Research On Three-dimension Dynamic Risk Ma-Nagement Method Of Ship Lock Based On SVR And PSO-BP Algorithm

The ship lock is a sustainable water transport infrastructure for industrial and agricultural development in the region.It can not only ensure the channels flow,enhance the regional economy,but also contribute to water supply and power generation,flood prevention and drainage,agricultural irrigation,and prevention of soil erosion influences.Today,while certain achievements have been made in the construction and management of ship lock facilities,they also face challenges such as ship collision,deformation of main facilities,and aging of the project.Therefore,it is of great practical significance to carry out research on ship lock risk management.This paper mainly completes the following tasks:uses machine learning algorithms and three-dimensional visualization technology to carry out research on three-dimensional dynamic risk management of ship locks;establish a three-dimensional digital ground feature model of the study area;build a ship lock deformation prediction model;design deformation prediction and comprehensive risk prediction optimization algorithms;propose a ship lock safety risk management visualization methods;study the visual simulation of 3D dynamic monitoring of ship locks;develop a 3D dynamic risk management system for ship locks;realize the visualization and intelligence of ship lock risk management.The specific research contents are as follows:(1)According to the actual engineering situation,the ship lock deformation prediction model is established;Based on support vector regression machine(SVR),a prediction method of lock deformation is proposed to realize dynamic prediction of lock deformation trend.The superiority and generalization ability of the algorithm are verified by comparing with many machine learning algorithms.(2)Considering the complexity of the algorithm,the risk data is reduced in dimension based on the principal component analysis method to extract key risk indicators.Aiming at the shortcomings of the BP neural network algorithm,which is easy to fall into local minima and slow convergence,a particle swarm algorithm(PSO)optimized BP neural network algorithm for ship lock comprehensive risk prediction method.By optimizing the weight and threshold in the BP neural network algorithm to improve the accuracy of the algorithm;Combining the actual risk value to verify the accuracy and superiority of the prediction model;Based on risk evaluation indicators,establish a comprehensive risk assessment module for ship locks,and realize the visualization of ship lock risk management.(3)Based on the research of system and development goals,design the system technical framework and functional modules.Based on GIS secondary development components,3DS Max and other software,carry out research on the creation method of 3D digital feature models in the study area,and build digital terrain and ship lock building models fusion of 3 D scenes to complete the 3D visual simulation of the ship lock;Based on the Visual Studio development platform and C#language to complete the storage and use of spatial data;Attribute data to realize the visual simulation of the ship lock dynamic risk management and deformation risk early warning.