A method of thermal network correction has been developed on the basis of regression analysis. The statistical tests were introduced into the correction process to obtain more detailed information and results. The method chooses important parameters using F-test and gives their estimated values with confidence intervals. It is also useful to establish the testing conditions which give sufficient temperature data. The normal equation is derived for minimizing the residuals of heat balance. The conjugate gradient method combined with the stepwise regression procedure is used to solve the equation. This technique greatly saves the computer time and its memory capacity. The numerical study with a model network showed that there is an optimum F-value to delete unimportant parameters. With this F-value, the network is corrected to satisfy most accurately the heat balance and possibly to give the most accurate prediction of temperatures. (Kursetz)
A new method of thermal network correction based on the statistical regression analysis
Eine neue Methode der Thermosystemkorrektur auf der Basis einer statistischen Regressionsanalyse
AIAA-Papers ; 1-9
1978
9 Seiten, 3 Bilder, 8 Tabellen, 8 Quellen
Aufsatz (Konferenz)
Englisch
Thermal network correction techniques
AIAA | 1970
|Thermal Network Correction Techniques
AIAA | 1971
|Thermal network correction by the optimization method
AIAA | 1986
|Differential correction schemes in nonlinear regression
TIBKAT | 1972
|