复杂装备多因素耦合安全性QHS-ESN度量

Metrics of complexity materiel system safety with multi-factor coupling based on QHS-ESN

针对装备安全事故耦合机理不明确、危险因素关联复杂的问题,提出场景分割耦合方法。将危险因素分割为危害故障、人为失误、致命环境、危险属性4个分量,从危险分量之间的非线性耦合关系拟合角度进行装备安全性度量;在此基础上,利用量子和声算法较强的全局寻优能力,构建一种新的量子和声搜索-回声状态网络(quantum harmony search echo state network,QHS-ESN)模型及其算法。并将其应用到某型飞机低空大表速飞行安全性度量中。仿真结果表明,该模型比原有的回声状态网络模型、和声神经网络模型在低空大表速飞行场景危险分量非线性耦合关系拟合上,兼顾拟合精度和稳定性能,具有更好的装备安全性度量效果。

To cope with the uncertain coupling mechanism and complicated hazard factors association of ma-teriel system accidents,a metrics method of system safety based on accident scenario division ＆ coupling is pro-vided.The hazard factors of the materiel system are divided into four variables ,i.e.hazardous fault,human er-ror,fatal environment and hazardous attribute.Then,the materiel safety metrics is equal to the nonlinear fitting of all the hazard variables.On the basis of this,a new quantum harmony search echo state network （QHS-ESN）model and its algorithm are proposed for the battleplan training safety metrics in subject of “high velocity at low altitude”,in which the quantum harmony search （QHS）is merged together with the echo state network （ESN）.The proposed model yields a superior optimization performance over the QHS.Compared with the mod-els of original ESN and harmony search back propagation （HS-BP）,further test is done in the nonlinear fitting of battleplan accident scenario hazard variables.And simulation result shows that it has a better effect on mate-riel safety metrics,taking fitting accuracy and stability into account simultaneously.