Toronto Metropolitan University
Browse

Estimation of Weibull parameters using artificial neural network

Download (13.45 MB)
thesis
posted on 2021-05-22, 13:37 authored by Md. Sujauddin Mallick

Weibull distribution is an important distribution in the field of reliability. In this distribution usually there are two parameters. The usual parameter estimation method is maximum likelihood estimation. Maximum likelihood estimation requires mathematical formulation and prior assumption. Non parametric method such as neural network does not require prior assumption and mathematical formulation. They need data to formulate the model. In this report feed forward neural network with back propagation is used to estimate the parameters of a two-parameter Weibull distribution based on four Scenarios. The Scenario consists of training and test data set. Training and test data set generated through simulated time to failure events using wblrnd function in MATLAB. The input to the network is time to failure, and the output is shape and scale parameters. The network is trained and tested using trainbr algorithm in MATLAB. The network performed better on Scenario 2 which has the larger number of training examples of shape and scale.

History

Language

English

Degree

  • Master of Engineering

Program

  • Mechanical and Industrial Engineering

Granting Institution

Ryerson University

LAC Thesis Type

  • Thesis

Year

2019

Usage metrics

    Mechanical and Industrial Engineering (Theses)

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC