Research On Distributed Simulation-based Optimization Based On HLA And Its Application

Nowadays, efficient and cost-effective logistics system is becoming a decisive factor of the operating performance in modern logistics. It is difficult to analyse logistics system with the traditional analytic method due to its characteristics of discrete, dynamic, random and complexity. The simulation-based optimization has become an important method for complex logistics systems.However, it holds some drawbacks in the practical application. The evaluation of optimization objective depends on simulation runs and the simulation model usually complex, huge constructed from the real complex logistics system, so it costs quite a lot of time in the simulation course. Furthermore, each simulation needs many replications to reduce the random of the system, which undoubtedly exacerbate the computational effort problem. Time and efficiency becomes a bottleneck in the development and application of the simulation-based optimization. Therefore, this paper focuses on shorten the time significantly.Key works of this paper can be summarized in the following five parts:(1) The theoretical framework of distributed simulation-based method (DSO) is anylysed. The fundamental of DSO is to distribute the simulation process to several computers to run the simulation model at the same time. Based on this analysis, the execution flow, modes and characteristic of DSO method are presented.(2) Based on the anylysis applicability of the high level architecture (HLA), the HLA-based DSO structure is proposed. And the function of each component is described.(3) Several issues must be done under the HLA framework are extracted. And then are solved one by one.(4) The key issues in the process of distributed parallel implementation are focused. The optimization mathematical model is reconstructed into a distributed simulation-based optimization mathematical model, the parallelization of algorithm is designed and the key technique to solve data distribution is discussed.(5) The time composition and node assignment are discussed. Based on the execution flow of DSO, each time component is analysed and final expression of the total execute time is gained. Furthermore, we analysised the main factors which affect node assignment. In the end, an experiment is implemented to validate the correctness of the proposed approach. And we develop a HLA-based container terminal logistic system federation using the approach this paper presented. The results show the significant efficiency improvement of proposed method on optimal resource allocation on container terminals. This approach proposes an effective solution to solve the time efficiency problem in simulation-based optimization on complex logistics system.