Particle Swarm Optimization And Its Application In Engineering

Stimulated by the innovation-driven development strategy of the 13 th Five-year Plan,scientific and technological innovation has become the necessary means to accelerate the innovation of China.The ability of the high precision industrial production and the independent research and development is becoming more and more important to the modern manufacturing industry,which has greatly promoted the development of optimal technology related to the field of precision manufacturing.The demand for the optimization change constantly and the difficulty of the optimization increases unceasingly.Therefore,an intelligent particle swarm optimization(PSO)algorithm with fewer parameters,and lower requirements for mathematical properties of the objective functions emerges at a historic moment.The above optimization algorithm can be used in engineering practice to optimize the process parameters,design lightweight structure,design optimal route,combine the optimization algorithm and the special algorithm to improve calculation results and so on.This dissertation is supported by National Natural Science Foundation of China “Study on rapid optimization of cross section parameters of multi body frame structure based on reanalysis theory”(No.: 50975121),Changchun science and Technology major support program “Analysis of Welding Deformation in the Bogie Frame of a High-speed Rail Passenger Car CRH3-350”(No.10KZ03),and Jilin science and technology innovation fund for small and medium sized enterprises “Design and Experiment for Grain Storage Monitoring System based on 3-D Laser Scanning Technology”(No.20130522150JH)to research PSO algorithm and its application in engineering.The main work includes:Firstly,the optimization theory and algorithm are analyzed and classified.According to the application of PSO algorithm in engineering,the background of the algorithm and the process design ideas of the standard PSO,discrete particle swarm optimization(DPSO)and Multi objective particle swarm optimization algorithm(MOPSO)algorithm are introduced.The parameter setting method of PSO algorithm and the improved method are analyzed.On this foundation,the improved discrete PSO based on the conception of stack and pointer and the hybrid algorithm combined the PSO algorithm with the point cloud data processing algorithm are put forward.Furthermore,the improved algorithms are verified in different enginnering cases in the follow-up research.Secondly,the improved DPSO algorithm is employed to find the optimal welding sequence and direction of the bogie frame of a high-speed rail passenger car,which lead to the minimum residual deformation and stress.The thermo-mechanical coupling simulation analysis method is used in ANSYS software to carry out numerical simulation.In order to minimize the number of combinations of welding deformation and stress,based on the concept of the pointer and the stack,the design of welding sequence and direction optimization test is presented.The mathematical model is established based on the experimental design method.Through the optimization of the improved DPSO algorithm proposed in this dissertation,the pointer with the optimal welding sequence and direction can be found.The method greatly reduces the computational expense needed in the weld process without an obvious loss of accuracy.The optimal solution effectively reduces the welding deformation and stress in the welding process,which has a guiding role in inproving the welding process,manufacturing quality and service performance.of the side beam of the high-speed rail passenger car.Thirdly,based on the concept of "stack and pointer",an improved multi-objectiveparticle swarm optimization algorithm is proposed.The improved algorithm is employed to study the influence of T joint model on the welding residual stress and strain.On the basis of the optimization design of the welding process of of the bogie frame of a high-speed rail passenger car mentioned above,the welding residual stress and strain are studied by using a simpler and more general T joint model.The optimized welding parameters are increased from welding sequence and welding direction to welding current,voltage,speed and discrete parameters.Furthermore,the factors that affect the welding residual stress and deformation are studied by extending the optimization targets from the single target to the multiple target.Results show,that welding parameters and welding sequence have a strong influence on residual welding stress and deformation,and welding parameters have a great influence on the welding heat input.In addition,the welding sequence and welding direction have great influence on the control of welding residual deformation and welding residual stress.A reasonable welding parameters and welding sequence can be chosen according to the results of Pareto front for diferent resuirements.Fourth,for laying the foundation of the improved point cloud data processing algorithm,the dissertation introduces the on-line monitoring system of grain reserves based on 3D laser scanning technology and the general point cloud data processing method.This dissertation firstly introduces the composition of the system,and the function of each part and the operating mode in detail.On this basis,the dissertation then presents the experimental process for collecting the grain surface point cloud data and the general point cloud data processing methods,i.e.,coordinate transformation,data simplification,de-noising processing,surface reconstruction,and grain volume calculation.Nevertheless,some shortages exist in those data processing methods,which need to be further studied.Fifth,a point cloud data simplification algorithm based on particle swarm optimization and a classified denoising method are proposed.The simplification algorithm can adaptively determine the simplification threshold of the average reduction method,which solves the problem of large amount and uneven distribution of the point cloud data by using the general algorithm.By using the algorithm,the distribution of the point cloud data is more uniform,and the redundant data are reduced more than the average distance method.The relative error of the grain volume obtained by this algorithm is lower than the relative error of the grain volume with the average simplification method.The point cloud data denoising algorithm is divided into three categories according to the noise types of point cloud data.The first and second types of noise points are removed by using the grid method.For the third types of noise points,the wavelet denoising method is used to remove the mixed noise points.By comparing the subjective and objective criteria of the point cloud data of the proposed method with the average denoising and median denoising methods,it can be concluded that the proposed method preserves the most point cloud data information and has a higher signal noise ratio(SNR),peak signal to noise ratio(PSNR)and lower root mean square error and(RMSE).Finally,the dissertation makes a systematic summary and puts forward the problems to be further improved.