Detection of Multiple Change-point in Non-linear Profiles

Khanzadeh, Mojtaba | 2015

766 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47300 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Akhavan Niaki, Taghi
  7. Abstract:
  8. This effort attempts to study the multiple change-point problem in the area of non-linear profiles. Two methods for estimating the times of multiple change-points is proposed. In the first method, a model consisting of two networks, which is based on artificial neural networks, is proposed. These networks are distinctive only in their training data. One network is trained for ascending segment of the profile and the other is trained for descending segments of the profile. In the second method, Bayesian approach is proposed for estimating multiple change-point. While using Bayesian approach the parameters of the Non-linear model must be estimated. However, this issue is complicated or impossible in quite a few cases. As a result, proposed Bayesian estimator is based on difference between the response variables and in-control profile curve. In this section our assumption is that the amount of the parameters are well defined after gradation. Consequently, by this method the inputs of Bayesian approach are being produced. Functionality of proposed estimators has been evaluated by simulation experiments and the results indicate that the estimators specifically Bayesian approach provide accurate and rather precise estimates of multiple change-points in Non-linear profiles in the selected case problems
  9. Keywords:
  10. Statistical Quality Control ; Nonlinear Profile ; Multiple Change Points ; Neural Network ; Bayesian Analysis ; Stochastic Approximation Monte Carlo (SAMC)Algorithm

 Digital Object List