Application of CNN in Active Surface Adjustment of Large Reflector Antennas

In view of the complex structure and multi-factor influence of large reflector antennas, an active adjustment analysis method based on deep learning is proposed in this paper. By constructing neural network model and emphasizing layer module, the adaptability and prediction accuracy of the model are improved. At the same time, a data preprocessing mechanism is designed to reorganize disordered data into ordered data to meet the needs of model learning. Taking 8-meter antenna as an example, through simulation training, the average adjustment accuracy of 48 actuators reached 0.001597mm and the final panel accuracy reaches 0.004936mm. This result provides an intelligent scheme for the active surface adjustment in engineering practice considering various influencing factors.