A simulation-optimization approach for supporting conservative water allocation under uncertainties

In this paper, a hybrid method integrated unbiased grey model (UGM) and artificial neural network (ANN) into an interval two-stage fuzzy credibility-constrained programming (ITFCP) framework is proposed for water resources allocation of the Yalong River research area. Through the grey correlation analysis and the eXtreme Gradient Boosting (XGboost) algorithm, the economic and social indicators are related to the water demands of different water sectors in different regions can be obtained for building water demand prediction model. According to the unbiased grey prediction of the socio-economic development data of each region in the Yalong River Basin (YRB), water demand prediction models are constructed by using neural network. The establishment of a hybrid two-stage interval fuzzy credibility-constrained programming model can analyze the uncertainties existing in the process of water resources allocation. Taking 2020, 2025, and 2030 as the planning years, the developed model studies and reveals the system benefits at different credibility levels, the water shortage of each user in sub-regions and the water resources allocation situation to provide suggestion for managers to optimize the allocation of water resources. Compared to the previous methods, this integrated model can help decision-makers set management policies more sustainably and profitably.