The probability density estimation (pdf) has been studied since 1950s[1].The non-parametric density estimation is used in signal processing applications such as classification, compression[2], and also in process control[3].Recursive method for updating the pdfs of non-stationary processes using orthogonal wavelets have been introduced in [4].In [5] compactly supported multivariate polynomial (CSP) functions have been used to approximate the solution of reduced Fokker-Planck-Kolmogorov (FPK) equation to find the stationary pdf of stochastic nonlinear dynamical system and compared with the pdfof Parzens estimator[6].