摘要：本文首次引进经济预警思想对港口投资效益进行预测及评价研究。首先构建了港口投资效益预警指标体系，建立了基于RBF神经网络的港口投资效益预警模型；其次，采集九五到十五期间的各大沿海港口投资历史数据和投资效益情况对该模型进行了学习，测算了在不同扩散系数下的模型误差，结果表明当扩散系数为06时，模型的误差为9507%，说明该模型能够很好地对港口投资效益进行预测；最后，依据港口投资效益历史警情指标值设置了三个警度输出区间。

Abstract: The benefit of port investment is studied by introducing the economic early warning system. First, a benefit early warning index system is constructed; the model of early warning system for the port investment effect is built based on RBF neural network. Then this model is tested through historic figures of investment and benefits from year 1996 to 2005 on Chinese costal harbors. By testing model error throughdifferent spread coefficients, it turns out that when the spreadcoefficient value is defined as 0.6, the model error is 95.07%, under which the model can best predict the investment benefits of the harbors. At the end, the warning intelligence indicators are divided into three different early warning zones based on the historical data of port investment benefit.