Process monitoring based on neural networks is getting more and more attention. Compared with classical neural networks, high-order neural networks have natural advantages in dealing with heteroscedastic data. However, high-order neural networks might bring the risk of overfitting and learning both the key information from original data and noises or anomalies. Orthogonal constraints can greatly reduce correlations between extracted features, thereby reducing the overfitting risk. This paper proposes a novel fault detection method called second-order component analysis (SCA). SCA rules out the heteroscedasticity of pro-cess data by optimizing a second-order autoencoder with orthogonal constraints. In order to deal with this constrained optimization problem, a geometric conjugate gradient algorithm is adopted in this paper, which performs geometric optimization on the combination of Stiefel manifold and Euclidean manifold. Extensive experiments on the Tennessee-Eastman benchmark pro-cess show that SCA outperforms PCA, KPCA, and autoencoder in missed detection rate (MDR) and false alarm rate (FAR).
Fault Detection via Occupation Kernel Principal Component Analysis
School of Mechanical and Aerospace Engineering, Oklahoma State University(俄克拉荷马州立大学机械与航空航天工程学院);Computer Science and Mathematics Division, Oak Ridge National Laboratory(橡树岭国家实验室计算机科学与数学分部);Automatic Laboratory, Ecole Polytechnique Federale de Lausanne(洛桑联邦理工学院自动实验室)
Lithological Controls on the Permeability of Geologic Faults: Surrogate Modeling and Sensitivity Analysis
Oden Institute for Computational Engineering and Sciences(奥登计算工程与科学研究所);The University of Texas at Austin(德克萨斯大学奥斯汀分校);Department of Earth and Planetary Sciences(地球与行星科学系);Harvard University(哈佛大学);Department of Civil and Environmental Engineering(土木与环境工程系);Massachusetts Institute of Technology(麻省理工学院);Department of Earth, Atmospheric, and Planetary Sciences(地球、大气与行星科学系)