Research On Situation Awareness,Threat Assessment And Autonomous Decision Method Of UUV Under Uncertain Conditions

Currently,various sensors,such as intelligence gathering,anti-submarine warfare,marine environment survey,and seabed topography mapping,have been employed in the unmanned underwater vehicles(UUV's)to perform military and civilian tasks.As a new concept,the intelligent underwater platform is capable of autonomous remote navigation and operations under unmanned control.Situation awareness,threat assessment,and autonomous decision-making are the critical guarantees for UUV's autonomous operation under unmanned on-site control and the essential manifestation of autonomy of the UUV under the complex and unknown marine environment.Therefore,studying the technical problems of situation awareness,threat assessment,and autonomous decision-making of the UUV has outstanding theoretical research value and engineering application significance.This paper mainly implements research work in the following areas:First,when the UUV performs tasks in a complex marine environment,the sensor is affected by different factors such as different depths,water temperature,and noise,making uncertainty in the detected sensing information.This uncertainty has an essential impact on the autonomous operation of the UUV.According to the types of UUV sensors and measurement information,a UUV situation awareness model is constructed.The potential uncertain events in the UUV situation awareness model is analyzed to establish an ontology model of an uncertain event using ontological methods.Now,semantic rules triggering uncertain events are designed,and a rule engine reasoning mechanism is utilized to update the ontology model of the uncertain event.Then,the uncertain event ontology is transformed into a Bayesian network.The influence degree of the uncertain event is obtained through Bayesian network reasoning,and the validity of the ontology method is verified through simulations.Second,a dynamic Bayesian network-based UUV threat assessment algorithm is proposed to effectively and accurately evaluate the UUV threat in a complex marine environment.According to the characteristics of the threat elements of each kind of uncertain event,the threat feature variables are extracted to determine the dynamic Bayesian network.Network parameters are determined by combining expert domain knowledge with genetic optimization algorithms,which capable of solving the problem of parameters with strong subjectivity and inaccurate reasoning.The superiority and effectiveness of dynamic Bayesian networks to the Bayesian network for uncertain reasoning are verified.Furthermore,the dynamic impact graph and expected utility theory are employed to propose an autonomous decision-making method for solving the problem of the UUV's autonomous dynamic decision-making in an uncertain marine environment.According to the threat assessment results,a dynamic impact graph model of UUV autonomous decision-making based on dynamic Bayesian network is constructed,which includes the probability model,the decision-making rule model,and the expected utility model.The maximum expected utility principle is employed to design the utility calculation model of a dynamic impact graph for achieving the best decision-making.The effectiveness and dynamics of the employed dynamic impact graph in autonomous decision-making of the UUV are verified through simulations.Finally,two typical mission tasks,including underwater area search and cruise are designed according to the requirements of UUV's operation tasks,and the trigger elements of uncertain events are chosen according to time series simulation.The effectiveness of situation awareness,threat assessment,and autonomous decision-making methods of the UUV is investigated via simulations.The results indicated that the UUV could complete threat perception and threat level assessment during task execution.Moreover,it can generate useful and reasonable autonomous decision-making strategies,which can effectively guarantee the UUV's operation safety.