Research On Prediction Method Of Ship Impact Environment Based On Probabilistic Neural Network

Science and technology are primary productive forces,technological innovation is the liberation of human labor.Computer instead of human beings to do high-cost experimental research and cumbersome scientific engineering calculation is the inevitable trend of scientific research.The calculation and analysis of the impact environment under the underwater non-contact explosion load is the basis of the anti-explosion and anti-shock research of surface ships.How to carry out research on ship impact environment with low cost and high efficiency is the starting point of this paper.The author firstly carries out secondary development of ANSYS to realize parametric rapid modeling of ships,solved the problem that the existing ship has fewer models,single sample and uneven distribution of main scale parameters.Quickly establish a number of virtual ship,fill the gap between the real models,and then establish a ship impact environment database.On this basis,the author builds a ship impact environment prediction model based on probabilistic neural network.And the particle swarm algorithm and ant colony algorithm are used to optimize the network structure,and the rapid prediction and analysis of ship’s spectral velocity,spectral displacement and spectral acceleration are realized.The specific work of this paper is as follows(1)For the structural characteristics of the hull,the secondary development of ANSYS is carried out using the parameterized design language APDL.The material and section properties are defined in detail,the hull profile design based on the parent transformation method,and the main structure of the hull are created.And the rapid creation of modeling processes such as attribute assignment to meshing,enabling rapid parametric modeling of ships.(2)Using the user interface design language UIDL,customized visual parameter assignment menu system for ship parameterized rapid modeling,and the operation method of the parameterized rapid modeling interface is introduced in detail.The modal analysis of the rapid established ship model is given,and its natural frequency and mode shape are given to verify the effectiveness of the parameteic rapid modeling method.(3)Three-dimensional modeling of 11 ships and numerical simulation of impact environment has been processed,batch processing of data was done through Python programming and extracted of feature parameters,and then established the database of ship impact environment,which is used for intelligent computing.Explain the mathematical principles and network structure of probabilistic neural network,particle swarm optimization and ant colony algorithm.At the same time,the feasibility of the probabilistic neural network model is verified by a simple example.(4)Training probabilistic neural network models for ship impact environment prediction,the key parameters of probabilistic neural network are optimized by particle swarm optimization and ant colony algorithm.Compare and analyze the impact of factors such as sample size,impact factor and assessment location on the impact environment,and compare the impact environment of multiple ships in the database.Finally,the intelligent prediction model completed by training is used to forecast the impact environment of the ship,and a higher forecasting accuracy is obtained.