Fourier transforms were introduced to convert aperiodic signals from time domain to frequency domain, in this way one can observe the spectra of the aperiodic signals. Discrete time fourier transform give the continuous spectrum of discrete time signal.
The real meaning of fourier transform is that to convert a signal which is in time domain into frequency domain and vise versa. The advantage to use fourier transform is there when a signal is in time domain of infinite length and you want to analyze its response for any given channel. Since the signal is of infinite time length so you convert it into frequency domain where it will be periodic for a frequency range (so you take its first part) and you also convert channel in freq domain and there you just convolute/ multiply (as required) with each other and easily gets the answer.
The book "Signals & Systems" by Oppenheim and Willsky has a four-page account on the historical development of the Fourier transforms, including why it was developed and what has been available already at these times. (IIRC the Fourier series was developed first.)
Fourier Transform may be seen as the further step after such mathematical tricks like Taylor series expansion or orthogonal polynomials. The task is to calculate, up to given precision, the values of a possibly very complicated function, with least effort. Taylor series is good only in a close vicinity of a given point, orthogonal polynomials are usable in wider domains but they fail to reproduce finite periodic signals (more precisely: even those with values contained in any finite section, periodic or not).
If you need infinite precision, then the correct Fourier series (discrete and finite) may become a Fourier Transform, also very helpful. Say you have to solve a difficult differential equation. The solution may be next to trivial for arbitrary Fourier component of a full solution (but inverse transform may be nevertheless hard again).
Physical significance: an X-ray picture of a single crystal is nothing else but the FT of its structure. The meaning of FT in quantum mechanics is more subtle, but priceless.
For over two centuries, Baron Jean Baptiste Joseph Fourier (03/21/1768 - 05/16/1830) found transform that bears his name.
But it has not been shown to be applicable to random signals. Hence the lack of interest in this discovery.
In 1929 Wiener shows that FT autocorrelation of a random signal is equal to the power spectral density. This means that the FT is applicable to random or reals signals. This is one of the most important theorem of signal processing.
In 1929 Wiener shows that FT autocorrelation of a random signal is equal to the power spectral density. This means that the FT is applicable to random or reals signals. This is one of the most important theorem of signal processing.