The optical quantum-well structure consists of plugging one photonic crystal into another one with different energy band structure making the state of photon have special characters due to different arrangements of wells and barriers. This study discusses two-parameter one-dimensional photonic crystal structure and corresponding quantum-well structures and analyses the transmission spectra under axial stress. Optical fiber communication signal classification is to identify modulation style of signal with much noise. Wavelet transformation has a good localization characteristic in time-frequency domain, and the neural network has characteristics of selfstudy, self-adaptation, and high stabilization, which can improve the automatization and intelligence of recognition, so we combined the advantages of wavelet and neural network to identify the modulation styles of optical fiber communication signal in the paper. Firstly, we used the wavelet to decompose the signal, and then extracted the characteristics through the wavelet coefficient. Lastly we adopted the probabilistic neural networks to classify 4 kinds of common optical fiber communication signal. The simulation results indicate that the presented method performs well.