中文摘要：

介绍了径向基概率神经网络（RBPNN）的一种启发式结构优化算法．该算法首先提出一种移动平均中心超球覆盖算法，并用于初选径向基概率神经网络的隐中心矢量，然后使用遗传算法进一步优选隐中心矢量，同时优化核函数控制参数．实验结果表明，该算法在实际应用中能够加快优化速度，降低计算复杂度，有效地简化RBPNN模型的结构．

英文摘要：

The paper introduces a heuristic structure optimization algorithm of the radial basis probabilistic neural network （RBPNN）. A moving medium centers covering hypersphere algorithm is proposed to initially select potential bidden centers vectors of the first bidden layer from all the training samples, then the genetic algorithms further prunes the structure of RBPNN, and determining the matching controlling parameters of kernel function of RBPNN. The experimental results showed that the algorithm can effectively quicken the optimization speed, decrease the computation complexity, and simplify the structure of the RBPNN.