the mechanisms of why azeotropes form are described in general terms on wikipedia (https://en.wikipedia.org/wiki/Azeotrope).

in order to confirm experimentally that a mixture forms an azeotrope you need to do a compositional analysis of the vapor and liquid phases at the conditions of interest (e.g. constant temperature or pressure).

You want to confirm experimentally or theoretically?

The mixture A and B will form an azeotrope (say a VLE azeotrope) when the vapor and the liquid, in phase equilibrium, have the same composition.

You can see this, for prediction of azeotrope: http://pubs.acs.org/doi/pdf/10.1021/i160027a006

Or this, for formation of azeotropes: http://pubs.acs.org/doi/pdf/10.1021/ie50480a018

Or even this, for azeotrope separation: http://pubs.acs.org/doi/pdf/10.1021/ie500043c

I think Johannes is right

The American Chemical Society has published several books in its "Advances in Chemistry" Series - "Azeotropic Data" which may be available in university libraries.

The CRC Handbook of Chemistry and Physics has a partial list.

If your institution has access to a process simulator (most chemical engineering departments), then this can predict many.

There are a number of theoretical (calculation/computation) methods which predict from properties.  Leonardo (above) has given some, but others are easy enough to find (though not necessarily easy to use).

Azeotropes are quite predictable using the boiling points and chemical nature of the molecules. See: "CORRELATION OF AZEOTROPE DATA" by G. H. EDULJEE and K. K. TIWARI in Chem Eng. Sci, vol 31, p 535, 1976. Little cited, but very useful, IMHO. Check it out on Google Scholar.

I want to confirm it theoretically.  I want to know  at what composition will the mixture of A and B will form an azeotrope and what difference will it make in VLE data or TXY/PXY diagram.  How can I do it theoretically?

one of the method is as Martin told, use a process simulator, enter the components and generate the t-x-y plot ;this is perhaps the most easiest.

Another method can be using the basic equation of relative volatility=((Ya/Xa)/(Yb/Xb))=Ka/Kb.

step-1 modify the equation by using Yb=1-Ya, Xb=1-Xa.

step-2 select a suitable K-value method based upon the nature of the components you are using.

step-3 find the values of Ka and Kb using the decided method

here you might have to use  a bit of thermodynamics equations to calculate some parameters in these methods or can take help of internet

step-4 if relative volatility is coming approximately 1 your concentration is nearly an azeotropic composition

you can go through some links for guidance-

http://www.jmcampbell.com/tip-of-the-month/2006/10/guidelines-for-selecting-k-value-method/

http://www.jmcampbell.com/tip-of-the-month/2006/09/how-to-determine-k-values/

Thank you all for your valuable answers

Theoretically, you may employ Aspen PLUS to do that.

How I can calculate the azeotrope composition using MATLAB? Models? Criterion?