistent. so i need your input to write the code as per the requirement.

Thanks

Thiyagi...

[Eric1977]

I believe billko was referring to posting your VI to the forum and not just a snapshot of the graphs.

[Thiyagu]

Hello billko, 

Here i have posted my VI.

In this VI, i didn't compare the standard result to tested result.

[Thiyagu]

Hello billko, 

Here i have posted my VI.

In this VI, i didn't compare the standard result with tested result.

[crossrulz]

Where's the limit testing you claimed you tried?  How do you plan to maintain the limits?

[BTC_Admin]

---

@Thiyagu wrote:

Hello billko, 

Here i have posted my VI.

In this VI, i didn't compare the standard result with tested result.

---

Are you trying to write LabVIEW code that programmatically detects good graphs vs bad ones by comparing it with a given graph? If that's the case, it involves more work than how you'd do it visually, you'll first need a defined set of rules to use as test criteria.

I would have suggested looking under Mathematics->Fitting->Polynomial Fit to start but seems like you are wanting to use some specific waveform characteristics as test criteria so you'd have to first figure out those criteria (you called it "standard data" in your question). Just looking at graph images I can't tell what exactly you are considering as passing criteria in that "OK Graph".

If you're looking to have at least some level of "Coil to Coil amplitude", you could use poly fit (as mentioned above) and then subtract it from original data to get those peaks (which you can compare against a desired minimum value), you would still need to do more tricks with math.

-BTC