| The garbage cleaning unmanned vessel can realize autonomous cruise,identify garbage,and clear floating objects on the water surface,which is of great significance and function for saving manpower,consuming fuel resources and improving the environment of the reservoir.At present,the quality of the hull is too large,and the structural optimization is mostly aimed at local components.The performance study of the optimization agent model itself has little impact on the hull endurance and garbage salvage.This paper takes the hull structure of the garbage cleaning unmanned vessel as a research object,designs a hull structure that meets the needs of the operation,and compares the performance of different hidden layer BP neural networks with the goal of the lightest weight of the whole ship,and selects the double hidden layer BP neural network The structure of the whole ship has been optimized.The specific research contents are as follows:(1)Analyze the characteristics of the navigation area and the technical requirements of the project,refer to the current data of the ship type and the shipbuilding specifications,complete the hull structure design of the garbage cleaning unmanned vessel,and check the main performance of the hull to prove the rationality of the hull structure design.(2)Establish a three-dimensional finite element model of the hull and calculate the load according to the relevant CCS codes,determine the constraints and dangerous conditions,and analyze the stress and strain of the whole ship and hull members and the relatively weak link of the hull structure;The whole ship conducts modal analysis to obtain the natural vibration mode and frequency of the hull,and determines the vibration frequency forbidden zone of the ship.(3)Use the parameter test method to change the geometric dimensions of the main components,and perform sensitivity analysis on the mass of the ship,maximum equivalent stress,maximum shear stress,first-order natural frequency,and second-order natural frequency to select the design variables of the ship;Based on the analysis and screening components,establish an orthogonal table test table,carry out orthogonal test design,and obtain training and test samples for establishing a structure-optimized neural network.(4)Based on the mathematical model of structural optimization,four types of hidden layer BP neural networks were constructed with sample data of orthogonal experiments,and their fitting and errors were compared.The double hidden layer BP neural network was selected as the optimization of the whole ship structure Proxy model;determine design variables,constraints and objective functions of hull structure optimization;use Optimization tool to optimize the structure of the whole ship to achieve the optimization goal,and the quality of the whole ship will be reduced by10.3% after optimization;through finite element analysis and real ship The navigationtest proves that the optimized garbage cleaning unmanned vessel’s strength and natural frequency meet the requirements of the specification after optimization,and the stability and speed of the actual ship’s navigation meet the project requirements. |
