Prediction model for very high frequency communication interruption based on random forest regression algorithm

Abstract

Abstract The onboard Very High Frequency (VHF) communication is susceptible to interruption due to antenna radiation signal blockage caused by the aircraft body at certain angles resulting from changes in the aircraft’s attitude. To establish a prediction model for communication interruptions, data on the radiation power level, attitude angle 1, and attitude angle 2 of the antenna were initially collected. Subsequently, a machine learning model was constructed to predict the radiation power at two specific frequency points of a VHF communication antenna for helicopters. The model was built by using the random forest (RF) regression algorithm. For the prediction model of frequency point 1, the Goodness of Fit (R2) and Mean Squared Error (MSE) were 0.997 and 0.023, respectively, with feature importance scores of 0.98 for attitude angle 1 and 0.02 for attitude angle 2. For the prediction model of frequency point 2, the R2 and MSE were 0.999 and 0.008, respectively, with feature importance scores of 0.91 for attitude angle 1 and 0.09 for attitude angle 2. The study results indicate that the variation in attitude angle 1 significantly impacts the antenna’s radiation power. The RF regression model can effectively predict the onboard VHF communication capability, thereby providing technical support for reducing the potential risk of communication interruption.