中文摘要：

声先偏转是一种高速非机械光束方向控制技术，近年来在卫星光通信领域日益受到重视。运用径向基（RBF）神经网络模型实现声光偏转器光束方向控制，通过训练对RBF网络的权值和网络层元数等结构参数进行优化，调整得到理想模型。研究结果验证了该模型的可靠性与准确性。

英文摘要：

As a high-speed non-mechanical beam steering technology, acousto-optic deflection has attracted more and more attention in the field of satellite optical communication recently. The beam steering control algorithms of acousto-optic deflector are based on a radial basis function （RBF） neural model. The ideal control model is established by optimizing the parameters and structure of the RBF neural network through training. Study results show its reliability and accuracy of the proposed control model.