为预测察/打无人机导弹发射过程机身温度,提出一种基于PSO神经网络的预测算法。为了寻求RBF神经网络的最优结构,建立了开关型RBF神经网络,并采用PSO算法寻求开关型神经网络的开关值和网络参数。实验结果表明,该算法生成的虚拟函数能够较好的反映参数的内在联系,提高了数值仿真效率。预测结果对察/打无人机武器发射安全性论证有重要价值。

A method of PSO neutral network was proposed, by which the temperature of R/S UAV during missile launch process (MLP) could be forecasted. In order to find optimal structure of RBF neutral network, a switch RBF neural network was established, and the PSO algorithm was used to get the switch value and parameters of the switch RBF neural network. Training results showed that the dummy function from PSO algorithm would indicate the relationship between parameters of R/S UAV surface temperature well. The efficiency could be improved through MLP by PSO method.