The Research Of Longtan Reservoir's Optimal Scheduling Based On Runoff Stochastic Simulation

Hydropower,being a renewable clean energy source,plays an extremely important role and underpins the sustainable development of the economy and society in the history of energy development in China.Longtan Hydropower Station with annual regulation capability plays an important part in the hydropower compensation regulation for the cascade Xijiang River Power Station.Working out a scientific and reasonable optimal scheduling model of po wer generation and formulating optimal scheduling chart accordingly from the model can bring better social and economic benefits to Longtan Hydropower Station and shows quite strong theoretical significance and practical application value.This paper takes the Longtan Hydropower Station Reservoir in Xijiang River Basin as the research object to study the establishment and solution of stochastic optimal scheduling model of reservoir power generation under the two scenarios of historical runoff sequence and stochastically-simulated runoff sequence.Based on this reservoir scheduling strategy,the reservoir operation chart is drawn.The main research results obtained in the paper are as follows:(1)The statistical characteristics and the trend,variability and periodic nonlinear property of the inbound runoff are revealed by using hydrological statistics as well as methods of Mann-Kendall,R/S,ordered clustering and Morlet wavelet analysis.(2)The maximum entropy principle and Copula function are in,troduced into Gibbs sampling algorithm to simulate the monthly runoff of 1000a.This method fully reflects the correlation structure of the runoff between adjacent months and effectively avoids the information distortion caused by fitting joint distribution.The fact that the simulated runoff sequence be similar to the measured runoff sequence in statistical characteristics proves that the Gibbs sampling algorithm derived from the maximum entropy principle and the Copula function could be used for the stochastic simulation of monthly runoff:(3)Using simple Markov chain to describe the inbound runoff process,the runoff transition probability matrix is deduced with statistical method.The transition probability matrices based on historical runoff sequence and simulated runoff sequence are compared and analyzed.The statistical law of transition probability matrix based on simulated runoff sequence is more obvious and consistent with the actual circumstances.(4)A stochastic optimal scheduling model for power generation in Longtan Reservoir is established and a scheduling strategy based on historical runoff sequence and simulated runoff sequence is obtained by using stochastic dynamic programming algorithm.Based on this reservoir scheduling strategy,the reservoir operation chart containing the information of the upcoming water in the future period is drawn which could solve the reservoir stochastic optimal scheduling problems.So,the obtained reservoir scheduling chart could provide references for the practical scheduling of Longtan Reservoir.