Method of principal components in solvsng task of GAS
TURBINE PARAMETERS indirect measurings
M.V.
Shevchenko, S.V. Yepifanov
The problem of the gas turbine engine nonmetering
parameters deviations estimation using deviation of monitored parameters
statistical analysis has been handled. Possible changing of the engine
technical condition and lack of information about this changing caused by
metering system limited capacity has been taken into account. Thrust was
considered as nonmetering parameter. Variance
analysis for linear system of equations in matrix form has been carried out. Multicollinearity of the assigned task design matrix
elements has been found out. The principal component analysis with succeeding
regression on principal components procedure that allows estimating biased
assessments has been suggested and analyzed. The GTE with five metering
parameters and twenty three parameters of technical condition was considered as
observation object The faults of the engine gas path
were modeled using nonlinear component–based model of observable object. The
assessments of parameters deviation have been obtained and have been compared
with nonlinear model results. The accuracy of assessments has been estimated.
Key words: gas turbine engine, diagnostics, thrust, correlation matrix, statistical
assessment, principal components.