To address the issue of inaccurate prediction of radio interference in transmission lines under the influence of multiple environmental factors at high altitudes

this paper proposes a hybrid prediction model based on variational mode decomposition (VMD)

convolutional neural network (CNN)

bidirectional gated recurrent unit (BiGRU)

and attention mechanism. First

a long-term observation station was established at an altitude of 2 420 m to collect radio interference data together with surrounding environmental factors

and Pearson correlation analysis was employed to screen the relevant variables. Second

considering the nonlinear characteristics of radio interference data

VMD was applied to decompose the original signal into a series of relatively stable subsequences. Third

these subsequences were used as inputs to the combined CNN-BiGRU-ATT model. Meanwhile

to enhance the prediction performance

the whale optimization algorithm was utilized to optimize two hyperparameters of the BiGRU

yielding the optimal parameter configuration. Finally

the prediction results of each subsequence were superimposed and reconstructed to obtain the final radio interference prediction. The results indicate that the proposed method achieves root mean square error (RMSE)

mean absolute error (MAE)

and mean absolute percentage error (MAPE) of 2.134

1.125

and 0.594

respectively

in one-step prediction

and demonstrates superior accuracy in multi-step prediction compared with other models.