nonlinear curve fitting for a lead acid battery PLZ HELP

[altenbach]

I don't think this is correct. I think time is related to the line number and not contained in your data.

To me it seems you have 2D data (which is also easy to fit). E.g. Q(I, t). Can you clarify?

[lil_zu]

what do u mean by line number?

the relationship is Q= It but our original data contains I and Q so i found t  by Q/I since our fitting formula does not contain I and it should be

Q(T, q,c,k)=(qkcT)/(1-exp(-kT) + c(kT+exp(-kT) - 1))

ill also attach dataset with I but i dun thnk it is needed

[altenbach]

---

@lil_zu wrote:
the relationship is Q= It but our original data contains I and Q so i found t  by Q/I since our fitting formula does not contain I and it should be

---

So, since x and y (I and t) are linearly related, we get four linear curves, each with a different slope. This cannot be fit to a complicated equation. 😉

There has to be something else! Here's a picture of the 3D data that could be fit if we had a model of Q(I,#).

Message Edited by altenbach on 09-16-2007 12:07 PM

[lil_zu]

hmm but from the link i sent u earlier they managed to get the curves fitted and found the parameters.And it is a 2D. Q vs t..btw is it possible to attach ur VI ?

[lil_zu]

U could also refer to the following link for more theory if u have time.

www.solenergi.dk/rapporter/sec-r-12.pdf

especially on page no 49

Thx mate 🙂

[lil_zu]

sorry about double post but.for now just ignore the dataset i provided i would appreciate if u just assist me in builiding the VI as datasheet i can import later

[altenbach]

Here's what I did. I first use the model to simulate some data, the the fit succeeds well. (make sure that X cannot be zero).

(The result is only exact if there is no noise, of course)

Please double-ckeuch the model VI to make sure it is correct.

I still think you are misinterpreting your data.

(LabVIEW 8.0)

Message Edited by altenbach on 09-16-2007 03:36 PM

[lil_zu]

thx a lot for your great help man.. about the data set ill post a new one but not sure how it will turn out

but now it seems that we would have to change equation from

Q(T, q,c,k)=(qkcT)/(1-exp(-kT) + c(kT+exp(-kT) - 1))    to  ========>>>>  I(T, q,c,k)=(qkc)/(1-exp(-kT) + c(kT+exp(-kT) - 1)) to this i removed the T in numeator

[altenbach]

See, NOW things start to make sense. 🙂

Message Edited by altenbach on 09-17-2007 08:11 AM

[lil_zu]

hehe ya..thx a lot man 😄

i have few other questions as well.. but i would post them later.. is it ok i keep all my questions to this thread?