Due to the relatively large error in the wind power forecast，the uncertainty message of forecast is more useful as a guide for the wind farm and the system operator.Based on the analysis on the short term and ultra-short term wind power forecast error distribution from small to mid scale wind farm，a piecewise exponential distribution model for wind power forecast error is put forward.The probability density function and the cumulative probability function of piecewise exponential distribution are introduced to estimate parameters of the model using nonlinear least square method.In the example analysis，the piecewise exponential model is compared with traditional error models by precision indexes and the curves fitting effect.Results demonstrate the advantages of piecewise exponential distribution model in the forecast precision and flexibility.