Reliability estimation using an integrated support vector regression – variable neighborhood search model
- Publisher:
- Elsevier BV
- Publication Type:
- Journal Article
- Citation:
- Journal of Industrial Information Integration, 2019, 15, pp. 103-110
- Issue Date:
- 2019-09-01
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| 1-s2.0-S2452414X18300876-main.pdf | Published version | 1.06 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
© 2019 Elsevier Inc. As failure and reliability predictions play a significant role in production systems they have caught the attention of researchers. In this study, Support Vector Regression (SVR), which is known as a powerful neural network method, is developed as a way of forecasting reliability. Generally, SVR is applied in many research environments, and the results illustrate that SVR is a successful method in solving non-linear regression problems. However, SVR parameters tuning is a vital task for performing an accurate reliability estimation. We propose variable neighborhood search (VNS) for continuous space, including some simple but efficient shaking and local search as its main operators, to tune the SVR parameters and create a novel SVR-VNS hybrid system to improve the reliability of estimation accuracy. The proposed method is validated with a benchmark from the former literature and compared with conventional techniques, namely RBF (Gaussian), AR (autoregressive), MLP (logistic), MLP (Gaussian), and SVMG (SVM with genetic algorithm). The experimental results indicate that the proposed model has a superior performance for prediction reliability than other techniques.
Please use this identifier to cite or link to this item: