基于自回归模型的自适应阈值算法

提出一种基于自回归模型的自适应阈值检测算法用于监测液体火箭发动机地面试验的稳态过程,该算法由两个阶段组成。第一阶段的检测算法阈值由不同试车台次的稳态参数测量值的自回归模型的估计值、标准偏差和带宽系数计算,第二阶段的检测算法阈值由稳态过程的参数均值的自回归模型的估计值、标准偏差和带宽系数计算。用某大型液体火箭发动机的地面试验数据进行验证,结果表明该算法能有效可靠地监测发动机的运行状态。

Adaptive Threshold Algorithm Based on Autoregressive Model for Fault Detection of LRE in Ground Test

An adaptive threshold algorithm consisting of two stages based on autoregressive model was presented for monitoring the steady process of Liquid Propellant Rocket Engine(LRE) in ground test.For the first stage,the thresholds are calculated with the estimation values,standard deviation and band-width factors estimated by autoregressive model for steady parameter measurement values obtained from many different engine tests.Standard deviation and band-width factors which are trained by test data with the training algorithm are used at the beginning of the steady process of the engine test.For the second stage,the thresholds are calculated with the prediction values estimated using autoregressive model for steady parameter measurement mean values from many different engine tests.The standard deviation and band-width factors were used in the next steady process.The adaptive threshold algorithm based on autoregressive model was verified with the LRE ground test data.The results show that the approach can effectively and reliably monitor the LRE operation status.