Грибова В. В., Клещев А. С., Inzartsev A., Pavin A., Eliseenko G.
OCEANS 2016 MTS/IEEE Monterey, OCE 2016
Inzartsev A., Pavin A., Kleschev A., Gribova V., Eliseenko G. Application of Artificial Intelligence Techniques for Fault Diagnostics of Autonomous Underwater Vehicles // Proceedings of the OCEANS 2016 MTS/IEEE Conference & Exhibition, Monterey, California, USA, September 19-23, ISBN 978-15090-1537-5. DOI: 10.1109/OCEANS.2016.7761098.
Autonomous underwater robotic vehicles (AUVs) are equipped with monitoring and emergency systems (MESs) to increase the mission success rate. The MES ensures the AUV's safety in water and the fault tolerance of its subsystems. The signals produced by the self-test functionality of robot subsystems as well as the parameters measured by sensors are the source information for the MES. Nowadays, MES actions in the AUVs designed by the Institute for Marine Technology Problems (IMTP) are based upon the hypothesis of the isolation of signals. As a result, the robot's reaction to several quasi-simultaneous signals is not always adequate. It is possible to overcome the disadvantages of the existing version of MES with the help of an ontological approach to be used for the implementation of an intelligent MES (IMES). An IMES functions as a tactic-level agent of the information and control system can carry out diagnostic actions (submissions) if necessary. The compiler forms the IMES with the help of a preset knowledge base and description of the robotic configuration with the application of the cloud platform IACPaaS. Knowledge is a semantic network. The architecture of the AUV's IMES consists of three basic blocks: a block for the formation of knowledge on AUV fault diagnostics, a translator that translates knowledge into program code, and the AUV MES itself. Extensive testing of IMES as part of AUV is planned soon. © 2016 IEEE.