In general case, using a combination of diagnostic features, with each feature characterizing the diagnostic object’s condition with a certain probability, it would be correct to establish a critical rule by which the combination of features would be considered to be related to one of the possible states (diagnoses). In particular case, it is important to distinguish two diagnoses (differential diagnostics or dichotomy), such as “good condition” and “faulty condition”.
Methods of statistics decisions, such as minimal risk method, minimal faulty decision method, minimax method, Neumann-Pearson method, maximum likelihood method allow to choose the critical rule using optimum conditions, such as minimum risk condition, reducing chance of a diagnostic error at a set level of another error.
The paper considers selection of a method for establishing standard values of diagnostic features using a theory of statistic decisions-making. The paper dwells upon calculation of probability of defect skipping and false alarm, as well as failure skipping risk while establishing a standard feature value by various methods of decision-making.
Kostyukov V.N., Naumenko A.P. Risk assessment for establishing standard values of diagnostic features // Dinamika sistem, mehanizmov i mashin (Dynamics of systems, mechanisms and machines). - Omsk: OmSTU, 2014. - Iss. 4. - p. 150-154. Download the publication
You can order the translation by e-mail: a@dynamics.ru
|