A BPNN-based method for modeling temperature characteristics of a radio frequency power amplifier

摘要

The present invention discloses a BPNN-based method for modeling temperature characteristics of a radio frequency power amplifier, comprising: selecting training data and test data of a BPNN model according to measured data of temperature characteristics of the radio frequency power amplifier obtained; determining input variables and output variables of the BPNN model; importing the training data into the BPNN model and and training the BPNN model; importing the test data into the trained BPNN model, comparing output results and test results of the BPNN model and calculating an error of the two; comparing the magnitudes of the test error MSB of the BPNN model and a precision expectation value of the BPNN model, wherein if the test error MSB is less than the precision expectation value, training of the BPNN model is completed; if the test error MSB is greater than the precision expectation value, retraining is performed by adjusting parameters until the MSB is less than the precision expectation value and thus the training ends. The present invention applies the BPNN to realize prediction of the change in performance indicators of the radio frequency power amplifier within a given temperature range.