Study On Optimization Model And Algorithm Of Irrigation District Under Uncertainty

With the rapid development of economy in China, agricultural development is lagging behind facing multiple problems, such as the reduction of agricultural water consumption because of water shortages year by year, the excessive ineffective supply because of unreasonable supply-side structure, and the huge waste of agricultural water resources due to low utilization. As the major base of grain production, irrigation districts in China are facing the severe problem of ever-increasing water resources shortages, leading to the urgency to optimize irrigation management schemes and agricultural planting structure. In addition, there are multiple uncertain factors in optimization models for irrigation management, such as precipitation, water availability and many economic parameters. These uncertainties can cause great losses or even catastrophic risks for an agricultural irrigation system. However, these risks cannot be reflected in the previous deterministic models. Therefore, it is important and meaningful to research and apply uncertainty optimization theory on agricultural irrigation efficiency of three main units including agricultural water resources, irrigation system and agricultural planting structure.In this paper, the study area is Shijin Irrigation District in Hebei province. Starting from the respects of water resources, canal system and crops and considering the uncertain factors in the irrigation area, three optimal irrigation models were established, and their corresponding algorithms or programs were proposed. The achievements are listed as follows:(1) Taking into account uncertainties of water quantity, a fuzzy interval programming model for irrigation water was established. The model has the characteristics of reflecting uncertainties expressed as interval and fuzzy, and can deal with the agricultural water optimization problems which include multiple water supplies, multiple sub-districts, multiple crops and multiple crop growing periods. The model is to achieve maximum incomes of local farmers and to maintain their equitable incomes. So a fairness constraint is considered to ensure the water quantity are distributed relatively fair rather than absolute mean to farmers in the irrigation district. The model can avoid the phenomenon that the farmers who live closer to water source are allocated larger scale water due to the higher water use efficiency than the farmers who live far away from water source. Meanwhile, the model takes into account a constraint of priority to apply surface water, which can give priority to use surface water in irrigation periods and make groundwater as a supplementary water source. Therefore, the model can also reduce the risk of losses caused by the exploitation of groundwater. Moreover, a single step method to solve the above model was developed on the basis of two step method and goal programming. The method can handle the models with high uncertainty on right hand side and avoid the situation of the non-complete solution or meaningless solution obtained by two step method.(2) Taking into account uncertainties of groundwater in sub-districts, an irrigation management model based on queuing theory was established. The model takes the minimum irrigation duration as the target, and gives a function relation between irrigation time and area irrigated by groundwater based on queuing theory. Then an adaptive particle swarm optimization algorithm was developed based on particle swarm optimization algorithm. The algorithm adopts a linear decline function to determine the inertia weight, and adds artificial random disturbance when it is possible to fall into a local solution.(3) Taking into account uncertainties of economic parameters, a fuzzy multi-objective fractional programming model was established. The model achieves multiple goals by adjustment the planting land of winter wheat, maize and cotton, which can protect and utilize limited agricultural water resources effectively, as well as ensure food supply security. Based on superiority and inferiority measures and combined with goal programming method, an algorithm to solve fuzzy multi-objective fractional programming was developed. The algorithm can transform a complex fuzzy multi-objective fractional programming model into a single object model with almost linear structure.(4) Taking into account multiple uncertainties in agricultural water resources management systems, an agricultural water optimal allocation model under uncertainty was established. Agricultural water allocation schemes were obtained based on the Monte Carlo simulation of the distribution characteristics of the uncertain parameters. Further, the linear relation between the difference in purchase price and the difference in water allocation of winter wheat and maize were obtained.(5) A general decision support system for water resources optimal allocation was developed. The system involves both fuzzy and interval uncertainties. The decision support system is capable of flexibly designing data structure and model structure, providing multiple solution methods, and displaying various results having no limit on crop species and irrigation districts. It is such a general tool that can be easily transplanted to other irrigation districts platform.