中文摘要：

弹药协调器定位故障是火炮弹药自动装填系统的典型故障,其动作可靠性受多个不确定性因素影响,显式的极限状态方程难以获得。为计算协调器动作可靠性的定量数值,对协调器进行了不确定性建模,并根据标准状态的测试数据对模型进行了修正,构建了基于径向基神经网络的代理模型以提高计算效率,将网络输出与原始样本数据进行对比。结果表明,径向基神经网络能够很好地拟合原始模型。对代理模型进行了Monte-Carlo仿真分析,求解了定位误差的概率分布,弹药协调器的动作可靠性估计值为92.48%。

英文摘要：

Positioning failure of shell transfer arm is a typical fault of automatic ammunition loading-system of howitzer.The action reliability depends on several uncertainty variables.The explicit limit state function is nearly unavailable.To calculate the quantitative action reliability, uncertainty model of shell transfer arm was built and modified by using the measured data to improve the simulation accuracy.RBF neural network was used as agent model to improve the simulation efficiency.The network outputs were compared with original data.The result shows that the RBF neural network is satisfactory for fitting purpose.Monte-Carlo simulation based on agent model was implemented.Probability distribution of positioning errors was obtained.The es-timated action reliability of shell transfer arm is 92.48%.