**Research on Foundation Problem about Computation Framework of Measurement Error**

Author:Dai Wei Hua

Supervisor:wang yue ke

Database:Doctor

Degree Year:2017

Download:3

Pages:146Size:2207K

Keyword:computation framework，finite computation，measurement error，noise identification，pdf

By evaluating the measurement error,one can directly determine whether the measurement accuracy meets the requirement,otherwise,the deficiency of the measuring method can be revealed its performance can be further improved.In order to promote the research on measurement error evaluation,it is necessary to establish a uniform computation framework for measurement error evaluation.Measurement errors exists in the results and should be evaluated on various measurement occasions.However,measurement errors in a different situations situation generally obeys different rules,therefore,it is difficult if measurement errors are evaluated according to an uniform computation framework.This leads to the basic problem of this paper-how to establish a uniform computation framework to achieve the measurement errors evaluation in general measuring situations,around this issue,this paper focuses on the following key problems:1.Design of computation framework for measurement errors.The purpose of establishing an uniform computation framework for measurement error evaluation is to provide a reference for others to compute the measurement errors with an framework that consists of general techniques.Measurement errors are always different from each other along with that the measurands,measuring methods and circumstances change in different measurement situations.However,we cannot provide a certain method of measurement error evaluation for each situation,a more reasonable choice is to design a general method that fits any situation in the sense of measurement principle.Therefore,design of computation framework for measurement errors is one of key issues of establishing uniform computation framework for measurement error evaluation.2.Computation of probability density function.In the classical theory of measurement error,a known probability density function(pdf)of measurement error is desirable for calculating the confidence interval(or coverage interval)and extreme error.Similarly,the problem "how to obtain the pdf of measurement error" would be inevitable when we perform the measurement error evaluation according with the uniform computation framework.Thus,the pdf computation is one of key issues of establishing an uniform computation framework for measurement error evaluation.3.Finite computation achievement.In the uniform computation framework,the major contents that are intended to compute include the pdf of measurement error and other evaluations.Furthermore,pdf is the basic of other evaluations in general.However,it performs a traverse above the total interval of measurand quantity in theory when we intend to compute pdf and other evaluations.This would desire infinite operations which is impossible to achieve.As a result,the finite computation achievement is also a key issue of establishing an uniform computation framework for measurement error evaluation.4.Noise identification.Noise generally exists on any physical process of measurement,and random error occurs in the measurement result just due to the existence of noise,whose random characteristic directly affects the form of errors’ PDF.It aims at obtaining the statistical characteristic of noise through noise identification,which is always the precondition of the pdf computation of measurement error.Therefore,noise identification also becomes a key issue of establishing an uniform computation framework for measurement error evaluation.To solve the key technical problems above,the paper designed a uniform computation framework for measurement error,the core of which is to compute the pdf of measurement error,and proposed a method based on geometric space analysis to compute the pdf of error,a method based on quantity discreteness to achieve the finite computation of measurement error evaluation,a method based on statistical-test to perform the noise identification.These technical methods corresponds to the four key technical problems above,and theirfeasibilities and effectiveness are verified by theoretical analysis and simulation.The work in this paper could provide a reference for others to establish more effective computation frameworks for measurement error evaluation.