| Popis: |
It is important and difficult for underwater weapons to get ship's structure parameters from limited physical field signals. The degree of recognition will directly affect weapons' attacking result. Nowadays, some researchers used acoustic model to recognize underwater object, however, few research about ascertaining these parameters such as object's length, width and tonnage have been found. In this paper, we proposed an improved Back Propagation (BP) neural network model that can escape local optimum thanks to optimizing the initial weight values and threshold values by Particle Swarm Optimization (PSO) algorithm to solve it. The method can study the relationship between the positions and values of magnetic field curve's extremums and structure parameters directly. Its high accuracy and good robustness have been validated by a test. |
