You need JavaScript to view this

Statistical Control of Measurement Quality; Controle Statistique de la Qualite de la Mesure; Statisticheskim kontrol' kachestva izmerenij; Control Estadistico de la Calidad de las Mediciones

Conference:

Abstract

Effective nuclear materials management, and hence design and operation of associated material control systems, depend heavily on the quality of the quantitative data on which they are based. Information concerning the reliability of the measurement methods employed is essential to both the determination of data requirements and the evaluation of results obtained. Any method of analysis should be (1) relatively free from bias and (2) reproducible, or, in more usual terminology, precise. Many statistical techniques are available to evaluate and control the reproducibility of analytical results. Economical and effective experimental designs have been developed for the segregation of different sources of measurement error. Procedures have been developed or adapted tot use in maintaining and controlling the precision of routine measurements. All of these techniques require that at least some measurements must be duplicated, but duplication of all measurements can be justified only when the detection of every gross error, or mistake, is extremely important. Three types of measurement bias can be considered: (1) bias relative to a standard, (2) bias relative to prior experience, and (3) bias relative to a group. The first refers to the degree to which the measurements obtained deviate systematically from some ''standard'' which is unbiased  More>>
Authors:
Bennett, C. A. [1] 
  1. Battelle Memorial Institute, Richland, WA (United States)
Publication Date:
Feb 15, 1966
Product Type:
Conference
Report Number:
IAEA-SM-67/35
Resource Relation:
Conference: Symposium on Nuclear Materials Management, Vienna (Austria), 30 Aug - 3 Sep 1965; Other Information: 13 refs., 2 figs., 3 tabs.; Related Information: In: Nuclear Materials Management. Proceedings of the Symposium on Nuclear Materials Management| 905 p.
Subject:
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS; ACCURACY; CONTROL SYSTEMS; DESIGN; ERRORS; EVALUATION; NUCLEAR MATERIALS MANAGEMENT; RELIABILITY; SEGREGATION; STANDARDS
OSTI ID:
22127241
Research Organizations:
International Atomic Energy Agency, Vienna (Austria)
Country of Origin:
IAEA
Language:
English
Other Identifying Numbers:
Other: ISSN 0074-1884; TRN: XA13M2336082320
Submitting Site:
INIS
Size:
page(s) 465-480
Announcement Date:
Sep 12, 2013

Conference:

