## Abstract

A geophysical inversion information based underwater object detection method is proposed by using the joint Gravity-Gradient and Magnetic-Gradient Inversion algorithms. The gravity-gradient and magnetic-gradient inversion equations are combined to estimate the orientation and distance of the underwater object. After calculating the relative positions of underwater object from the gravity-gradient inversion equations and magnetic-gradient inversion equations, the BP Neural Network is exploited to obtain an optimal geophysical inversion equation applied to underwater object detection. A typical three layered neural network of 6 input and 3 output neurons with a single hidden layer is constructed to realize information fusion. The leading characteristics of such neural network are strong parallel computing, learning and adaptive capabilities, as well as good fault-tolerance. With the proposed method, the trajectories of an underwater object can be detected accurately. Simulation results show that our method is more efficient than the joint gravity-gradient and magnetic-gradient inversion methods.

Original language | English |
---|---|

Title of host publication | Autonomous Underwater Vehicles |

Subtitle of host publication | Dynamics, Developments and Risk Analysis |

Publisher | Nova Science Publishers, Inc. |

Pages | 1-24 |

Number of pages | 24 |

ISBN (Electronic) | 9781536118315 |

ISBN (Print) | 9781536118193 |

Publication status | Published - 1 Jan 2017 |

Externally published | Yes |

## Keywords

- BP neural network
- Gravity gradient inversion
- Magnetic gradient inversion
- Underwater object detection

## ASJC Scopus subject areas

- Engineering (all)
- Computer Science (all)