A wavelet-based genetic algorithm using real-number coding and arithmetical crossover method in signal processing is described in this work. Due to the characteristic of the wavelet, an analytical signal can be represented by a finite linear combination of wavelet-based functions. Using a wavelet-based genetic algorithm to find the coefficients to such representation, an analytical signal can be reconstructed by the coefficients and the corresponding elementary function. Therefore the method can be used to compress and de-noise analytical signals because the insignificant information such as noise will not be reserved in the reconstructed signal. Both simulated signals and experimental multicomponent chromatograms are successfully compressed and de-noised with the proposed algorithm.