The Influence of Raw Data Selection on the Rain Attenuation Forecast of Grey System

In this study, the application of grey system theory in short-term prediction of rain attenuation is explored, with the objective of enhancing the accuracy and reliability of rain attenuation forecasting in the field of wireless communications. Based on the measured data from a rainfall event in 2021 at Qingdao Station, the GM(l,l) model within grey system theory is employed to forecast rain attenuation at the next 30s mark. Through systematic analysis of the influence of varying numbers of raw data on prediction results, it is found that selecting an appropriate number of raw data is crucial for improving prediction accuracy. Experimental results indicate that during periods of significant variation in rain attenuation, increasing the number of raw data can lead to abnormal values in prediction outcomes, thereby magnifying prediction errors. Consequently, using a smaller dataset for prediction effectively mitigates errors. By optimizing the program and eliminating outliers, the predicted values of rain attenuation at the next 30s mark generated by the grey system theory align closely with the actual values, validating the effectiveness and practicality of this method.