Citation Formats

Bennett, C. A. Statistical Control of Measurement Quality; Controle Statistique de la Qualite de la Mesure; Statisticheskim kontrol' kachestva izmerenij; Control Estadistico de la Calidad de las Mediciones. IAEA: N. p., 1966. Web.
Bennett, C. A. Statistical Control of Measurement Quality; Controle Statistique de la Qualite de la Mesure; Statisticheskim kontrol' kachestva izmerenij; Control Estadistico de la Calidad de las Mediciones. IAEA.
Bennett, C. A. 1966. "Statistical Control of Measurement Quality; Controle Statistique de la Qualite de la Mesure; Statisticheskim kontrol' kachestva izmerenij; Control Estadistico de la Calidad de las Mediciones." IAEA.
@misc{etde_22127241,
title = {Statistical Control of Measurement Quality; Controle Statistique de la Qualite de la Mesure; Statisticheskim kontrol' kachestva izmerenij; Control Estadistico de la Calidad de las Mediciones}
author = {Bennett, C. A.}
abstractNote = {Effective nuclear materials management, and hence design and operation of associated material control systems, depend heavily on the quality of the quantitative data on which they are based. Information concerning the reliability of the measurement methods employed is essential to both the determination of data requirements and the evaluation of results obtained. Any method of analysis should be (1) relatively free from bias and (2) reproducible, or, in more usual terminology, precise. Many statistical techniques are available to evaluate and control the reproducibility of analytical results. Economical and effective experimental designs have been developed for the segregation of different sources of measurement error. Procedures have been developed or adapted tot use in maintaining and controlling the precision of routine measurements. All of these techniques require that at least some measurements must be duplicated, but duplication of all measurements can be justified only when the detection of every gross error, or mistake, is extremely important. Three types of measurement bias can be considered: (1) bias relative to a standard, (2) bias relative to prior experience, and (3) bias relative to a group. The first refers to the degree to which the measurements obtained deviate systematically from some ''standard'' which is unbiased either (1) by definition, or (2) because all known sources of bias have been removed. The second in concerned with the presence of systematic differences over a period of time. The third type of bias concerns the relationship between different physical entities or individuals at a given time. Recent developments in statistical methodology applicable to the evaluation of all three types of bias are discussed. Examples of the use of the statistical techniques discussed on Hanford data are presented. (author) [French] La gestion efficace des matieres nucleaires et, par consequent, la conception et le fonctionnement des systemes de controle de ces matieres dependent beaucoup de la valeur des donnees quantitatives sur lesquelles ils sont fondes. Il importe donc de connaitre le degre de precision des methodes de mesure employees, afin de determiner les exigences en matiere de donnees et d'evaluer les resultats obtenus. Toute methode d'analyse doit etre 1. relativement exempte d'erreur systematique, 2. reproductible, autrement dit precise. On dispose de nombreuses methodes statistiques pour evaluer et controler la reproductibilite des resultats des analyses. On a mis au point des moyens economiques et experimentaux pour separer les diverses sources d'erreurs de mesure. On a cree ou adapte des methodes permettant de maintenir et de controler la precision des mesures courantes. Toutes ces techniques exigent qu'au moins un certain nombre de mesures soient faites deux fois; mais la repetition de toutes les mesures n'est justifiee que lorsqu'il est extremement important de detecter toute erreur grossiere. On peut considerer trois sortes d'erreurs systematiques dans les mesures: 1. l'erreur par rapport a un etalon, 2. l'erreur par rapport aux resultats de mesures precedentes, 3. l'erreur propre a un groupe.. La premiere est la deviation systematique des mesures obtenues par rapport a un ' etalon ' exempt de deviation, soit par definition, soit parce que l'on a supprime toutes les causes connues de deviation. La deuxieme concerne des differences systematiques observees sur un certain laps de temps. La troisieme concerne le rapport entre differentes entites ou individualites physiques a un moment donne. L'auteur expose les moyens les plus recents de la methodologie statistique applicables a l'evaluation de chacune de ces trois sortes d'erreurs systematiques. H donne des exemples d'utilisation des methodes statistiques appliquees aux donnees concernant le reacteur de Hanford. (author) [Spanish] La eficacia de la administracion de los materiales nucleares y, por tanto, la organizacion y el funcionamiento de los sistemas de control dependen, sobremanera de la exactitud de los datos cuantitativos en que estan basados. La informacion sobre la seguridad de los metodos, de medicion empleados es esencial tanto para determinar las necesidades en lo que respecta a los datos como para evaluar los resultados obtenidos. Todos los metodos de analisis deben estar relativamente exentos de desviacion y ser reproducibles o, por decirlo de una manera mas corriente, precisos. Para evaluar y controlar la reproducibilidad de los resultados de los analisis se dispone de varias tecnicas estadisticas. Se han ideado procedimientos experimentales eficaces y economicos para descubrir el origen de los errores. Tambien se han ideado o adaptado procedimientos para mantener y controlar la exactitud de las mediciones corrientes. Todas esas tecnicas requieren la repeticion de algunas mediciones, pero la repeticion de todas ellas solo esta justificada cuando es de suma importancia detectar todos los errores serios. En las mediciones cabe considerar tres tipos de desviacion o inclinacion viciosa: 1) desviacion respecto de un patron aceptado, 2) desviacion respecto de experiencias anteriores y 3) desviacion respecto de un grupo. El primer tipo se refiere el grado en que los valores-obtenidos se separan sistematicamente de un patron que esta exento de desviaciones ya sea por definicion ya sea por haberse eliminado todas las causas de desviacion conocidas. El segundo tipo se refiere a la presencia de diferencias sistematicas despues de un determinado periodo de tiempo. El tercer tipo concierne a las relaciones entre distintas entidades fisicas en un momento determinado. En la memoria se analizan algunos procedimientos recientes de metodologia estadistica que pueden servir para evaluar esos tres tipos de desviacion. Se dan, ademas, algunos ejemplos del empleo de esas tecnicas estadisticas con datos obtenidos en Hanford. (author) [Russian] Jeffektivnyj kontrol' i uchet jadernyh materialov i, sledovatel'no, organizacija i dejstvie svjazannyh s jetim sistem kontrolja za materialami, v znachitel'noj stepeni zavisjat ot kachestva kolichestvennyh dannyh, na kotoryh oni osnovany. Informacija otnositel'no nadezhnosti primenjaemyh metodov izmerenij neobhodima kak dlja opredelenija potrebnostej v dannyh, tak i dlja ocenki poluchennyh rezul'tatov. Pochti ljuboj metod analiza mozhno bylo by schitat' horoshim, esli by on 1) byl otnositel'no svoboden ot sistematicheskih oshibok i 2) byl vosproizvodimym, ili, upotrebljaja bolee prinjatuju terminologiju, tochnym. Dlja ocenki i kontrolja za vosproizvodimost'ju analiticheskih rezul'tatov imeetsja mnogo statisticheskih metodov. Byli razrabotany jekonomicheskie i jeffektivnye jeksperimental'nye konstrukcii dlja vydelenija razlichnyh istochnikov oshibok izmerenij. Byli razrabotany ili prisposobleny procedury dlja podderzhanija i kontrolja za tochnost'ju povsednevnyh izmerenij. Vse jeti metody trebujut dublirovat' po krajnej mere nekotorye izmerenija, odnako dublirovanie vseh izmerenij mozhet byt' opravdano tol'ko v tom sluchae, kogda obnaruzhenie kazhdoj gruboj bshibki ili pogreshnosti imeet iskljuchitel'nuju vazhnost'. Mozhno uchityvat' sledujushhie tri vida oshibok v izmerenijah: 1) sistematicheskaja oshibka, otnosjashhajasja k jetalonu, 2) sistematicheskaja oshibka, otnosjashhajasja k predydushhemu opytu i 3) sistematicheskaja oshibka, otnosjashhajasja k gruppe. Pervaja oshibka kasaetsja stepeni, s kotoroj poluchennye izmerenija sistematicheski otklonjajutsja ot predel'nogo ''jetalona'', kotoryj ne imeet oshibok, libo 1) soglasno opredeleniju, libo 2) potomu chto ustraneny vse izvestnye istochniki oshibok. Vtoraja oshibka svjazana s nalichiem sistematicheskih rashozhdenij togo ili inogo perioda vremeni. Tretij vid oshibok kasaetsja svjazi mezhdu razlichnymi svojstva- ' mi ili individuumami v dannoe vremja. Obsuzhdajutsja nedavnie dostizhenija v oblasti statisticheskoj metodologii, primenjaemoj k ocenke vseh treh vidov oshibok. Privodjatsja primery ispol'zovanija statisticheskih metodov na osnove dannyh Hjenforda. (author)}
place = {IAEA}
year = {1966}
month = {Feb}
}