| Because of the low cost and high efficiency, the container transportation has been taken extensive attention by various circles at present. As a key part of container transportation, the railway container modal plays an important role in the whole system. Railway container terminal is not only the main place to conduct the arrival-departure, storage, and process of goods, but also the important components of the modern integrated logistics system. The only approach of survival in the intense market competition is to improve the production efficiency and decrease the operating costs. However, the railway container terminal logistics system is relatively large, and has the characteristics of highly non-linear and dynamic. Meanwhile, the affect factors of the processes are numerous, and the logic between each segment is complex. Therefore, the research of this system optimization is very difficult. In this dissertation, the relative optimization problems of railway container terminal logistics system are researched, and different methods are adopted to analyze and optimize the segments and resources of the system, so as to attain the goals of scientific management and control.The state space model is a kind of dynamic time domain model which is come from the stationary time series analysis, and the fundamental assumption of it is the dynamic system conforms to the markov property. This model can describe the internal state of the system accurately and explain the relations between internal state, external input and output variables very well. Mutual information technology is a typical feature selection high-dimension data separation measurement approach, the internal relations are established between the high-dimension feature extraction vectors and the output classification information by which, in order to achieve the aim of reducing the high-dimension of the original feature space. Because the state space time series model is suitable for the problems with multiple input and output variables, and makes no require of massive historical data to describe the system states, so this model is used to establish the railway container terminal location’s regional freight demand prediction. Furthermore, the mutual information technology is introduced to reduce the high dimension of the original input data. Besides, a weighted mutual information calculation method is proposed to deal with the exceptional affect factors as natural disaster and policy changes, so that to improve the abilities of comprehensive information extraction and data dimension reduction. Through the comparison experiments with the LIBSVM support vector regression model and the local linear wavelet neural network model, the effectiveness of this model to solve the regional freight demand prediction problem with small sample and high-dimension is proved.Queuing theory is an effective method to research the system operation strategy related problems, the models established from this theory can be further divided into steady queuing model and the transient queuing model. Because the description of the system state is difficult and the calculation is relatively complicated, so the applications of the transient queuing model are limited. Railway container terminal gate system has a strong dynamic nature, so it is hard to be described accurately by the tranditional steady queuing model. For the congestion problem of railway container terminal gate system, the relevant data are collected and the distributions for inter arrival time of outer trucks together with the service time of gate system are estimated. Then the transient queuing model and system optimization model are established accordingly. The equally likely combinations optimization solution method is brought to solve the queuing model. From the system simulation comparison experiments and three sensitivity analysises, the model and method are proved to be reasonable and effective.For the railway container terminal crane scheduling and storage allocation problem, the factors such as storage direction and the safity distance between the cranes are considered comprehensively, and the mathematical model is carried out on the objective of minimizing the maximum process completion time in this dissertation. As a significant approximate solution method to the NP-Complete problem, the heuristic algorithm is act an important role in solving the railway container terminal related optimization problem. Backtracking search optimization algorithm is a comparatively new evolutionary algorithm. Due to the simple overall structure, it can be used to solve the high-dimension multimodal optimization problem rapidly and effectively. On the purpose of enhancing the optimal solution searching ability, and overcoming the defect of trapping into the local optimal solution, the backtracking search optimization algorithm is improved to get a better performance and fitness to the railway container terminal crane scheduling and storage allocation problem. The numerical example analysis shows the railway container terminal crane scheduling and storage allocation model as well as the improved backtracking search optimization algorithm is feasible and valid.It is difficult to describe the railway container terminal logistics system by using the tranditional mathematical model. Therefore, system simulation method is an efficient way to solve this problem. Simio is new generation intelligent object oriented 3D system simulation software. It has a distinct three-layer structure, and can provide the definition to the behavior, property, process of the system object. Consequently, it has favorable simulation ability to the discrete system. The Simio system simulation model is established in accordance with the entire processes of railway container terminal logistics system in this dissertation. Through the actual input parameters, the system operation statuses are simulated under some specific conditions. Meanwhile, for the sake of observing the results changes and drawing some conclusions, the system related affect factors and the numbers of resources are altered to conduct repeatedly experiment respectively. The railway container terminal logistics system resource allocation optimization calculation method is proposed and integrated into the system simulation logic. The optimal numbers of resources are calculated with the running of the simulation model. It can be concluded from the experiment results, system simulation method is able to give an effective imitate and supervisory control to the railway container terminal logistics system, and the optimization results are contributed to the managers to implement the relevant decisions.In this dissertation, the mutual intelligent technology, state space time series, equally likely combinations algorithm, improved backtracking search algorithm and Simio system simulation technology are adopted to research the railway container terminal logistics system. The system intelligent control is implemented from the aspects of regional freight demand prediction, gate system congestion optimization, crane scheduling and storage allocation, system resource allocation optimization. The researches of this dissertation have vital theoretical significance and practical application value, and lay some foundations for the future studies. |
PDF Full Text Request