低转速航空发动机滚动轴承故障深度异常检测方法

针对航空发动机滚动轴承在低转速状态下故障难检测的问题,提出了一种基于Transformer框架的深度支持向量描述方法用于检测低转速滚动轴承的故障。首先,构建了基于Transformer模型的振动特征提取主干网络。然后,将所提取的特征输入一个三层自编码器结构,用于计算网络模型的损失函数。为减少网络计算量,提高训练速度,在预处理中将滚动轴承的振动加速度时域信号通过快速傅里叶变换(FFT)得到的频谱结果作为网络的输入,且仅依靠正常数据完成模型的训练。最后,在带机匣的航空发动机转子试验器和某型真实的航空发动机上分别进行了试验验证。结果表明,所提方法能够准确的实现对低转速滚动轴承故障的检测,且检测精度分别为93%和100%,充分表明该方法具有很好的异常检测能力及应用价值。

泵出存储式测井深度异常现象分析

国内油田目前水平井、大斜度井和各类复杂井况较为普遍,常规电缆测井和湿接头钻具输送电缆测井在资料采集方面面临较大困难,泵出存储式测井技术的应用在解决此类测井施工难题中发挥了重要作用。与电缆测井直接记录测井深度不同的是:泵出存储式仪器是通过记录测量时间,通过时深转换来间接获取测井深度,在精度控制上还存在一些问题,本文以Y60型泵出存储式测井系统在复杂井况的应用为例,分析产生深度异常的原因。

基于半参数模型的海洋深度数据滤波

在传统海洋深度数据滤波的趋势面模型基础上,考虑到海底地形的变化的复杂性,对于那些海底地形变化较大,尤其是呈波浪形变化的海底地形,通常多项式模型不能对海底地形进行较好的拟合,存在有较大的模型误差。由于半参数模型可以较好地将参数值与模型误差进行分离,因此在趋势面模型的基础上引入非参数分量来表示趋势面模型中的模型误差,将常规的趋势面模型转化为半参数趋势面模型。并为了使所建立的模型具有抗差性,进一步引入半参数抗差趋势面模型,对于含有大量异常值的测深数据能够进行有效的滤波。

华南陆块均衡重力状态及其地质意义

重力均衡是一普遍的地球物理现象,可为区域深部构造、岩石圈形变、地壳结构以及应力场状态研究提供参考信息,是研究地壳内部结构和划分大地构造单元的重要方法之一。华南陆块是欧亚板块重要组成部分,也是我国矿产资源“大粮仓”,虽然对华南陆块的研究已持续近百年,但在构造、成矿等诸多地质问题上仍存在较大争议。为了解华南陆块均衡程度及其对构造、成矿的影响,本文尝试从均衡重力异常和均衡深度异常两个方面来探索。首先利用卫星布格重力计算得到了均衡剩余重力异常,然后利用高程和壳幔密度差计算了均衡深度异常。结果表明华南陆块大部分区域地壳处于均衡状态,相对正均衡异常主要位于东部沿海地区和武陵山一带,相对负均衡异常反映了秦岭—大别造山带、江南造山带及华南陆块西缘造山活动。地震多沿正、负均衡深度异常的转换带或梯度带发生,认为这些地带通常为深部构造转换部位,应当成为地震活动研究关注的重点区域。均衡剩余重力异常也间接约束了不同类型金属矿床分布,正均衡剩余重力异常区多分布地幔来源金属矿床,而负均衡异常区则多产出壳源相关金属矿床。

内蒙二连盆地北部重磁电资料综合解释预测含油有利区

针对中石化在内蒙二连盆地北部新区的勘探工作,收集、整理并连片处理了区内及外围航磁、重力与电法资料,在此基础上应用常规重磁数据处理技术和有特色的重磁异常优化滤波技术、航磁异常全梯度深度标定技术,结合其它物探资料和已知地质资料进行了综合解释并对成油远景进行了分析,提出了下一步勘探部署建议。进一步证明非地震方法在沉积盆地圈定和含油有利区带预测中具有良好的应用效果,也显示了老资料重新处理解释的不菲效益。

双频激电法在云南禄劝新发铜矿区的应用

以禄劝新发铜矿区为例,研究在山区运用双频激电法进行矿产远景调查的效果。结果表明:在交通不便、自然条件差的山区,双频激电法具有轻便、快速等优点。找矿效果理想,与地质、化探结果吻合,为双频激电法研究提供了一个可供参考的实例。

Identifying Anomaly Aircraft Trajectories in Terminal Areas Based on Deep Autoencoder and Its Application in Trajectory Clustering

