基于改进粒子群蚁群算法的多目标双边匹配问题研究

发布时间：2018-03-20 00:22

本文选题：双边匹配　切入点：蚁群算法　出处：《广西大学》2016年硕士论文　论文类型：学位论文

【摘要】：近年来我国电子商务发展迅猛,在中介模式的电子商务交易过程中,电子中介在尽可能满足买卖双方各自需求条件并考虑各方利益的情况下促成交易,该问题为电子商务中双边匹配问题。现实生活中存在着大量的双边匹配问题。目前解决双边匹配问题的多数算法都是将多目标问题转化为单目标问题来求解,但对同一问题不同的决策者往往会从不同的角度考虑。因此,本文建立一种多目标双边匹配问题模型,并结合改进的粒子群蚁群算法对其进行求解J主要工作内容如下：(1)针对粒子群算法在连续优化问题的搜索效率方面提出一种全局最优变异策略,并将其融入带收缩与发散操作的自适应粒子群优化算法当中,使得粒子整体保持高效搜索的同时及时逃离局部最优。之后用几组经典测试函数进行测试,实验结果表明该策略能够增强粒子的搜索功能,提高算法的寻优精度和速度。(2)针对蚁群算法收敛速度慢、易陷入局部最优的特点,设计一种完全自适应信息素调节多目标蚁群算法,通过动态调控蚁群系统信息素使得算法能够较快地收敛,避免算法停滞不前,所得多目标非劣解具有较好的收敛性且分布均匀。之后使用国际公认的TSP多目标测试函数对其进行测试,并对测试结果进行分析。(3)设计了一种针对双边匹配问题的满意度计算方法,该方法能更准确地反映满意程度情况,并在此基础上设计一类双边匹配问题数学模型。针对多目标双边匹配问题的特点将(1)(2)当中的粒子群算法与蚁群算法分别进行改进,并将二者有机结合用于求解己建立的双边匹配问题数学模型。实验结果表明改进后的多目标粒子群蚁群算法可以较好地解决该类双边匹配问题。
[Abstract]:In recent years, electronic commerce in China has developed rapidly. In the process of electronic commerce transaction of intermediary mode, electronic intermediary can meet the requirements of both parties as far as possible and take into account the interests of all parties to facilitate the transaction. This problem is a bilateral matching problem in electronic commerce. There are a lot of bilateral matching problems in real life. At present, most of the algorithms to solve the bilateral matching problem are to transform the multi-objective problem into a single-objective problem to solve the problem. However, different decision makers on the same problem tend to consider it from different angles. Therefore, a multi-objective bilateral matching model is established in this paper. Combining with the improved particle colony ant colony algorithm, the main work of solving J is as follows: 1) A global optimal mutation strategy is proposed for the search efficiency of the particle swarm optimization problem. It is integrated into the adaptive particle swarm optimization algorithm with shrinkage and divergence operations to make the whole particle keep efficient search and escape from the local optimum in time. Then several groups of classical test functions are used to test. The experimental results show that the strategy can enhance the searching function of particles and improve the accuracy and speed of the algorithm. (2) aiming at the characteristics of the ant colony algorithm, the convergence rate is slow and the ant colony algorithm is prone to fall into local optimum. A fully adaptive pheromone adjusting multiobjective ant colony algorithm is designed. By dynamically adjusting pheromone of ant colony system, the algorithm can converge quickly and avoid stagnation. The obtained multiobjective noninferior solution has good convergence and uniform distribution, and then it is tested by the internationally recognized TSP multiobjective test function. Finally, a satisfaction calculation method for bilateral matching problem is designed, which can reflect the degree of satisfaction more accurately. On this basis, a mathematical model of a kind of bilateral matching problem is designed. According to the characteristics of the multi-objective bilateral matching problem, the particle swarm optimization (PSO) algorithm and the ant colony algorithm (ACO) are improved respectively in accordance with the characteristics of the multi-objective bilateral matching problem. The experimental results show that the improved multi-objective particle colony ant colony algorithm can solve the two-sided matching problem.
【学位授予单位】：广西大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP18

【相似文献】