| As all known, Shandong is one of the largest provinces for poplar plantation. Although the analysis and evaluation of poplar growth with yield has made some development, systematical research related poplar united management has not been completed, especially in poplar operating models and risk management. In this paper, based on the poplar plantation in Shandong province, by statistical analysis of the matlab software for operation platform, discussed the research methods of poplar regional integration management, established a nonlinear bi-level programming model, which comprehensively considerate the targets of upper evel and low level. An effective hierarchic genetic algorithm is adopted to solve the model. Next, the price risk was incorporated into the model, Monte Carlo method was used to simulate price factors. Finally we proved the scientificalness of the model and the availability of the algorithm by two examples of fast-growing plantation.The main research content and summary are as follows:1. According to the present management situation of fast-growing plantation and the modeling principle and technology, the fast-growing plantation optimization model of bi-level programming was built. In the model, the upper level decision maker is forestry administrative unit and the objective is to balance the age structure of fast-growing plantation in the whole area, the lower-level decision makers are direct operators and their objectives are to maximize the economic benefits of harvest. The purpose of this model is to explore how to maximize the economic benefits while balancing the age structure. So this model could be very useful in reality plantation management work for providing scientific theoretical basis for healthy management and reasonable references for efficiency operation of fast-growing plantation.2. Based on the established bi-level programming model and collected timber price data of recent ten years, with price risk incorporated into the model, the stochastic bi-level model is established. And Monte Carlo method was used to simulate price factor. The specific steps of Monte Carlo simulation are as follows: (1) formulate the random mathematic model; (2) generate random numbers; (3) random sample simulation; (4) genetic algorithm is presented to solve the model. 3. A kind of effective hierarchic genetic algorithm for nonlinear bi-level programming model and uncertainty bi-level programming model is presented. First generate a group of upper variables in the upper bounds, then the lower level can optimize its objective on this base. When the upper level get the feedback of lower level, the upper level objective and fitness can be calculated with the optimization result of lower level.After that, the upper level continue to optimize objective with genetic algorithm. Finally the global optimal solution will be outputed.4. At last with examples of fast-growing plantation, the scientificalness of the two models and the availability of the genetic algorithm is proved. With the results of examples, the model was proved that it has a great practical application value for considering the needs of different levels of forestry system and combining with practical condition when optimizing. And the economic benefit would be lose more with the age structure improved much. Besides, different data ranges of felling limitation value were tested in this article which showing us that the forest age structure could be optimized better if the upper level loosen control on felling quantity.On the other side, the availability of Monte Carlo simulation was expressed in the example. A hundred price scenarios were used as realized prices in the model. Under price risk, the age structure was well improved and economic benefit is more than the fixed price condition. Through simulation of uncertainty in timber price, which could more close to reality situation,the harvast risk could be minimum and the price risk can be known more comprehensively, so the decision making would be more accurately.
|
PDF Full Text Request