Simply level recalculation using statistical approach

Shih Fang Huang*, Chun Sung Chen, Ren-Jye Dzeng

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


A railway is a strip-shaped corridor usually divided into multiple sections that are constructed separately. Associated construction contracts must consider scheduling. Because a railway is a continual linear structure, high-precision level surveys are needed for rail sections to connect smoothly. However, the different phases and sections of construction require multiple-level surveys, which often lead to bias in the benchmarks. Since the railway is a continuous alignment, and an inconsistent elevation system affects the civil work and subsequent track laying. This study therefore used a statistical method to eliminate inconsistencies in benchmarks. Statistics are widely used in engineering and in daily life, to solve decision making problems involving uncertainties. This study used the expected value of closing error between benchmarks as an index for recalculating level, and used the standard error of expected value as the accuracy index. Application of the proposed method using measurement data for a Taipei underground project of the Taiwan Railway Administration showed that it eliminates bias in benchmarks and provides the required accuracy for closure between adjacent points.

Original languageEnglish
Title of host publicationInformation Technology for Manufacturing Systems II
Number of pages6
StatePublished - 22 Jul 2011
Event2011 International Conference on Information Technology for Manufacturing Systems, ITMS 2011 - Shanghai, China
Duration: 7 May 20118 May 2011

Publication series

NameApplied Mechanics and Materials
ISSN (Print)1660-9336
ISSN (Electronic)1662-7482


Conference2011 International Conference on Information Technology for Manufacturing Systems, ITMS 2011


  • Benchmarks
  • Level
  • Railway
  • Statistics

Fingerprint Dive into the research topics of 'Simply level recalculation using statistical approach'. Together they form a unique fingerprint.

Cite this