Aiming　to　solve　the　difficult　problem　of　detecting　the　incipient　faults　in　planetary　gearboxes,　a　comprehensive　methodology　that　combines　Teager　energy　operator　(TEO)　demodulation　and　stochastic　resonance　to　extract　fault　features　is　proposed.　Firstly,　the　vibration　signal　of　a　planetary　gearbox　is　decomposed　by　empirical　mode　decomposition　(EMD)　method　to　select　the　component　signal　containing　the　fault　information.　TEO　is　used　to　extract　the　demodulated　signal　from　the　component　signal.　Secondly,　the　demodulated　signal　is　sub-sampled　and　properly　compressed　to　meet　the　requirement　of　small　parameter　condition　of　the　stochastic　resonance　system.　Then,　particle　swarm　optimization　(PSO)　algorithm　is　further　used　to　optimize　the　system　parameters　of　the　stochastic　resonance　system　according　to　the　predefined　output　signal-noise　ratio　as　the　fitness　function.　Based　on　the　optimized　parameters,　the　stochastic　resonance　system　is　reconstructed　to　achieve　the　best　match　of　signal,　noise　and　nonlinear　system.　Finally,　the　signal　is　re-input　into　the　optimized　stochastic　resonance　system　to　realize　the　resonance　enhancement　extraction　of　fault　features.　The　simulation　results　and　the　experimental　tests　both　validate　that　the　proposed　methodology　can　achieve　large　signal-to-noise　ratio　output　of　weak　period　signals　with　intensive　noise,　providing　an　efficient　and　accurate　solution　for　fault　feature　extraction　of　planetary　gearboxes　with　incipient　damages.　©　2019,　Nanjing　Univ.　of　Aeronautics　an　Astronautics.　All　right　reserved.