Concerning the problem that the traditional wavelet neural network is difficult to set initial parameters and easily fall into local extremum, this paper proposes a wavelet functional link neural network (WFLN) based on adaptive chaotic particle swarm optimization (ACPSO). Firstly, wavelet neural network and random vector functional link net are combined to construct wavelet neural network and strengthen parallel computing ability. Secondly, chaos optimization factors and adaptive weight coefficient are introduced into particle swarm optimization to improve premature convergence of particle swarm, and a dynamic balance between global and local search capabilities can be realized. Finally, optimize the WFLN neural network using the ACPSO algorithm, a short-term wind power forecasting model is established.The experimental results show that the ACPSO-WFLN wind power prediction model can significantly reduce the number of hidden neurons and the amount of iterations compared with other networks, and the root mean squared error is reduced to 0.0723, which has a higher prediction accuracy.