The windmill start process of a missile turbojet engine was simulated and also modeled before ignition using radial basis function networks(RBFN).Under the condition of few sample data,some experiential relations were introduced to reduce the input variable numbers.Compressor characteristic maps were generated with moving least square based on experimental data.Start process after ignition was modeled and the model was solved by particle swarm optimization(PSO) algorithm.PSO can avoid the problem of divergence for nonlinear equations caused by Newton-Raphson(N-R) method when the initial condition is far from real solution.Calculated results show a great agreement with the test data.The presented method provides some reference for the performance analysis and optimization of turbo engine windmill start process.