Research On Routing Optimization Problem Of4PL With Completion Risk

In the21st century, the competitive approach has changed dramatically. With the development of supply chain management, a new supply chain management pattern-the4th party logistics (4PL) has appeared which has caused widespread concern in academia and the business community. As a client manager, based on third-party logistics,4PL has used powerful ability to integrate social resources to provide customers integrated supply chain solutions. But now theoretical discussion of the logistics system is very broad for4PL, so the relevant research has significant meaning.When4PL integrates the external resources to optimize decision-making, the crucial two issues:the logistics routing optimization and the evaluation and selection of the Third Party Logistics (3PL) providers. And various uncertain factors make the logistics operation exist time delay. If the time delay leads to due time over a given period,4PL will bear the corresponding losses. Therefore, based on the4PL routing optimization, this paper further investigates the4PL routing optimization problem with completion risk.After reviewing4PL routing optimization, completion risk management, stochastic theory and genetic algorithm, immune algorithm, BP neural network algorithm, our research achievement as follows:Firstly, for the characteristics of randomness, we start with model of dependent-chance programming based on multigraph. Under various constraints, we solve the best path of the maximum completion probability.Moreover, because routing optimization problem is a NP-hard problem, the genetic algorithm (GA) based on stochastic simulation and the immune algorithm (IA) based on stochastic simulation were designed respectly. GA based on stochastic simulation, in addition to encoding, crossover, mutation, selection of the characteristics of GA, has used the stochastic simulation method to determine the fitness function; IA based on stochastic simulation which has introduced the concept of affinity, concentration, expected reproductive rate, and memory mechanisms, can better maintain the diversity of groups. Through three different instances of scale simulation, GA based on stochastic simulation can get better solutions on a small scale, but with the expansion of the scale it falls into a local optimum; and GA based stochastic simulation can get better solution at a faster convergence speed to overcome shortcomings.Finally, note that the stochastic simulation process is time-consuming, we then transform the dependent-chance programming model to the equivalent deterministic model. The results of the algorithm are better than the algorithm results based on stochastic simulation, especially the results mean, computing time and stability.The research of this paper provided the effective method for routing problem and its optimization in4PL in real life.