观测永久丢失下随机离散事件系统故障预测的验证算法

针对随机离散事件系统在故障预测时可能出现系统观测永久丢失,导致预测不准确的问题,提出一种观测永久丢失下故障预测验证的算法。首先对观测永久丢失的随机离散事件系统的U-可预测性进行了形式化。其次使用随机预测器构造了一个随机离散事件系统的U-预测器,实现了系统的故障预测。基于U-预测器,提出了随机离散事件系统U-可预测性的充分必要条件及验证算法,并且引入成对的方式,明显地改进了该验证算法的复杂度。仿真结果表明,该验证算法使得观测永久丢失下系统故障预测准确。最后,实例说明观测永久丢失下故障预测验证算法的应用。结果表明,该验证算法相比现有同类验证算法应用范围更广,验证结果更精确。

Verification algorithm for fault prediction of stochastic discrete-event systems under permanent loss of observations

Aiming at the problem of permanent loss of system observations during the fault prediction of stochastic discrete-event systems(SDESs), resulting in inaccurate predictions, this paper proposed an algorithm for verification of fault prediction under permanent loss of observations. Firstly, it formalized the notion of U-predictability of SDESs under permanent loss of observations. Secondly, it constructed a U-predictor from the given system using stochastic predictors to perform fault prediction of SDESs. Based on the U-predictor, it proposed a necessary and sufficient condition and an algorithm for verifying U-predictability of SDESs, and the introduction of a pairwise approach significantly improved the complexity of the verification algorithm. Simulation results show that the verification algorithm made the fault prediction of the system accurate under permanent loss of observations. Finally, an example illustrated the application of fault prediction verification algorithm under permanent loss of observations. Compared with the existing similar algorithm, the verification algorithm has a wider range of applications and the verification results are more accurate.