Author(s): Panuwat Pinthong; Ashim Das Gupta; Mukand Singh Babel
Linked Author(s): Ashim Das Gupta
Keywords: Neurofuzzy; Fuzzy logic; Decision-making modeling; Reservoir operation and management; Pasak River Basin
Abstract: Reservoir is used to regulate the flow for water resources management to meet the spatial and temporal demand in terms of quantity and quality. A decision-making procedure is needed for reservoir operation and management as it plays an important role in balancing demand and supply for optimal social, economic and environmental benefits. A decision-making model for reservoir releases is developed in this study. The model is composed of three modules: the first module involves a simulation model of physical system based on water balance concept. The second module deals with the decision-making process using neurofuzzy technique for switch control, water supply, and flood control. The third module is for the performance evaluation of water supply and flood control using three indicators such as reliability, vulnerability and resiliency. The hybrid learning algorithms combining the gradient descent and the least-square method is used for training the neural network to generate the fuzzy rule-based and membership functions. The inflows to and storage state of the reservoir are the input for the model and the water releases form the reservoir are the corresponding output. The decision-making model is applied for reservoir operation study of the Pasak Jolasid Dam, Thailand. From the comparison of the actual operation with the results of the neurofuzzy and simulation using rule curve, the reservoir releases derived based on neurofuzzy model indicate higher reliability for water supply and flood protection. With the simulation using the rule curve, the derived water releases do not exceed the downstream flow capacity of the Pasak River, but releases do not meet the water demand with the level of reliability as can be achieved with neurofuzzy rule based. The appropriate water releases are decided based on neurofuzzy logic developed for reservoir operation and management. All membership 281 functions in neurofuzzy model are expressed in nondimensional form, which provides a generalized model to be used for reservoir operational study.