Abstract: The authors use MATLAB to build the neural network to do delineation of the potential seismic sources. There are many factors to be impacted on the delineation. The factors and the result are of high non-linear relationship. The authors take the Coastal Region of South China(CRSC) as study area, after the samples' checking up and then optimizing the network, simulating again. The results accord with the actual conditions. This show that the Radial Basis Probabilistic Neural Network(RBPNN) has the practicability in delineation of the potential seismic sources.

Key words: delineation of the potential seismic sources, Coastal Region of South China, non-linear relationship, RBPNN, MATLAB

摘要： 用MATLAB语言建立径向基概率神经网络来求解地震潜在震源区的划分问题，地震潜在震源区划分是地震危险性安全评价工作的重点，影响潜在震源区的客观因素与潜在震源区划分结果间是一种高度非线性关系，将实际问题分析为网络的模式识别，以华南沿海地区为例检验优化网络，结果较好地对应了中国地震烈度区划图（1990），该方法是对潜在震源区智能划分的一次有效尝试。

关键词: 潜在震源区划分, 华南沿海, 非线性关系, 径向基概率神经网络, MATLAB