Measurement and inspection technology are important parts of manufacturing industry today. They provide important feedbacks to modern control systems, detect failures and flaws either online or offline, and hence enhance boost the productivity and improve the quality of final products. The continued development and evolution of measurement principles and systems have produced increasingly precise and reliable results. Published literature and industry practice show that better measurement results usually require both a "ruler" of finer scales and a better method to make more stable readings; however, these are bottlenecks in developing new methods and improving existing systems. In this thesis a probability-based method (PBM) for length measurement is proposed which produces measurement results more accurate than the physical scales on the "ruler". As a natural extension to the classic approach, this measurement model uses probability readings as indicators and evaluates the measurands. Around this method, a theory is established and proven for various levels of priori knowledge. Following this, real-world obstacles in implementation and practice usage of PBM are considered. For example, mis-alignments which not only cause trigonometric errors but also induce a discrete random behavior on the measurand, sources of sampling randomness consisting of synthetic parts and those from random errors, and impacts of laboratory gantry systems are fully investigated. A validation process is conducted using existing hardware with newly developed software. Multiple tests are conducted. The experiment results confirmed the validity of PBM and therefore shows that this method will make a positive contribution to the industrial inspection.
|Date of Award||10 Jun 2019|
- University Of Strathclyde
|Supervisor||Yi Qin (Supervisor) & Xichun Luo (Supervisor)|