Study Of Inventory Management Problem In Close-Loop Supply Chain Based On The Internet Of Things

A Closed-Loop Supply Chain is a whole material circle from production to recycling material,which aims to reduce environmental pollution and resource waste.Closed-Loop Supply Chains can supply products and services to customers in an economic and scientific way.The advance in model information technology,especially the Internet of Things,brings a great revolution in methods of collecting,handling,utilizing information.In the past some key information can only be predicted by fuzzy or probability methods,while the Internet of Things can provide real-time and accurate information of tracking and monitoring which covers every phase of product whole life cycle.The Internet of Things makes managers of supply chains grasp current states of systems,and has caused a big change of operation modes in Closed-Loop Supply Chains.For the problems of inventory management in Closed-Loop Supply Chains,the Internet of Things visualizes some key information,such as inventory level,use information,and recovery information,where the tradition inventory management tool lacks corresponding handling methods and application modes.The works of designing information handing methods and novel inventory models can guide enterprises to build advanced inventory management systems,enhance the competitiveness,and bring about the power of sustainability for society.Thus,this thesis studies the problems of inventory management in Closed-Loop Supply Chains based on the Internet of Things.Focus on the inventory system with embedded sensors in Closed-Loop Supply Chains,this thesis studies the information handling method,and proposes a complex event detection method under the condition of harsh working.Then,this thesis systemically analyzes inventory problems under serval circumstances,which includes the consideration of containers' recovery information,the spare inventory management with the use information of Product Service Systems,and the coordinate optimization of inventory and logistics.The first two problems are in the area of stochastic dynamic programming problems,which encounters the trouble of enormous calculation requirements with increasing scales of problem,namely "the curse of dimensionalities."Thus,we propose high-performance heuristic strategies for reducing calculating requirements.For the problem of coordinate optimization of inventory and logistics,this thesis designs an intellectual algorithm based on adaptive large neighborhood searching to solve this problem due to its NP-hard feature.The research works of inventory management of Closed-Loop Supply Chains in the environment of Internet of Things are concluded as follows.(1)The information handling method under the condition of hard working is studied.Complex events are composed by a group of senor data in some logic structures.We analyze the noise of data from the Internet of Things during the collection and transmission processes.A complex event detection system is proposed based on supervised learning model.In this system,we design a detection algorithm for handling the noise and disordered errors in data streams for the requirement of real-time information handling.Meanwhile,the algorithm of calculating reliable values are designed based on Taboo searching,which enhances the accurate level of complex event detection system by minimizing the bias of training data.Through a simulation study based on a real manufacturing case,the calculation efficiency of the detection system is shown.The comparison between the proposed system and the Viterbi algorithm of Hidden Markov shows the accuracy level and the robustness of the detection algorithm under different circumstances of sensor noise.(2)The inventory problem with the recovery information of reusable container is studied based on the Internet of Things.This thesis considers the condition where the manufacturer of the Closed-Loop Supply Chain supplies a product by reusable containers with the information of when the container is ready to recover.Based on the theory of Markov Decision Process,the inventory model of single kind product and single retailer is built.Under different levels of information provided by the Internet of Things,myopic and farsighted heuristic strategies are proposed through the consideration of minimizing expected operation costs.In the numerical studies,the performances and features of these two strategies are analyzed in different utilization rates.The model is extended to the conditions of selective disposal behavior and uniform distributed in-use time in the simulation study.The performances and robustness of the farsighted strategy and the corresponding values of information are also evaluated.(3)The problem of spare inventory of Product Service System based on the Internet of Things.The product replacement and spare inventory model is built for maximizing the net service revenue based on the Markov Decision Process which views the service process as a Markovian deterioration.This thesis utilizes the technology of Internet of Things to monitor working status of in-use product,and considers the condition that the service supplier of Product Service System receive the revenue based on the quantity of providing service.Based on the characters of optimal strategy of multi-customer system,this thesis proposes a heuristic replacement policy and a heuristic inventory control policy.In 20 numerical studies with random parameters,the average gaps between the heuristic strategy and the optimal strategy are 3.27%and 5.66%under the conditions of identical and unidentical customers.In the end,the sensitive analysis evaluates the impacts of varying parameters on the performance of the heuristic strategy.(4)The coordinate optimization problem of inventory and logistics is studied based on the Internet of Things.This thesis utilizes the technology of Internet of Things to monitor the status of distributed inventories and transportation vehicles.The mathematic model of coordinate optimization is built for minimizing the sum cost of inventory and transportation.This model generalizes the Inventory Routing Problem with the consideration of material balance at the manufacturer and simultaneously delivery and pickup in same routes,which is definitely a NP-hard problem.For solving this problem,this thesis proposes a modified mata-heuristic algorithm based on the adaptive large neighborhood searching algorithm.Meanwhile,a dynamics algorithm is proposed for the sub-problem of deliver and pickup planning.Utilized into a standard testing example,the performance of the proposed algorithm is evaluated.In the end,the robustness of algorithm is evaluated by the sensitivity analysis.