论文信息 - External Balance Calibration Curve Prediction Using Polynomial and MLP Artificial Neural Network

External Balance Calibration Curve Prediction Using Polynomial and MLP Artificial Neural Network

The central aim of this study is improve the accuracy of Calibration Curves employing Polynomial and Multi Layer Perceptrons Neural Networks. The system to be calibrated consists of the external aerodynamic balance of the subsonic wind tunnel number 2, the TA2, of the Institute of Aeronautics and Space, Brazil. The process involves the estimation of the calibration parameters and two subsequent calibrations using the same configurations, where the prediction under repeatability conditions of this combination is studied. Previous studies employing the MLP have shown that this class of artificial neural network can be trained to fit an external balance calibration curve with arbitrary levels of accuracy, but this behavior is not so for temporal predictions. Nevertheless, as the MLP convergence for a particular calibration data set improves, its capacity for predicting future calibrations worsens. The deviations of a polynomial are presented to the output of the MLP in the learning process and the outputs of both are combined in order to reduce the resulting errors. The impact of the employment of the combination of different polynomials and

I. M. Barbosa | E. del Moral Hernandez

[1] P. R. Bevington,et al. Data Reduction and Error Analysis for the Physical Sciences , 1969 .

[2] Emilio Del Moral Hernandez,et al. Calibration Curve of a Multi-Component Balance Using MLP Artificial Neural Network with Learning Endowed with Loading Uncertainty , 2006 .