Application of neural network and its differential to the mutual coupling effects of sonar arrays

This paper contains two stages. In the first stage, the neural network is utilized to model and predict the mutual coupling effect of sonar array. In the second stage, the differential on such a neural network is utilized to optimize the mutual coupling effect of sonar array. The main advantage of the proposed method is that the differential of a neural network can be directly carried out from the original neural network. It is well known that the neural network requires only some training data sets and can predict the remaining data sets. Therefore, with the use of these two stages, one can easily optimize the mutual coupling effect of sonar array from some training data sets. Optimization can be obtained without knowing the whole distribution of the mutual coupling effect of sonar array.