To enhance the pre-warning applicability and accuracy of maritime traffic safety risk, a pre-warning system in restricted visibility weather of the risk was set up, and it was composed of the risk matrix knowledge base, traffic flow density prediction subsystem and visibility warning subsystem. By collecting large samples, the expert survey method was modified by using the fuzzy information distribution theory under the condition of incomplete information, and the maritime traffic risk matrix was determined. The traffic density was calculated by using the short-time prediction algorithm of traffic density based on the limit learning machine theory in the artificial neural network. The regional atmospheric model system was used to divide the visibility forecast data provided by the meteorological and marine forecasting departments into spatial-temporal fine meshes, and the visible distance was calculated. The system was used to predict the visibility distance and traffic flow density of the focused sea area with spatial grids of 2 n mile by 2 n mile and time step of 10 min, so as to verify the effectiveness of the system. Simulation result shows that at 12 time points in two different time periods, the prediction accuracy rates of visible distance are 75%, 75%, 80%, 75%, 80%, 75%, 75%, 75%, 80%, 80%, 80% and 75%. The prediction accuracy rates of corresponding traffic flow densities are up to 80%. Therefore, the forecast result is reliable, and the system can realize the visualization and intelligent monitoring of navigation risk in sea area in restricted visibility weather. 6 tabs, 8 figs, 31 refs. © 2018, Editorial Department of Journal of Traffic and Transportation Engineering. All right reserved.