【摘要】：論文主要研究了量子行為粒子群算法的改進(jìn)措施,提出了兩種改進(jìn)的算法,結(jié)合成品油配送方案設(shè)計(jì)問(wèn)題,將理論算法推向了實(shí)用。粒子群優(yōu)化是群智能算法的典型代表,而基于量子勢(shì)阱中粒子的動(dòng)態(tài)行為設(shè)計(jì)的量子行為粒子群算法,其優(yōu)良的性能已引起國(guó)內(nèi)外學(xué)者的廣泛關(guān)注,現(xiàn)已成為國(guó)際上的研究熱點(diǎn)。然而該算法也隸屬于群智能算法,并未擺脫所有群智能優(yōu)化所固有的易于陷入早熟收斂的弊端。因此研究該算法的改進(jìn)措施,在豐富群智能優(yōu)化理論和拓展群智能優(yōu)化應(yīng)用兩方面都將有重要意義。論文主要研究?jī)?nèi)容如下。第一,針對(duì)量子行為粒子群優(yōu)化在迭代過(guò)程中,粒子多樣性迅速下降,算法易收斂于局部最優(yōu)解的問(wèn)題,提出了一種基于選擇策略的量子行為粒子群算法。該算法同樣采用量子勢(shì)阱作為尋優(yōu)機(jī)制,但在勢(shì)阱中心的構(gòu)建方面,提出了新的建立方法。在每一步迭代中,取前K個(gè)最優(yōu)個(gè)體作為候選集,采用輪盤(pán)賭策略在候選集中選擇一個(gè)作為勢(shì)阱中心,調(diào)整其它個(gè)體向該中心移動(dòng)完成一步優(yōu)化。在優(yōu)化過(guò)程中使K值單調(diào)下降,以期達(dá)到探索和開(kāi)發(fā)的平衡。函數(shù)極值優(yōu)化的實(shí)驗(yàn)結(jié)果表明,該算法的優(yōu)化能力比原算法有明顯提高。第二,目前的量子行為粒子群算法采用實(shí)數(shù)編碼,搜索能力不夠理想。針對(duì)這一問(wèn)題,提出一種采用量子比特編碼的量子行為粒子群算法。該算法在Bloch球面建立搜索機(jī)制,用泡利矩陣建立旋轉(zhuǎn)軸,用Delta勢(shì)阱模型計(jì)算旋轉(zhuǎn)角度,用量子比特在Bloch球面上的繞軸旋轉(zhuǎn)實(shí)現(xiàn)搜索,用Hadamard門(mén)實(shí)現(xiàn)變異,以避免早熟收斂。該算法可增強(qiáng)對(duì)解空間的遍歷性,提高收斂概率。實(shí)驗(yàn)結(jié)果表明該算法的優(yōu)化能力優(yōu)于原算法。最后,針對(duì)當(dāng)前優(yōu)化算法在求解成品油配送優(yōu)化問(wèn)題方面存在的不足,本文研究了新算法在求解成品油配送車(chē)輛路徑優(yōu)化問(wèn)題上的工程應(yīng)用。本文將提出的算法應(yīng)用于求解大慶油田儲(chǔ)運(yùn)銷(xiāo)售分公司的成品油配送車(chē)輛路徑優(yōu)化問(wèn)題,該實(shí)例的對(duì)比結(jié)果表明,新算法的優(yōu)化結(jié)果明顯優(yōu)于該公司的現(xiàn)有設(shè)計(jì)結(jié)果。同時(shí)也證明了,新算法是求解成品油配送路徑優(yōu)化問(wèn)題的有效方法,對(duì)于今后解決類(lèi)似組合優(yōu)化問(wèn)題具有一定的參考價(jià)值。
[Abstract]:In this paper, the improvement measures of quantum behavior particle swarm optimization algorithm are studied, and two improved algorithms are proposed. Combined with the design of product oil distribution scheme, the theoretical algorithm is put into practice. Particle swarm optimization is a typical representative of swarm intelligence algorithm, and the excellent performance of quantum behavior particle swarm optimization algorithm based on the dynamic behavior of particles in quantum potential wells has attracted extensive attention of scholars at home and abroad. Now it has become a hot research topic in the world. However, the algorithm also belongs to the swarm intelligence algorithm, and does not get rid of the inherent disadvantage of all swarm intelligence optimization, which is easy to fall into premature convergence. Therefore, it will be of great significance to study the improvement measures of the algorithm in both enriching the theory of swarm intelligence optimization and expanding the application of swarm intelligence optimization. The main research contents of this paper are as follows. Firstly, in order to solve the problem that the particle diversity decreases rapidly and the algorithm converges to the local optimal solution in the iterative process of quantum behavior particle swarm optimization, a quantum behavior particle swarm optimization algorithm based on selection strategy is proposed. The algorithm also uses quantum potential well as the optimization mechanism, but in the construction of potential well center, a new method is proposed. In each iteration, the first K optimal individuals are taken as the candidate set, and the roulette strategy is used to select one as the center of the potential well in the candidate set, and the other individuals are adjusted to move to the center to complete the one-step optimization. In the optimization process, the K value decreases monotonously in order to achieve the balance between exploration and development. The experimental results of function extremum optimization show that the optimization ability of the algorithm is obviously higher than that of the original algorithm. Secondly, the current quantum behavior particle swarm optimization algorithm uses real number coding, and the search ability is not ideal. In order to solve this problem, a quantum behavior particle swarm optimization algorithm based on quantum bit coding is proposed. In this algorithm, the search mechanism is established in Bloch sphere, the rotation axis is established by Pauli matrix, the rotation angle is calculated by Delta potential well model, the search is realized by the rotation of quantum bits around the axis of Bloch sphere, and the mutation is realized by Hadamard gate in order to avoid premature convergence. The algorithm can enhance the ergodicity of the solution space and improve the convergence probability. The experimental results show that the optimization ability of the algorithm is better than that of the original algorithm. Finally, in view of the shortcomings of the current optimization algorithm in solving the optimization problem of refined oil distribution, this paper studies the engineering application of the new algorithm in solving the problem of vehicle routing optimization of refined oil distribution. In this paper, the proposed algorithm is applied to solve the routing optimization problem of finished oil distribution vehicles in Daqing Oilfield Storage and Transportation sales Branch. The comparison results of this example show that the optimization results of the new algorithm are obviously better than the existing design results of the company. At the same time, it is proved that the new algorithm is an effective method to solve the problem of product oil distribution path optimization, which has certain reference value for solving similar combinatorial optimization problems in the future.
【學(xué)位授予單位】：東北石油大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類(lèi)號(hào)】：TE83;TP18

【參考文獻(xiàn)】

相關(guān)碩士學(xué)位論文 前1條

1 楊路燕;基于風(fēng)險(xiǎn)分析的成品油二次配送路徑優(yōu)化研究[D];大連海事大學(xué);2014年

，

本文編號(hào)：2478299