Optimization design for the installation position of hold bracket in railway bogie system by uniform design method

ABSTRACT

The sensor hold bracket design is an essential element for running safety in the railway vehicle bogie braking system. The installation position of the hold bracket in the bogie system affects the brake sensing performance. This paper aims to determine the optimal installation position of the hold bracket. Three installation angles and one geometry dimension of a holding bracket are selected for the control factors. The uniform design of the experiment method creates a set of experiments. The surrogate models from input and output data are obtained and compared by applying Kriging interpolation and radial basis functions. The final optimal design models of the holding bracket for two surrogate models are determined using the genetic algorithm. The results show that the optimized parameters could substantially improve the installation angle error of the pressure switch from 4° to 0.04°, which aligns with expectations. Finally, to confirm the effect of the control parameters and their reliability and validity, the impact of the pressure switch is expected when the optimal hold bracket model is installed in the railway vehicle bogie system. As a result, the optimal design procedure can improve and reduce the risk of sensor failure during train operation.

CO EDITOR-IN-CHIEF:

• Hsiau, Shu-San

ASSOCIATE EDITOR:

• Tsai, Jhy-Cherng

KEYWORDS:

• Hold bracket system
• uniform design
• Kriging interpolation
• radial basis function

Disclosure statement

No potential conflict of interest was reported by the author(s).

Nomenclature

$\mathit{c}$	=	the center of input values
$\mathit{F}$	=	n-dimensional column vector with all ones
$\mathit{h}\left(\mathit{x}\right)$	=	radial basis function
$\mathit{r}\left(\mathit{x}\right)$	=	n-dimensional vector
$\mathit{R}$	=	known square matrix
$R_1$	=	the metric
$\mathit{y}\left(\mathit{x}\right)$	=	unknown response function
$\mathit{x}$	=	vector with unidentified input variables
$\mathit{Y}$	=	known response vector
$\mathit{\beta }$	=	known constant
$\mathit{\varphi }$	=	the function used by Gaussian

Additional information

Funding

The authors would like to express our gratitude and appreciation for the financial support of the Ministry of Science and Technology (MOST) under grant nos. [109-2221-E-992 -022 and 111-2221-E-992-057]. Furthermore, I would like to thank the undergraduate research team for their collaborative effort.