Recursive least squares processing of fuzzy measurement

Проведен анализ традиционного подхода к задаче обработки временного ряда. Сформированы его недостатки, связанные с необходимостью хранения непрерывно пополняемого массива измерений и неучетом различной полезности "старых" и "свежих" измерений. Обосновано применение рекуррентного варианта метода наименьших квадратов. Предложена процедура реализации рекуррентной обработки измерений для случая, когда они заданы нечетко. Определена функция принадлежности нечеткого результата оценки параметров временного ряда.The problem of statistical processing of time series, given by a set of fuzzy measurements. The analysis of the traditional approach to solving this problem and formulate its drawbacks associated with the need to continuously replenished storage array measurements and the neglect of various utility of "old" and "fresh" measurements. The application of the recursive version of the method of least squares. Relationships for recursive calculations of the vector of parameter estimates of the regression model describing the behavior of the controlled parameter. The purpose of the article - the development of a method of implementing the traditional procedure of recursive processing of measurements for the case when these measurements are fuzzy values. To describe the values of fuzzy measurement of the controlled variable used membership function (L-R)-type. Described step by step technique for calculating the parameters of membership functions of the fuzzy result of applying the recursive procedure. Defined function of fuzzy values of the controlled variable at the time of the forecast