基于扰动收缩粒子群算法的物联网配送车辆调度

Scheduling Distribution Vehicles in Internet of Things Based on Perturbation Contraction Particle Swarm Optimization

为了提高物联网配送车辆的调度效率,采用扰动收缩粒子群算法。首先建立物联网配送车辆优化调度问题的数学模型,考虑到货物品种及数量、需求时间和地点、运输线路以及运输时间的不确定性,包括运输成本、时间惩罚成本、固定成本;接着对基本粒子群算法增设非线性扰动因子用来平衡粒子的全局和局部搜索,在进化前期值比较小,让粒子主要进行局部搜索,而在后期设置值比较大,进行全局搜索,同时增设收缩算子,避免粒子的过度振荡,粒子编码涉及到收货点、车辆编序、行驶顺序,给出了算法流程;最后,仿真试验和实例分析验证了算法的合理性与可行性。结果表明:增设收缩算子对任务目标点寻优地理位置偏差值最小,避免了总成本增加;带有非线性扰动因子调整策略的粒子群优化算法具备更强的跳出局部最优的能力,优化后的算法运行速度加快;对于每次试验的搜索成功率以及违约惩罚成本占总成本比例,与遗传算法、蚁群算法、粒子群算法、混沌量子粒子群算法、模拟退火粒子群算法和柯西变异粒子群算法预测方法相比,扰动收缩粒子群算法预测方法具有更高的搜索成功率和较低的违约惩罚成本,能够满足物联网配送车辆系统对预测精度的需求,对实现实时交通控制具有重要意义。

In order to improve the efficiency of distribution vehicle scheduling by internet of things, perturbation contraction particle swarm optimization is used. First, the mathematic model for optimization of vehicle scheduling in Internet of Things is established. It considers the uncertainty of the variety and quantity of goods, the time and place of demand, the route and time of transport, including transport cost, time penalty cost and fixed cost. Second, the non-linearly perturbation factor is added to the basic particle swarm optimization to balance the global and local search of the particles, the initial value of evolution is smaller, the particle is mainly carried out local search, while the latter setting value is larger for global search, and contraction operator is added to avoid the excessive oscillation of particles, the particle coding involves receiving point, vehicle sequencing and driving sequence, and the algorithm flowchart is given. Finally, the rationality and feasibility of the algorithm are verified by simulation experiment and case analysis. The result shows that(1) the added shrinkage operator can minimize the deviation of the location optimization of the task target point, which avoids the increase of the total cost;(2) the particle swarm optimization with the adjustment strategy of non-linear disturbance factor has a stronger ability to jump out of the local optimum, and the optimized algorithm runs faster;(3)for the search success rate of each experiment and the proportion of default penalty cost to total cost, compared with GA, AC, PSO, CQPSO, SAPSO, CMPSO, the proposed PCPSO prediction method has higher search success rate and lower default penalty cost, it can also satisfy the demand of prediction accuracy of Internet of Things distribution vehicle system, which is of great significance for real-time traffic control.