Models And Algorithms For Multiobjective Transportation Optimization Problems

This thesis studies two classes of decision-making optimization problems widely concerned in the modern logistics industry;transportation service procurement problems and online-to-offline(O2O)food pickup and delivery problems.With the rapid development of trade globalization,transportation plays a critical role in global trade,and outsourcing transportation has become a very common practice in the modern logistics industry.A global company(shipper)often uses reverse auctions to purchase transportation service from global transportation service providers(carriers)to meet its shipping requirements.Optimized transportation service procurement solutions help shippers save transportation costs before any freight is moved in transportation operations.In addition to transportation costs,other factors such as service quality and service reliability are also very concerned by most shippers.The first problem studied in the thesis is the full truckload transportation service procurement with transit time(FTL-TPTT).It is an extension of the classic transportation service procurement problem by minimizing the total transportation cost and the total transit time simultaneously.A solution with a shorter transit time is more attractive for a shipper even if additional transportation cost is required,since agile supply chains can better help the shipper to improve its revenue.Compared to sea shipping,the number of carriers for trucking freight transportation is often dozens,and multiobjective exact methods are not capable of finding all the nondominated solutions.Therefore,we design a two-phase evolutionary algorithm(TPEA)to solve the FTLTPTT.Computational results show that the biobjective heuristic outperforms two multiobjective exact methods and the classic NSGA-Ⅱ,and it is able to solve most of the practical size instances optimally.The second problem is the biobjective transportation service procurement with transit time and total quantity discounts(TPTT-TQD).We develop a biobjective branchand-bound algorithm.Two upper bound sets are designed and two strong fathoming rules are proposed to enhance the efficiency of computing the Pareto front.Two bounding methods with search region reduction are developed to improve the lower bound sets.Moreover,a hybrid branching method is also proposed;it firstly branches in the decision space and then branches in the objective space.A set of test instances that simulates the full container load sea freight service procurement is generated.Computational results on the test instances show that the algorithm is capable of solving the instances of practical sizes.Further analysis shows that decisions can be improved if total quantity discounts are introduced in the procurement of transportation services.Finally,the thesis studies the biobjective O2O food pickup and delivery problem with time window(O2O-PDPTW),which minimizes the total duration time of all the riders and the total waiting time of all the customers simultaneously.For most O2O takeout platforms,in addition to daily logistics operating costs typically considered,the service level needs to be considered as it may be also likely to affect the competitiveness of the platform in the takeout industry.A biobjective memetic algorithm is proposed.It embeds a multi-directional local search in the multiobjective evolutionary framework to search the nondominated solutions.The algorithm also uses a route combination method to further enhance the final Pareto front.Experiments show that the algorithm can help O2O takeout platforms improve decisions.