No abstract

Aircraft fault detection and diagnosis (FDD) is a key technology to ensure flight safety. However, multiple simultaneous fault detection is still a difficult problem. To deal with it, a new FDD approach is proposed in this article, in which the expected-mode augmentation (EMA) approach is applied for the variable-structure interacting multiple model (VSIMM) estimation. The EMA approach is used for online model set adaption. In the EMA–VSIMM FDD approach, the failure model set is composed of a basic fixed model set and an adaptive EMA set. The EMA set is generated based on the basic fixed model set according to their predicted probabilities. Single or multiple simultaneous faults can be detected and diagnosed by applying the EMA set properly. The simulation results show that the EMA–VSIMM method is effective in detecting faults of various types and estimating unknown fault parameters.

In this study, oil shale resources in the Middle Jurassic Shimengou Formation in the Tuanyushan area along the northern margin of the Qaidam Basin in Northwest China are evaluated. The total organic carbon (TOC) content of oil shale in the study area is positively correlated with resistivity and negatively correlated with acoustic travel time and natural gamma. Based on TOC and oil yield, as well as the log response differences between the resistivity logging curve, the natural gamma logging curve and the acoustic travel time logging curve of mudstone with differing TOC contents, ΔlogR and stepwise regression models can be created to quantitatively determine the TOC content of oil shale. The results of the two prediction models and the measured values show the mean absolute deviation (MAD) of the ΔlogR model to be 0.95 wt% and the coefficient of determination, R 2 , 0.67. The MAD of the stepwise regression model is 1.15 wt% and R 2 is 0.54. Of these two prediction models, the ΔlogR model has a higher recognition ability, but the predicted results are easily disturbed by silty mudstone. Therefore, the stepwise regression model is used to quantitatively identify oil shale resources in the study area. The comparison of the predicted and measured oil yields of oil shale from exploratory wells shows their R 2 to be 0.49. This suggests that in the whole study area, oil shale resources can be identified by the logging curve. Based on log interpretation, the volume method is used to estimate the area's oil shale resources. The results show the oil shale resources in this area to total 392 million tons and converted shale oil resources 24 million tons.