Search

The anomaly detection of aeroengines faces significant challenges, including high noise, complex parameter correlations, and imbalanced data. Current methods primarily rely on the reconstruction error of autoencoders to isolate anomalies. However, such methods are ineffective in detecting minor anomalies, as the reconstruction error for minor anomaly samples closely resembles that of normal samples. To address this problem, this paper proposes an innovative Deep Autoencoder Anomaly Detection Model (DAADM) for the anomaly detection of aeroengines. DAADM consists of two sub-networks: Anomaly Score Calculation Network (ASCN) and Deep Feature Extraction Network (DFEN). Firstly, ASCN introduces the adversarial training strategy into deep autoencoders, ensuring training stability while effectively isolating minor anomalies. Secondly, DFEN extracts deep-level features from input samples, providing a different perspective from ASCN to enhance sample information extraction. The combination of ASCN and DFEN compensates for their respective deficiencies, comprehensively considering input sample information, thereby promoting the capability of anomaly detection. Finally, the proposed DAADM is validated on real an aeroengine dataset and a public dataset. On the aeroengine dataset, the accuracy, recall, and F1 score of DAADM exceed the state-of-the-art method by 4.73%, 4.79%, and 5.76%, respectively. It is noteworthy that experimental results have proven that DAADM has strong anti-noise capabilities. [ABSTRACT FROM AUTHOR]

Collective entity linking always outperforms independent entity linking because it considers the interdependencies among entities. However, the existing collective entity linking methods often have high time complexity, do not fully utilize the relationship information in heterogeneous information networks (HIN) and most of them are largely dependent on the special features associated with Wikipedia. Based on the above problems, this paper proposes a novel collective entity linking method based on relationship path in heterogeneous information networks (PathEL). The PathEL classifies complex relationships in HIN into 1-hop paths and 3 types of 2-hop paths, and measures entity correlation by the path information among entities, ultimately combining textual semantic information to realize collective entity linking. In addition, facing the high complexity of collective entity linking, this paper proposes to solve the problem by combining the variable sliding window data processing method and the two-step pruning strategy. The variable sliding window data processing method limits the number of entity mentions in each window and the pruning strategy reduces the number of candidate entities. Finally, the experimental results of three benchmark datasets verify that the model proposed in this paper performs better in entity linking than the baseline models. On the AIDA CoNLL dataset, compared to the second-ranked model, our model has improved P, R, and F1 scores by 1.61%, 1.54%, and 1.57%, respectively. [ABSTRACT FROM AUTHOR]

To address existing methodological limitations, a novel method for constructing a simulation surrogate model for a spiral shaft is proposed based on a double-layer progressive architecture, which consists of two steps. Firstly, the Goal-oriented autoencoder (GOAE)-classifier model discriminates the feasibility of the samples, ensuring accuracy through differential feature extraction by GOAE. Then, the Self-attention Artificial Neural Network (Self-attention ANN) weight allocation model assigns adaptive weights to surrogate models under different working conditions, which implements a dynamic combination of sub-model outputs and improves the model accuracy. Validation through a dataset generated via Latin hypercube sampling (LHS) and simulation affirms the efficacy, demonstrating fast and accurate simulation. The results underscore its potential for collaborative optimization. In summary, the proposed method has obvious advantages and can be extended to other fields to meet the needs of engineering. • Simulation surrogate is divide into two steps by a double-layer progressive model. • The proposed GOAE-classifier is used to discriminate the feasibility of samples. • An adaptive hybrid surrogate model is proposed to predict the feasible samples. • The weights of sub-models are adjusted by the proposed Self-attention ANN. • The performances of different models are compared based on generated dataset. [ABSTRACT FROM AUTHOR]

The rapid collaborative optimization (CO) has an increasing demand for high-fidelity surrogate models. However, the traditional surrogate model cannot be applied to all working conditions due to the limitations of model applicability. A hybrid surrogate model is proposed, which uses the attention mechanism to automatically decide the weights of the sub models according to the working conditions to ensure its approximation ability under all working conditions. First, according to the characteristics of the finite element analysis (FEA) parameters, a comprehensive design of experiment (DOE) is proposed, which ensures the space-filling property of the samples. Secondly, a Self-attention artificial neural network (ANN) is proposed to automatically adjust the weights of sub-surrogate models, improving the attention to the working condition-related features. The proposed Self-attention ANN is a general framework that can provide support for the adaptive weight decision in other equipment simulation hybrid surrogate models. The experiment on the database shows that the error of the two hybrid surrogate models established by the proposed method is 36.04% and 33.31% lower than that of the advanced model, respectively, and is significantly superior to other methods. This achievement not only combines the spatial approximation ability of sub models to establish nonlinear model, achieving the purpose of high-fidelity simulation of FEA systems, but also enables surrogate model covering complete space using limited samples, making the model suitable for various practical engineering problems. In summary, the proposed method has obvious advantages in solving existing problems, provides strong support for research and practical applications. • The rapid adaptive hybrid surrogate model is proposed to replace the FEA model. • The weights of sub-surrogate models are adjusted by the proposed Self-attention ANN. • A comprehensive DOE is used to ensure the quality and validity of sampling points. • The performance of different surrogate model establishing methods is compared. [ABSTRACT FROM AUTHOR]