Due to the non-stability and non-linearity characteristic of missile equipment spare parts demand

the thought of wavelet analysis theory used on missile equipment spare parts demand forecasting.The best grading of its decomposition of wavelet was determined in terms of the collective evaluation index

and spare parts demand time series were decomposed into different scales in order to reduce the randomicity and volatility of original time series;The high frequency signals were forecasted with ARMA model optimized by the improved self-adaptive intergeneration projection genetic algorithm and damping least-squares method

and the low frequency was forecasted with GM(1

1) model;the respective forecast result were integrated to get the forecast value of the original time series.Through an experiment of missile equipment spare parts demand

the feasibility and effectiveness of this method was proven.