基于DE-RLSSVM算法的航空发动机传感器故障诊断

为了使约简最小二乘支持向量机(RLSSVM)具有更好的稀疏性和泛化能力,利用微分进化(DE)算法选择RLSSVM的支持向量,提出了DE-RLSSVM算法.在benchmark回归数据集上的仿真试验表明该算法具有很好的稀疏性和泛化能力.然后将该算法用于航空发动机传感器故障的诊断,提出了基于DERLSSVM算法的航空发动机传感器故障诊断方法.该方法利用DE-RLSSVM算法对传感器故障进行监测,然后进行定位和隔离.数字仿真结果表明该传感器故障诊断系统能够实现对航空发动机传感器硬故障的检测与隔离.

Sensor fault diagnosis of aero-engine based on DE-RLSSVM algorithm

In order to improve the generalization capability and sparsity of reduced least square support vector machine (RLSSVM), DE-RLSSVM was presented, whereby differential evolution (DE) algorithm was used to select the support vector of RLSSVM. Experiments on benchmark regression data sets show that the presented algorithm has good generalization and sparsity capabilities. Then, the presented algorithm was applied to sensor fault diagnosis of aero-engine. DE-RLSSVM algorithm was used to detected sensors faults, then to locate and isolate the faults. The simulations show that the sensor fault diagnosis system using the presented algorithm can detect and isolate hardware faults of sensor of aero-engine.