Anomalous trajectory detection and traffic flow classification for complicated airspace are of vital importance to safety and efficiency analysis.Some researchers employed density-based unsupervised machine learning method to exploit these trajectories related to air traffic control(ATC)actions.However,the quality of position data and the tiny density difference between traffic flows in the terminal area make it particularly challenging.To alleviate these two challenges,this paper proposes a novel framework which combines robust deep auto-encoder(RDAE)model and density peak(DP)clustering algorithm.Specifically,the RDAE model is utilized to reconstruct denoising trajectory and identify anomaly trajectories in the terminal area by two different regularizations.Then,the nonlinear components captured by the encoder of RDAE are input in the DP algorithm to classify the global traffic flows.An experiment on a terminal airspace at Guangzhou Baiyun Airport(ZGGG)with anomaly label shows that the proposed combination can automatically capture non-conventional spatiotemporal traffic patterns in the aircraft movement.The superiority of RDAE and combination are also demonstrated by visualizing and quantitatively evaluating the experimental results.

Variations of SST and Thermocline Depth in the Tropical Indian Ocean During Indian Ocean Dipole Events

Interannual variations in the surface and subsurface tropical Indian Ocean were studied using HadlSST and SODA datasets. Wind and heat flux datasets were used to discuss the mechanisms for these variations. Our results indicate that the surface and subsurface variations of the tropical Indian Ocean during Indian Ocean Dipole （IOD） events are significantly different. A prominent characteristic of the eastern pole is the SSTA rebound after a cooling process, which does not take place at the subsurface layer. In the western pole, the surface anomalies last longer than the subsurface anomalies. The subsurface anomalies are strongly correlated with ENSO, while the relationship between the surface anomalies and ENSO is much weaker. And the subsurface anomalies of the two poles are negatively correlated while they are positively correlated at the surface layer. The wind and surface heat flux analysis suggests that the thermocline depth variations are mainly determined by wind stress fields, while the heat flux effect is important on SST.

Anomaly detection of earthquake precursor data using long short-term memory networks

Earthquake precursor data have been used as an important basis for earthquake prediction.In this study,a recurrent neural network(RNN)architecture with long short-term memory(LSTM)units is utilized to develop a predictive model for normal data.Furthermore,the prediction errors from the predictive models are used to indicate normal or abnormal behavior.An additional advantage of using the LSTM networks is that the earthquake precursor data can be directly fed into the network without any elaborate preprocessing as required by other approaches.Furthermore,no prior information on abnormal data is needed by these networks as they are trained only using normal data.Experiments using three groups of real data were conducted to compare the anomaly detection results of the proposed method with those of manual recognition.The comparison results indicated that the proposed LSTM network achieves promising results and is viable for detecting anomalies in earthquake precursor data.

Effi cient modeling of the gravity anomaly caused by a sedimentary basin with lateral variable density contrast and its application in basement relief estimation

The forward calculation of gravity anomalies is a non-negligible aspect contributing to the time consumption of the entire process of basement relief estimation.In this study,we develop a fast hybrid computing scheme to compute the gravity anomaly of a basement.We use the vertical prism source equation in a given region R centered at a certain gravity observation point and the vertical line source equation outside R to derive the gravity anomaly.We observe that the computation with the vertical line source equation is much faster than that of the vertical prism source equation,but the former is slightly inaccurate.Therefore,our method is highly effi cient and able to avoid the errors caused by the low accuracy of the vertical line source equation near the observation point.We then derive the general principle of choosing the size of R via a series of prism model tests.Our tests on the gravity anomaly over the Los Angeles Basin confirm the correctness of our proposed forward strategy.We modify Bott’s method with an accelerating factor to expedite the inversion procedure and presume that the density contrast between the sediments and the basement in a sedimentary basin varies laterally and can be obtained using the equivalent equation.Synthetic data and real data applications in the Weihe Basin illustrate that our proposed method can accurately and effi ciently estimate the basement relief of sedimentary basins.

Three-dimensional gravity inversion based on 3D U-Net++

The gravity inversion is to restore genetic density distribution of the underground target to be explored for explaining the internal structure and distribution of the Earth.In this paper,we propose a new 3D gravity inversion method based on 3D U-Net++.Compared with two-dimensional gravity inversion,three-dimensional(3D)gravity inversion can more precisely describe the density distribution of underground space.However,conventional 3D gravity inversion method input is two-dimensional,the input and output of the network proposed in our method are three-dimensional.In the training stage,we design a large number of diversifi ed simulation model-data pairs by using the random walk method to improve the generalization ability of the network.In the test phase,we verify the network performance by using the model-data pairs generated by the simulation.To further illustrate the eff ectiveness of the algorithm,we apply the method to the inversion of the San Nicolas mining area,and the inversion results are basically consistent with the borehole measurement results.Moreover,the results of the 3D U-Net++inversion and the 3D U-Net inversion are compared.The density models of the 3D U-Net++inversion have higher resolution,more concentrated inversion results,and a clearer boundary of the density model.

