Research On Multi-reservoir Flood Control Optimization Dispatch And Flood Evolution Model

With the successive completion of a large number of flood control reservoirs,cascades,parallel-connection and mixed-connection reservoirs with different scales have formed in major river basins in China.Implementing joint flood control scheduling on reservoirs is an effective way to ensure downstream flood control security.Different reservoir systems have different topological structures,flood storage capacity,operation modes,and flood characteristics.When faced different floods,it's a key technical issues that making the reservoirs operation to achieve the optimal allocation of flood storage capacity between reservoirs and to maximize the effectiveness of flood control.This paper takes the reservoir flood control system consisting of five controlled reservoirs in the upper reaches of the Yangtze River as the research object,and studies the flood control optimization scheduling of reservoirs and the related flood evolution.When the downstream flood control section accepts multiple substreams and intervals in the upper stream,the traditional flood evolution model cannot effectively predict the downstream flow process.To solve this problem,this paper establishes a BP neural network flow calculation model which extracting the flood lag from the upstream to the downstream by mutual information.The results of the case study in the upper reaches of the Yangtze River show that the certainty coefficients of the BP neural network flow calculation model in the training phase and the forecast phase are 0.9870 and 0.9560,respectively.The prediction accuracy proves the applicability and validity of this model using for flow calculation.This paper defines the system safety degree to characterize the safety degree of reservoirs during flood control scheduling,analyzes and compares two kinds of safety characterization forms of a single reservoir.What's more,two calculation methods of system safety degree are proposed,namely static weights and dynamic weights,and four kinds of scheduling strategies are formed.With the goal of minimizing the excess water in the downstream flood control section and maximizing system safety degree,a flood control optimization model of the reservoir group is constructed and solved by the elite variation particle swarm optimization algorithm.The results show that the dynamic weights piecewise-linear safety scheduling strategy can not only effectively reduce the amount of excess water but also achieve optimal allocation of flood storage capacity between reservoirs,and is more conducive to the system's flood control security.The scheduling coefficient and equal proportion impoundment idea are introduced to improve the divergence distribution method of the aggregate decomposition model,solving the problem that the traditional decomposition model violates the constraint conditions and can not allocate discharge of series reservoirs.Meanwhile,this paper proposes a multi-factor scheduling function,and uses the stepwise regression method to obtain the characterization of the scheduling function,and the parameters of the scheduling function are further optimized by using the chicken swarm optimization algorithm.The results of the case study show that the over-limit water reduction rate of multi-factor scheduling function is significantly better than the piecewise linear scheduling function's.