基于粒子群算法和支持向量機(jī)的船舶結(jié)構(gòu)優(yōu)化
發(fā)布時間：2019-01-30 21:31
【摘要】：船舶結(jié)構(gòu)優(yōu)化是船舶設(shè)計的重要方面，其主要目的在于以尋求優(yōu)化的結(jié)構(gòu)形式。由于船舶結(jié)構(gòu)優(yōu)化過程中涉及的設(shè)計變量數(shù)目眾多、種類多樣、所受的約束條件復(fù)雜，這導(dǎo)致了目標(biāo)函數(shù)的非線性程度強(qiáng)，很難尋求到優(yōu)化問題的最優(yōu)解。需選擇合適的優(yōu)化算法進(jìn)行結(jié)構(gòu)優(yōu)化，粒子群算法作為一種新型的智能算法，可實現(xiàn)性強(qiáng)，收斂性好，有優(yōu)秀的全局搜索能力。本文將粒子群算法應(yīng)用于優(yōu)化問題之中，先以三個經(jīng)典的桁架結(jié)構(gòu)驗證了粒子群算法由于結(jié)構(gòu)優(yōu)化的有效性，在此基礎(chǔ)之上提出MATLAB粒子群算法工具箱和有限元程序相結(jié)合應(yīng)用于結(jié)構(gòu)優(yōu)化的技術(shù)路徑，建立了一三艙段結(jié)構(gòu)有限元模型，將上述優(yōu)化路徑用于結(jié)構(gòu)優(yōu)化，，得到良好的優(yōu)化結(jié)果，驗證單目標(biāo)粒子群算法應(yīng)用于船舶結(jié)構(gòu)優(yōu)化的可行性。 船舶結(jié)構(gòu)優(yōu)化過程中，往往需要調(diào)用有限元軟件進(jìn)行迭代計算，以獲得結(jié)構(gòu)響應(yīng)作為優(yōu)化過程中的目標(biāo)函數(shù)或者約束條件，而因為船舶結(jié)構(gòu)優(yōu)化的復(fù)雜性，迭代次數(shù)會較大，這使得船舶結(jié)構(gòu)優(yōu)化過程需要消耗比較大的時間成本。在優(yōu)化過程之中借助近似模型可以減少優(yōu)化所需的時間成本，提升優(yōu)化效率，支持向量機(jī)作為一種有效的近似模型，可以對各種復(fù)雜非線性問題進(jìn)行回歸。在結(jié)構(gòu)優(yōu)化問題中，其可以用來建立結(jié)構(gòu)響應(yīng)近似模型，以預(yù)測結(jié)構(gòu)響應(yīng)、代替復(fù)雜費(fèi)時的有限元計算。支持向量機(jī)的參數(shù)選取是支持向量機(jī)應(yīng)用的難點之一，一般的基于經(jīng)驗的方法很難尋求到適合特定問題的支持向量機(jī)參數(shù)，本文將支持向量機(jī)參數(shù)的選取抽象為一優(yōu)化問題，建立了優(yōu)化的方法尋求支持向量機(jī)參數(shù)的方法，為支持向量機(jī)參數(shù)的選取找到了一條切實可行的路徑，利用粒子群算法選取支持向量機(jī)的參數(shù)，得到了具有最優(yōu)參數(shù)的支持向量機(jī)近似模型，并與基于經(jīng)驗參數(shù)的支持向量機(jī)做了對比，以驗證本文所提的參數(shù)選取方法的有效性。 在近似模型的參數(shù)選卻基礎(chǔ)之上，本文提出了基于支持向量機(jī)和粒子群算話的結(jié)構(gòu)優(yōu)化方法，在支持向量機(jī)參數(shù)選取方法的基礎(chǔ)之上，建立支持向量機(jī)近似模型，并與粒子群算法相結(jié)合，用于結(jié)構(gòu)優(yōu)化，為驗證上述方法的有效性，利用上述方法對船舶結(jié)構(gòu)進(jìn)行優(yōu)化。
[Abstract]:Ship structure optimization is an important aspect of ship design. Due to the large number of design variables involved in the process of ship structure optimization, the variety of design variables and the complexity of constraints, this leads to a strong degree of nonlinearity of the objective function, and it is difficult to find the optimal solution of the optimization problem. As a new kind of intelligent algorithm, particle swarm optimization has the advantages of strong realizability, good convergence and excellent global searching ability. In this paper, particle swarm optimization (PSO) algorithm is applied to the optimization problem. Firstly, three classical truss structures are used to verify the effectiveness of PSO due to structural optimization. On the basis of this, the technical path of the combination of MATLAB particle swarm optimization toolbox and finite element program for structural optimization is proposed, and the structural finite element model of the first and third cabins is established, and the above optimized path is applied to the structural optimization. Good optimization results are obtained, and the feasibility of applying single objective particle swarm optimization algorithm to ship structure optimization is verified. In the process of ship structure optimization, it is often necessary to use finite element software for iterative calculation to obtain structural response as the objective function or constraint condition in the optimization process. However, because of the complexity of ship structure optimization, the iteration times will be larger. This makes the ship structure optimization process requires a relatively large time cost. In the process of optimization, the time cost of optimization can be reduced and the optimization efficiency can be improved by using approximate model. As an effective approximate model, support vector machine (SVM) can be used to regress various complex nonlinear problems. In the structural optimization problem, it can be used to establish the approximate model of the structural response to predict the structural response, instead of the complicated and time-consuming finite element calculation. The parameter selection of support vector machine is one of the difficulties in the application of support vector machine. It is difficult to find support vector machine parameters suitable for a specific problem in general experience-based methods. In this paper, the selection of support vector machine parameters is abstracted as an optimization problem. The optimization method is established to find the parameters of support vector machine. A feasible path is found for the selection of support vector machine parameters. The particle swarm optimization algorithm is used to select the parameters of support vector machine. The approximate model of support vector machine with optimal parameters is obtained and compared with that of support vector machine based on empirical parameters to verify the effectiveness of the proposed parameter selection method. On the basis of parameter selection of approximate model, a structure optimization method based on support vector machine (SVM) and particle swarm optimization (PSO) is proposed in this paper. On the basis of parameter selection method of support vector machine (SVM), an approximate model of support vector machine (SVM) is established. The method is combined with particle swarm optimization algorithm to optimize the structure of ships. To verify the effectiveness of the above methods, the above methods are used to optimize the ship structure.
