Improved Harmony Search Algorithms And Their Application To Location-Inventory-Routing Problem

With the development of information technology and transportation,nationwide procurement and distribution has become a trend.Making determinations considering the whole supply chain,including procurement,location,distribution,not only can help large-scale retail enterprises to save operating costs,but also has gradually become the core of the competitiveness of enterprises.Thus,this thesis focuses on Location-Inventory-Routing problems based on joint replenishment strategy,designs two improved harmony search algorithms for the proposed model,and utilizes experiments to test the validity of the algorithms.Firstly,the advantages and disadvantages of differential evolution and harmony search are analyzed to provide the theoretical basis of hybridizing them,and an improved differential harmony search is proposed.Two experiments are taken to verify the performance of the improved differential harmony search.The first one consists of 22 benchmark functions and the second one consists of 15 benchmark functions.The experiments show that improved differential harmony search has better convergence accuracy and stability than other tested algorithms.Secondly,to overcome the difficulty of presetting the scale factor,a learning mechanism based on past experience is proposed and an improved self-adaptive harmony search is designed.Improved self-adaptive harmony search can dynamically adjust the scale factor based on the searching history.An experiment shows that improved self-adaptive harmony search has faster convergence and better accuracy than other tested algorithms.Lastly,a practical joint replenishment-location-inventory-routing model is proposed considering a three-layer supply chain consists of suppliers,distribution centers,and customers.The experiment shows that improved self-adaptive harmony search is better than other harmony search algorithms in solving joint replenishment-location-inventory-routing problems.