Abstract: To address the problems of inaccurate wind speed measurement and poor dynamic response of maximum power point tracking (MPPT) control in large-scale wind turbines, this paper proposes a joint optimization method for wind speed estimation and MPPT based on a back-propagation (BP) neural network. First, an improved particle swarm optimization (IPSO) algorithm is introduced to optimize the initial weights and thresholds of the BP neural network, constructing an IPSO-BP wind speed estimation model. Based on easily measurable parameters such as rotor speed and power, high-precision online estimation of the equivalent effective wind speed is achieved, with a root-mean-square error of 0.16?m/s and a latency below 0.01?s. Second, the estimated wind speed is used to generate the optimal rotor speed reference trajectory, and an MPPT controller is designed based on a neural network and robust integral of the sign of the error (NN/RISE) scheme. In this controller, the IPSO-BP model provides high-quality feedforward action, an online neural network compensates for residual unmodeled dynamics, and the RISE term suppresses approximation errors and external disturbances. The asymptotic stability of the closed-loop system is rigorously proved through Lyapunov theory. Simulation results demonstrate that under low-wind-speed turbulent conditions, the proposed method increases the average power coefficient to 0.476, reduces the rotor speed tracking error by 72.4% compared with the conventional optimal torque control, and lowers the low-speed shaft torque damage equivalent load by 35%. Moreover, it exhibits excellent robustness against parameter perturbations and sudden wind gusts.