Research On Trajectory Tracking Of Underwater Vehicle Aquaculture Cage Inspection Based On Model Prediction

In recent years,due to the gradual decline of fishery resources,the development of marine fisheries had gradually shifted from traditional fishing techniques to aquaculture technology.The vigorous development of deep-sea cage aquaculture had created a new situation of marine fishery aquaculture in the world.With the change of breeding methods,the artificial method of cage detection based on experience,physical strength and traditional equipment had been gradually phased out.And the inspection methods were constantly updated,orientated in intelligence.Using intelligent equipment such as underwater vehicles to inspect aquaculture cages is an inevitable trend in the future,which has important practical significance for the detection of the growth state of fish in the cages and the maintenance of the surrounding environment of the cages.Scholars of domestic and overseas had done a lot of work on the damage detection of net clothes and biological attachment corrosion detection of net cage by using visual recognition,image processing and other technologies,but there were few studies on using the characteristics of underwater robot for motion control and realizing the inspection track tracking of breeding net cage.Model Predictive Control(MPC)had the advantages of multiple inputs and multiple outputs,being able to incorporate constraints,and low requirements for model accuracy.Therefore,under the condition of rolling optimization,the uncertain factors caused by model mismatch and interference could be solved quickly.In order to realize the intelligent inspection of aquaculture cages,to improve the work efficiency of cage inspections,and to reduce the economic losses of farmers,In this paper,ROV(Remotely Operated vehicles)combined with MPC method were used to carry out relevant research and experiments.The specific work contents were as follows:(1)Modeling the ROV was the basis and premise of studying ROV motion control.In order to design the subsequent controller and carry out the simulation experiments smoothly,the kinematic model and dynamic model of 6-DOF ROV were constructed.Firstly,the geodetic coordinate system and the fuselage coordinate system were established.Secondly,the transformation relationship between the coordinate systems was established,and then the motion modeling and force analysis of the ROV were carried out.In view of the actual situation,the marine environment and experimental conditions were relatively harsh,and it was dangerous to carry out real ROV experiments.And due to the difficulties in operation,a virtual simulation model of ROV for the experimental environment and patrol inspection of virtual underwater aquaculture cages was constructed.(2)Aiming at the problems of high cost and low efficiency of manual regular inspection of breeding cage,a method of inspection track tracking control of ROV breeding cage based on MPC was proposed.Firstly,according to the physical and kinematic constraints of the ROV,combined with the ROV kinematic model and dynamic model,a MPC controller based on speed control was designed.Using the traditional PID(Proportion Integration Differentiation)control algorithm as the baseline,a model-free PID controller was designed.Secondly,the simulation experiment of two-dimensional horizontal plane trajectory and three-dimensional space trajectory tracking in the cage culture environment was carried out and a comparison was made.Finally,the experimental results showed that the proposed method had the advantages of good trajectory tracking performance,stable ROV operation and small fluctuation.(3)In view of the difficulties of ROV manual operation and the complex environment around the breeding cage in actual situation,an inspection track tracking control method of ROV monocular vision breeding cage based on MPC was proposed.Firstly,the pipeline was laid around the virtual breeding cage environment.Secondly,the pipeline information was obtained by loading ROS(robot operating system)virtual camera with ROV.Then,the reconstruction process from two-dimensional plane pipeline image information to three-dimensional space was realized through image pre-processing,pipeline edge extraction,feature point marking and coordinate conversion.The obtained trajectory information was sent to MPC controller,and then the trajectory tracking control of monocular vision cage inspection was carried out.Next,the interference factors such as ocean current were included in the experiment.Finally,the simulation results showed that the proposed method had good robustness,feasibility and anti-interference.