【學(xué)位授予單位】：上海交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：U662;TP18
本文編號：2418545
[Abstract]:Ship structure optimization is an important aspect of ship design. Due to the large number of design variables involved in the process of ship structure optimization, the variety of design variables and the complexity of constraints, this leads to a strong degree of nonlinearity of the objective function, and it is difficult to find the optimal solution of the optimization problem. As a new kind of intelligent algorithm, particle swarm optimization has the advantages of strong realizability, good convergence and excellent global searching ability. In this paper, particle swarm optimization (PSO) algorithm is applied to the optimization problem. Firstly, three classical truss structures are used to verify the effectiveness of PSO due to structural optimization. On the basis of this, the technical path of the combination of MATLAB particle swarm optimization toolbox and finite element program for structural optimization is proposed, and the structural finite element model of the first and third cabins is established, and the above optimized path is applied to the structural optimization. Good optimization results are obtained, and the feasibility of applying single objective particle swarm optimization algorithm to ship structure optimization is verified. In the process of ship structure optimization, it is often necessary to use finite element software for iterative calculation to obtain structural response as the objective function or constraint condition in the optimization process. However, because of the complexity of ship structure optimization, the iteration times will be larger. This makes the ship structure optimization process requires a relatively large time cost. In the process of optimization, the time cost of optimization can be reduced and the optimization efficiency can be improved by using approximate model. As an effective approximate model, support vector machine (SVM) can be used to regress various complex nonlinear problems. In the structural optimization problem, it can be used to establish the approximate model of the structural response to predict the structural response, instead of the complicated and time-consuming finite element calculation. The parameter selection of support vector machine is one of the difficulties in the application of support vector machine. It is difficult to find support vector machine parameters suitable for a specific problem in general experience-based methods. In this paper, the selection of support vector machine parameters is abstracted as an optimization problem. The optimization method is established to find the parameters of support vector machine. A feasible path is found for the selection of support vector machine parameters. The particle swarm optimization algorithm is used to select the parameters of support vector machine. The approximate model of support vector machine with optimal parameters is obtained and compared with that of support vector machine based on empirical parameters to verify the effectiveness of the proposed parameter selection method. On the basis of parameter selection of approximate model, a structure optimization method based on support vector machine (SVM) and particle swarm optimization (PSO) is proposed in this paper. On the basis of parameter selection method of support vector machine (SVM), an approximate model of support vector machine (SVM) is established. The method is combined with particle swarm optimization algorithm to optimize the structure of ships. To verify the effectiveness of the above methods, the above methods are used to optimize the ship structure.
【學(xué)位授予單位】：上海交通大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：U662;TP18
【參考文獻(xiàn)】
相關(guān)期刊論文 前10條
1 孫文彩;楊自春;;結(jié)構(gòu)非概率可靠性分析的支持向量機(jī)分類方法[J];工程力學(xué);2012年04期
2 孫利;王德禹;;支持向量機(jī)和遺傳算法組合策略的VLCC船中結(jié)構(gòu)優(yōu)化設(shè)計(英文)[J];Journal of Marine Science and Application;2012年01期
3 孫光永;李光耀;鐘志華;張勇;;基于序列響應(yīng)面法的汽車結(jié)構(gòu)耐撞性多目標(biāo)粒子群優(yōu)化設(shè)計[J];機(jī)械工程學(xué)報;2009年02期
4 朱家元,楊云,張恒喜,任博;支持向量機(jī)的多層動態(tài)自適應(yīng)參數(shù)優(yōu)化[J];控制與決策;2004年02期
5 袁小芳;王耀南;;基于混沌優(yōu)化算法的支持向量機(jī)參數(shù)選取方法[J];控制與決策;2006年01期
6 劉昌平;范明鈺;王光衛(wèi);馬素麗;;基于梯度算法的支持向量機(jī)參數(shù)優(yōu)化方法[J];控制與決策;2008年11期
7 段雪厚;王石剛;徐威;唐成龍;;基于徑向基神經(jīng)網(wǎng)絡(luò)的薄板平整軋制力預(yù)報模型[J];上海交通大學(xué)學(xué)報;2011年06期
8 王安麟;董亞寧;周鵬舉;吳小鋒;周成林;;面向液壓滑閥卡滯問題的健壯性設(shè)計[J];上海交通大學(xué)學(xué)報;2011年11期
9 ;Parameter selection of support vector machine for function approximation based on chaos optimization[J];Journal of Systems Engineering and Electronics;2008年01期
10 金偉良;袁雪霞;;基于LS-SVM的結(jié)構(gòu)可靠度響應(yīng)面分析方法[J];浙江大學(xué)學(xué)報(工學(xué)版);2007年01期
本文編號：2418545
本文鏈接：http://sikaile.net/kejilunwen/chuanbolw/2418545.html
最近更新
教材專著
