Abstract: The kernel recursive least square(KRLS) identification theory and three typical algorithm,i.e.ALD-KRLS,SW-KRLS,FB-KRLS,is discussed.The KRLS algorithm and traditional RLS algorithm is compared for nonlinear system identification ability by the simulation example.The simulation results show that the KRLS identification method is obviously superior to the traditional RLS identification method whether in the identification accuracy,stability or anti-interference.However in the three KRLS identification method,based on the SW-KRLS method has the unique sliding window performance,the SW-KRLS method obtains better identification effect than the ALD-KRLS and FB-KRLS method.the SW-KRLS identification method is particularly suitable time-varying nonlinear system identification.