Research On Risk Optimal Decision Models And Application For Environmental System Based On Uncertainty

Uncertainty optimization model has become a significant and effective tool to deal with the uncertainty decision making in environmental system analysis. The building and algorithm of optimization model system have important academic and applicable values. This thesis aims to build the model system of risk explicit optimization to directly reflect the relationship between risk and benefit as well as the designing of related effective heuristic method. The building process of uncertainty optimization model under unified measure, algorithm design on the basis of uncertainty simulation technique, the test of searching quality and efficiency when algorithm solving the uncertainty optimization problems were made systematically in this thesis. Meanwhile, water resources system and watershed environmental system were taken as the cases to verify the science of the model and the effectiveness of the algorithm in the application of the environmental system decision making. The following contributions have been made:The influence of parameters in the objective function and constrained condition on risk was considered on the basis of existing uncertainty theoretical method. The uncertainty optimization models which reflected the corresponding relationship between risk and aspiration level directly were established, including: risk explicit stochastic programming model, risk explicit fuzzy programming model, risk explicit stochastic-fuzzy programming model, modified risk explicit interval programming model as well as expended model under different measure. If stochastic, fuzzy, stochastic-fuzzy parameter or interval parameters are included simultaneously, various forms of coupling optimization model could be established. The definition of uncertainty measure and risk index were proposed, the unification of optimization model under different measure was obtained by the transformation of the maximum uncertainty measure model to the minimum risk index model, and then a system of theory and method on uncertainty risk explicit optimization model was formed.To combine the chaotic monkey algorithm (CMA) with the technique of stochastic simulation and fuzzy simulation, stochastic simulation-based and fuzzy simulation-based CMAs were designed respectively to solve stochastic and fuzzy risk explicit uncertainty optimization problems. CMA was used to achieve the solving of risk explicit interval linear programming (REILP) model. Taking credibility fuzzy Thus, it is verified that the universal CMA parameters were also applied to the solving of uncertainty optimization model. A numerical experiment on land-use decision making under total maximum daily load (TMDL), three kinds of uncertainty optimization models were sovled to verify the validity of the algorithms. The sensitivity of parameter distribution was also tested, and then the characteristics of the three uncertainty optimization models were discussed.The risk explicit fuzzy nonlinear programming (REFNLP) model for the optimal allocation of water resources under the fuzzy uncertainty environment, and the risk explicit fuzzy-interval linear programming of water resources (WR-REFILP) model under fuzzy-interval uncertainty environment were built respectively to fully consider the uncertainty and complexity in the urban water resources and water environmental system. In addition, fuzzy simulation-based CMA and the improved REILP model algorithm were designed respectively, obtaining the risk and system return tradeoff schemes of water allocation of different sectors of all regions at a certain confidence and aspiration level. Hence, decision schemes under acceptable risk level were provided for the decision makers. It is proved that the new intelligent algorithm: CMA has the characteristics of simpleness, high efficiency and robustness when solving multi-dimension problems by comparing with other algorithm results. Moreover, the differences of the two models were discussed, and a solving method of multi-objective REILP (MOREILP) model was provided.Take watershed ecosystem as another research case to further verify the validity of the proposed method system as well as the advantage of the new algorithm when solving large scale uncertainty optimization problems. Credibility fuzzy chance constrained programming (CFCCP) and REFILP model were built respectively, optimal temporal and spatial distribution was obtained at a certain confidence and aspiration level. The decision proposal in the ideal area was chosen by exploring the change rule of risk level and cost according to the risk cost and benefit trade-off curve. The model comparison research shows that REFILP model overcomes the limitation of the previous ILP model, which can provide specific and feasible scheme to directly reflect the trade off of risk cost and benefit with the advantages of short calculation time and high efficiency; the comparison with CFCCP model shows that the two models are consistent in solving extreme optimal value, and the respective applicable fields were also provided. The above research results enrich the uncertainty environmental system optimization method theory to provide support for environmental system optimization and scientific decisions, which helps to achieve the coordination of society, economy and ecology environment as well as sustainable development.