A fault estimation algorithm based on iterative learning strategy is proposed for a class of continuous nonlinear systems with fault signals.The algorithm mainly adopts the rolling optimization idea in predictive control theory.Firstly，when the nonlinear system is subjected to bounded state interference and measurement interference，a fault tracking estimator is constructed using state errors and output residuals，and the differential signal of adjacent two output residuals is added on the iteration axis to obtain a virtual fault signal that approximates the actual fault signal.Secondly，in the sense of λ-norm，the convergence and complexity of output residuals and fault estimation errors are analyzed，and the convergence is judged through Gronwall inequality，which provides a sufficient condition convergence of the algorithm.Finally，through numerical simulations，the forms of common failure functions，and the comparison between the proposed methods with the P-type algorithm，the feasibility and effectiveness of the proposed algorithm are demonstrated.