论文信息 - Sliding Dynamic Data Window: Improving Properties of the Incremental Learning Methods

Sliding Dynamic Data Window: Improving Properties of the Incremental Learning Methods

Abstract For updating Fault Detection and Diagnosing (FDD) systems, various techniques have been successfully proposed and implemented. Among them, Incremental Learning (IL) algorithms have shown to be reliable for updating FDD while it has own shortcomings. The FDD system with IL algorithms learns from recent available data without forgetting any previous redundant information. In order to address this drawback of IL for updating the FDD, the use of an Incremental Learning Dynamic Window (ILDW) is proposed. Based on ILDW dated information comparing with the current concept of the process is disregarded and the FDD within the sliding dynamic window is retrained. In this study advantageous of the ILDW are proved while Support Vector Machines (SVM) is employed as fault detection method with its application on simulation of the CSTR reactor. Results show that employing ILDW for updating SVM instead of just IL not only leads to similar outcomes regarding accuracy, but also significantly reduce the number of required support vectors and training time.

[1] Sameer Singh,et al. Novelty detection: a review - part 1: statistical approaches , 2003, Signal Process..

[2] Moisès Graells,et al. Incremental Learning Fault Detection Algorithm Based on Hyperplane-Distance , 2016 .

[3] Tao Wang,et al. Study on the classification of data streams with concept drift , 2011, 2011 Eighth International Conference on Fuzzy Systems and Knowledge Discovery (FSKD).

[4] Sameer Singh,et al. Novelty detection: a review - part 2: : neural network based approaches , 2003, Signal Process..

[5] Antonio Espuña,et al. Optimal Features Selection for Designing a Fault Diagnosis System , 2016 .

[6] Cunhe Li,et al. The incremental learning algorithm with support vector machine based on hyperplane-distance , 2011, Applied Intelligence.

[7] Junhong Wang,et al. A fast incremental extreme learning machine algorithm for data streams classification , 2016, Expert Syst. Appl..

[8] Gerhard Widmer,et al. Learning in the Presence of Concept Drift and Hidden Contexts , 1996, Machine Learning.

[9] Vincent Lemaire,et al. A Survey on Supervised Classification on Data Streams , 2014, eBISS.

[10] Dimitrios Gunopulos,et al. Incremental support vector machine construction , 2001, Proceedings 2001 IEEE International Conference on Data Mining.

[11] Qingtian Zeng,et al. An incremental learning algorithm for Lagrangian support vector machines , 2009, Pattern Recognit. Lett..