On the use of compound noise factor in parameter design experiments
This paper deals with theoretical analysis of a compound noise factor technique which replaces a noise array by a compound noise factor that has only a few levels, typically two. The author considers the model \[ y=f(x_{1},\ldots,x_{l})+\sum_{j=1}^{m}\beta_{j}z_{j}+ \sum_{j=1}^{m}\sum_{i=1}^{l}\gamma_{ij}x_{i}z_{j}, \] where \(x_{1},\ldots,x_{l}\) are the control factors, \(z_{1},\ldots,z_{m}\) are the noise factors, \(f(x_{1},\ldots,x_{l})\) is a general function, and \(z_{j}\) is symmetric about zero. Two settings \(z^{*}\) and \(z_{*}\) are called extreme compound noise settings if \(y(x;z^{*})=\max_{z}\{y(x;z)\}\) for any \(x\), resp. \(y(x;z_{*})=\min_{z}\{y(x;z)\}\) for any \(x\). The author proves that extreme settings exist if and only if \[ |\beta_{j}|\geq\sum_{i=1}^{l}|\gamma_{ij}|\quad \text{for}\quad j=1,\ldots,m, \] where \(x_{i}\) are rescaled to be in \([-1,1]\). Examples are given to show that extreme settings may not exist and to illustrate the ineffectiveness of extreme settings in some cases. Conditions that guarantee the effectiveness of the compound noise technique are obtained and numerical illustrations are presented.
- Effectiveness of a random compound noise strategy for robust parameter design
- Correlation coefficient of response variances in compound noise and product array experiments
- Response model analysis for robust design experiments
- Some Useful Compound Dispersion Experiments in Quality Design
- On the determination of robust settings in parameter design experiments
- Effectiveness of a random compound noise strategy for robust parameter design
- scientific article; zbMATH DE number 418925 (Why is no real title available?)
- Incorporating Noise Factors into Experiments with Censored Data
- Correlation coefficient of response variances in compound noise and product array experiments
- A confidence region for zero-gradient solutions for robust parameter design experiments
This page was built for publication: On the use of compound noise factor in parameter design experiments
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4797848)
