Research On Distributionally Robust Optimal Dispatch Of Hydrothermal Power System Considering Flood Losses Risk

With increasing energy shortages and deepening of the power system, priority and vigorously develop hydropower is the inevitable choice of sustainable development strategy implementation and energy structural adjustment. Cascade reservoirs hydropower optimal scheduling significant economic benefits, which has been widespread concern. But in recent years, floods render intensity increases, increased frequency, loss worsening trend. A wide range of heavy rainfall so extreme cascade reservoir inflow mutations to hydro-thermal system optimization scheduling an enormous challenge, seriously interfere with the safe and stable operation of the power system. Thus, for the consideration of the cascade reservoirs flood loss risk of hydro-thermal scheduling problem, both in theory and reality has important research significance and value.Firstly, the paper analysis cascade reservoir spillway safety margin, building hydro-thermal scheduling model based on security constraints of cascade reservoir flood prevention. However, the model does not reflect the impact of cascade reservoir extreme flood inflow on scheduling, for which the extreme value theory based peak over threshold(POT) model is adopted to build Generalized Pareto Distribution(GPD) of single reservoir extreme flood inflow. And then, Copula function is used to build a joint probability distribution of cascade reservoirs extreme inflow during flood period. Based on excellent performance of conditional value-at-risk(CVaR) in characterizing the tail risk of cascade reservoir spillway safety margin, a CVaR constraint for cascaded reservoir flood prevention is proposed, and a scheduling model for hydro-thermal power system considering cascade reservoir flood prevention risk is further proposed. Secondly, With Rockefeller and Uryasey theory to solve the CVaR constraints, which arising multidimensional integration problem after treatment, Sample average approximation is employed to transform the non-smooth CVaR constraint into smoothed inequality constraint, which can be then introduced into objective function as a penalty term. The smooth hydro-thermal system optimization scheduling problem is effectively solved by nonlinear primal-dual interior point algorithm.Finally, take a modified IEEE-14 bus system for example, the impacts of flood prevention confidence level and Monte Carlo sample number on the optimal scheduling results are analyzed quantitatively. The results show that Copula function better portray joint probability distribution of cascade reservoir extreme flood inflow, improve the security of cascade reservoir for flood prevention and the power generation efficiency of hydro-thermal. Based onthe excellent features of CVaR characterize the probability distribution of tail risk,Conditional Value at Risk theory can well describe the cascade reservoir flood risk.As the distribution function of cascade reservoir extreme flood inflow can not accurately estimate is considered, also based on the moment(mean and covariance matrix) uncertainty of cascade reservoir extreme flood inflow, a Worst Case Conditional Value at Risk constraint of cascade reservoir flood safety is proposed, and then a distributionally robust optimization scheduling model considering flood damage risk of hydro-thermal system is established. This model overcome the problem that the random variation in the interval of robust linear optimization was too much conservative, by using the mean and covariance matrix of historical date of cascade reservoir extreme flood inflow, and constructing the set of the uncertainty moment of cascade reservoir extreme flood inflow. Then, the distributionally robust optimization problem is transformed into semi-definite programming(SDP) problem by Lagrange duality theory, meanwhile, solve SDP problem by using interior point method.Finally, take a modified IEEE-14 bus system for example, Quantitative analysis the impact of moment uncertainty parameters of extreme flood inflow and confidence level of flood prevention on the total system cost, the generation flow and spillway flow of upstream and downstream. The results show that change of moment uncertainty set parameters characterize the extreme flood inflow volatility information of probability distribution, which closer to the actual situation. The Worst Case Conditional Value at Risk of cascade reservoirs is considered to ensuring the safety of cascade reservoirs flood prevention, while improving the efficiency of hydro-thermal power generation systems.