Optimal Precursor Perturbations of El Ni?o in the Zebiak-Cane Model for Different Cost Functions

Optimal precursor perturbations of El Nino in the Zebiak-Cane model were explored for three different cost functions. For the different characteristics of the eastern-Pacific （EP） El Nino and the central-Pacific （CP） El Nino, three cost functions were defined as the sea surface temperature anomaly （SSTA） evolutions at prediction time in the whole tropical Pacific, the Nino3 area, and the Nino4 area. For all three cost functions, there were two optimal precursors that developed into El Nino events, called Precursor Ⅰ and Precursor Ⅱ. For Precursor Ⅰ, the SSTA component consisted of an east-west （positive-negative） dipole spanning the entire tropical Pacific basin and the thermocline depth anomaly pattern exhibited a tendency of deepening for the whole of the equatorial Pacific. Precursor Ⅰ can develop into an EP-El Nino event, with the warmest SSTA occurring in the eastern tropical Pacific or into a mixed El Nino event that has features between EP-El Nino and CP-El Nino events. For Precursor Ⅱ, the thermocline deepened anomalously in the eastern equatorial Pacific and the amplitude of deepening was obviously larger than that of shoaling in the central and western equatorial Pacific. Precursor Ⅱ developed into a mixed El Nino event. Both the thermocline depth and wind anomaly played important roles in the development of Precursor Ⅰ and Precursor Ⅱ.

ADS-B Anomaly Data Detection Model Based on Deep Learning and Difference of Gaussian Approach

Due to the influence of terrain structure,meteorological conditions and various factors,there are anomalous data in automatic dependent surveillance-broadcast(ADS-B)message.The ADS-B equipment can be used for positioning of general aviation aircraft.Aim to acquire the accurate position information of aircraft and detect anomaly data,the ADS-B anomaly data detection model based on deep learning and difference of Gaussian(DoG)approach is proposed.First,according to the characteristic of ADS-B data,the ADS-B position data are transformed into the coordinate system.And the origin of the coordinate system is set up as the take-off point.Then,based on the kinematic principle,the ADS-B anomaly data can be removed.Moreover,the details of the ADS-B position data can be got by the DoG approach.Finally,the long short-term memory(LSTM)neural network is used to optimize the recurrent neural network(RNN)with severe gradient reduction for processing ADS-B data.The position data of ADS-B are reconstructed by the sequence to sequence(seq2seq)model which is composed of LSTM neural network,and the reconstruction error is used to detect the anomalous data.Based on the real flight data of general aviation aircraft,the simulation results show that the anomaly data can be detected effectively by the proposed method of reconstructing ADS-B data with the seq2seq model,and its running time is reduced.Compared with the RNN,the accuracy of anomaly detection is increased by 2.7%.The performance of the proposed model is better than that of the traditional anomaly detection models.

基于双流网络与多示例学习的异常事件检测

面对当前复杂场景下异常事件检测算法过度依赖帧级别标记,以及I3D模型耗时长、内存占用大等问题,设计了一种基于I3D的M-I3D模型并将其作为特征提取器,提出一种了基于深度时空特征和多示例学习的异常检测方法。所提方法将正常视频和异常视频作为包,并将视频片段作为多示例学习中的示例。利用M-I3D模型提取每个视频片段的特征,并将提取到的特征向量输入到三层全连接层中,进而自动学习一个深度异常排序模型,该模型可以预测异常视频片段的分数。此外,为了在训练过程中较好地定位异常,在排序损失函数中引入稀疏函数和约束性函数。结果表明,与其他方法相比,所提算法在UCF-Crime数据集上具有更高的准确率和更好的实时性。

Abnormality monitoring model of cracks in concrete dams

The abnormality monitoring model (AMM) of cracks in concrete dams is established through integrating safety monitoring theories with abnormality diagnosis methods of cracks. In addition, emphasis is placed on the influence of crack depth on crack mouth opening displacement (CMOD). A linear hypothesis is proposed for the propagation process of cracks in concrete based on the fictitious crack model (FCM). Abnormality points are detected through testing methods of dynamical structure mutation and statistical model mutation. The solution of AMM is transformed into a global optimization problem, which is solved by the particle swarm optimization (PSO) method. Therefore, the AMM of cracks in concrete dams is established and solved completely. In the end of the paper, the proposed model is validated by a typical crack at the 105 m elevation of a concrete gravity arch dam.