Research On Multi Type Load Identification Of Unmanned Surface Vehicle

Unmanned surface vehicle plays an important role in promoting marine development,protecting marine environment and safeguarding marine safety.When performing the above tasks,the unmanned craft has high requirements for its own weight,strength,speed and its own physical field protection performance.Therefore,a large number of lightweight,low magnetic/non-magnetic materials have been widely used in unmanned surface vehicle.In the process of carrying out tasks,unmanned surface vehicle will inevitably suffer from a variety of random loads,such as wave load,wind load,current load,collision load,underwater explosion load and so on.General loads such as wind and current will not cause structural damage quickly,but when special loads such as collision load,underwater explosion load and wave load act on the hull structure,it is easy to cause structural damage and reduce the strength and stability.Such special loads are difficult to predict accurately,but they can be identified,recorded and analyzed by structural response monitoring combined with load identification technology.As an important part of structural health management,load identification can effectively monitor and evaluate the system state in real time and avoid major accidents,which is of great significance to the operational efficiency of unmanned craft.Therefore,aiming at the problem of hull structure load identification,this paper mainly focuses on the analysis,classification and load identification of explosion and impact response.The main research contents are as follows:(1)The classification of impact load of unmanned craft is studied.The basic principles of wavelet transform,BP neural network and one-dimensional convolution neural network are described.The numerical simulation of underwater explosion and impact response is carried out based on the unmanned boat model.The energy characteristics of wavelet packet are analyzed by using wavelet packet transform,and the recognition and classification research is carried out by using BP neural network and one-dimensional convolution neural network respectively.(2)Based on the similarity analysis theory,the impact load location identification method of hull structure is studied.The impact location identification accuracy of Euclidean distance,cosine similarity distance,Pearson correlation coefficient,Chebyshev distance,Manhattan distance and Minkowski distance as similarity parameters are systematically compared.(3)The time domain point impact load identification method is studied.The basic principle of time-domain point impact load identification method is introduced.The time-domain impact load mathematical model is established based on green kernel function method and Chebyshev orthogonal polynomial.Tikhonov regularization method and truncated singular value decomposition method are used to solve the load identification mathematical model,and the time history of point impact load is obtained.The feasibility of this method is verified by numerical simulation.(4)The time-domain distributed impact load identification method is studied.The basic principle of time-domain distributed impact load identification method is described.Considering the load characteristics of underwater explosion spherical shock wave,the time-domain distributed impact load mathematical model is established based on the simplified three-dimensional Chebyshev orthogonal polynomial.Tikhonov regularization method and truncated singular value decomposition method are used to solve the load identification mathematical model,and the spatial and temporal distribution of impact load are obtained.Numerical simulation is used to verify the feasibility of this method.