Avionic Air Data Sensors Fault Detection and Isolation by means of Singular Perturbation and Geometric Approach.

Singular Perturbations represent an advantageous theory to deal with systems characterized by a two-time scale separation, such as the longitudinal dynamics of aircraft which are called phugoid and short period. In this work, the combination of the NonLinear Geometric Approach and the Singular Perturbations leads to an innovative Fault Detection and Isolation system dedicated to the isolation of faults affecting the air data system of a general aviation aircraft. The isolation capabilities, obtained by means of the approach proposed in this work, allow for the solution of a fault isolation problem otherwise not solvable by means of standard geometric techniques. Extensive Monte-Carlo simulations, exploiting a high fidelity aircraft simulator, show the effectiveness of the proposed Fault Detection and Isolation system.

Meta-Learning With Distributional Similarity Preference for Few-Shot Fault Diagnosis Under Varying Working Conditions.

Few-shot fault diagnosis is a challenging problem for complex engineering systems due to the shortage of enough annotated failure samples. This problem is increased by varying working conditions that are commonly encountered in real-world systems. Meta-learning is a promising strategy to solve this point, open issues remain unresolved in practical applications, such as domain adaptation, domain generalization, etc. This article attempts to improve domain adaptation and generalization by focusing on the distribution-shift robustness of meta-learning from the task generation perspective. In fact, few-shot fault diagnosis under varying working conditions allows to address the distribution shift problem in a natural way. An unsupervised across-tasks meta-learning strategy with distributional similarity preference is proposed, where the core is the distribution-distance-weighting mechanism. Differently from the naive random meta-train task generation strategy used in existing meta-learning methods, the source instances that present a more similar distribution with respect to the target instances gain larger weightings in the task generation. This strategy leads to a meta-task training set that is enough diverse, and at the same time can be easily learned due to the distribution similarity features of the source tasks. The proposed method introduces the concept of maximum mean discrepancy that is applied to derive the distribution distance of the measurements. Moreover, a model-agnostic meta-learning is applied to realize few-shot fault diagnosis under varying working conditions. The proposed solutions are verified and compared by considering two public datasets used for bearing fault diagnosis. The results show that the proposed strategy outperforms different related few-shot fault diagnosis methods under varying working conditions. Moreover, it is thus proved that, meta-learning with distribution similarity feature represents an effective approach for domain adaptation and generalization.

A novel algorithm for high compression rates focalized on electrical power quality signals.

This research proposes a high-performance algorithm for the compression rate of electrical power quality signals, using wavelet transformation. To manage the massive amount of data the telecommunications networks are constantly acquiring it is necessary to study techniques for data compression, which will save bandwidth and reduce costs extensively by avoiding having massive data storage facilities. First biorthogonal wavelet level six transform is applied, however after compression, the reconstructed signal will have a different amplitude and it will be shifted when compared to the original one. Then, normalization is used (for amplitude correction between the original signal and reconstructed one) by multiplying the reconstructed signal by the result of the division between the original signal maximum magnitude and the reconstructed signal maximum magnitude. Thirdly, the ripple in the reconstructed signal is eliminated by applying a moving average filter. Finally, the shifting is corrected by finding the difference between the maximum points in a cycle of the original signal and the reconstructed one. After the compression algorithm was performed the best rates are 99.803% for compression rate, RTE 99.9479%, NMSE 0.000434, and Cross-Correlation 0.999925. Finally, this works presents two new performance criteria, compression time and recovery time, both of them in a real scenario will determinate how fast the algorithm can perform.