Data mining for structural damage identification using hybrid artificial neural network based algorithm for beam and slab girder / Meisam Gordan

Meisam , Gordan (2020) Data mining for structural damage identification using hybrid artificial neural network based algorithm for beam and slab girder / Meisam Gordan. PhD thesis, Universiti Malaya.

[img] PDF (The Candidate's Agreement)
Restricted to Repository staff only

Download (238Kb)
    PDF (Thesis PhD)
    Download (5Mb) | Preview


      One of the approaches for structural health monitoring (SHM) consists of two major components, i.e. a network of sensors to collect the response data and an extraction method to obtain information on the structural health condition. Data mining (DM) is a novel data extraction technology which can employ for development of inverse analysis. Implementation of DM techniques in different areas of civil engineering has recently given very good results. However, application of DM in SHM is not used as much as expected, thus, many challenges are still ahead. Therefore, it is necessary to develop the applicability of DM in SHM. Hence, the main objective of this research is to address the feasibility and demonstrate the potential of DM in SHM, as well as apply DM technology which includes machine learning, artificial intelligence and statistical methods for damage detection of a lab-scale slab-on-girder bridge and four steel I beams structures using dynamic parameters of structures. To this end, experimental modal analysis of aforesaid structures was carried out by introducing different damage scenarios to generate the modal parameter database. Single-type and multiple-type damage cases were inflicted in the composite bridge structure to measure the first four natural frequencies and corresponding mode shapes of the structure at different locations, while only single-type damage using the first four flexural modes was considered in beam-like structures. The collected datasets were used as inputs for the DM process. In this research, a fault diagnosis methodology based on Cross Industry Standard Process for Data Mining (CRISP-DM) model was proposed for the purpose of damage detection. In the modeling phase, amongst all DM algorithms, the applicability of machine learning, artificial intelligence and statistical data mining techniques were examined using Support Vector Machine (SVM), Artificial Neural Network (ANN) and Classification and Regression Tree (CART) to detect the hidden patterns in vibration data. After evaluating the results of these algorithms, a hybrid Artificial Neural Network-based Imperial Competitive Algorithm (ANN-ICA) was presented in the deployment step of the proposed methodology to identify the structural damage of illustrative structures. According to the obtained results, the pre-developed ANN achieved generally more reliable capability of prediction in compare to SVM and CART. In contrast, CART showed the lowest performance of prediction fitness amongst all the patterns. It was also concluded that the ICA could successfully improve the learning process of the neural network. Therefore, the obtained results through the proposed DM procedure using ANN-ICA (1) confirmed the robustness of the hybrid network in compare to the pre-developed network, and (2) indicated that, the proposed damage identification model can be considered as a precise approach for monitoring the structural condition subjected to vibrational loads.

      Item Type: Thesis (PhD)
      Additional Information: Thesis (PhD) - Faculty of Engineering, Universiti Malaya, 2020.
      Uncontrolled Keywords: Data mining-based damage identification, Artificial neural network; Imperialist competitive algorithm; Support vector machine; Classification; Regression tree
      Subjects: T Technology > TA Engineering (General). Civil engineering (General)
      Divisions: Faculty of Engineering
      Depositing User: Mr Mohd Safri Tahir
      Date Deposited: 01 Oct 2021 01:54
      Last Modified: 10 Jan 2023 06:40

      Actions (For repository staff only : Login required)

      View Item