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

With the rapid development of technology and the acceleration of urbanization, energyshortage and environmental pollution are increasingly serious. Due to its advantages ofnon-polluting, renewable, abundant reserves, hydropower has become the most importantform of renewable energy. But in recent years, the floods trends to be frequency increasing,strength increases, loss worsening. The inflow change of the cascade reservoirs because of awide range of heavy precipitation bring enormous challenge to the optimal scheduling of ahydrothermal power system, seriously affecting the safe and stable operation of the powersystem. So research on stochastic optimal dispatch of hydrothermal power system consideringthe risk of flood losses is of great important significance theoretically and practically.The improved genetic algorithm based on the ELM is proposed in this paper to improve theefficiency of the optimization problem solving. The algorithm utilizes the extreme learningmachine (Extreme Learning Machine, ELM) to depict the fine features of the complexnonlinear mapping, quantitative simulation the process of the nonlinear evolution betweenfather generation and progeny population, and applies it to in the evolution process of thegenetic algorithm population. The simulation shows that, improved genetic algorithm basedon ELM can choose the inter-generational population numbers and population size andimprove accuracy and convergence speed of operation for solving stochastic optimizationproblems of the hydrothermal power system.In this paper, the stochastic optimization model of a hydrothermal power systemconsidering the risk of flood losses is proposed based on using conditional value at risk(CVaR) to depict the risk of flood losses through a systematic analysis to the characteristics ofthe water to depict its uncertainty. The risk of flood losses in the optimal scheduling of ahydrothermal power system is depicted based on good control of CVaR on the distribution oftail risk by using over storage losses of the reservoir and the flooding losses of thedownstream as an index to the risk measurement indicators. The model is proposed by usingconditional risk constraints in the hydropower station reservoir instead of the securityconstraints of the traditional hydrothermal scheduling model. It makes the theory and methodsand expands the hydrothermal optimal scheduling. The stochastic optimization problem istransformed into a deterministic optimization problem by combining the Monte Carlosimulation and analytical conditional hazard function and equivalent transformation anddiscrete through the conditions risk function due to the difficulty in expressing the probability density function parses of the conditional hazard function. Finally;through the example ofIEEE -9 bus system analysis to verify the feasibility and effectiveness of the stochasticoptimization model of a hydrothermal power system considering the risk of flood losses.There is a certain correlation between the inflows of the cascade reservoirs by furtherconsideration to the cascade reservoirs controlled by the same or similar weather. So the jointprobability distribution of the inflows of the cascade reservoirs is proposed by using Copulaconnection function;and then the risk of flood losses considering the probabilisticdependency of the inflow of the cascade reservoirs. On this basis;the stochastic optimizationmodel of a hydrothermal power system considering the risk of cascade reservoirs’flood lossesis proposed that using conditional risk constraints in the cascade reservoirs instead of thesecurity constraints of the traditional hydrothermal scheduling model. It improves thehydrothermal optimal scheduling theories. The simulation from IEEE-14 nodes system showsthat the optimal hydrothermal scheduling program considering the risk of flood lossesconstraint for a hydrothermal optimal scheduling method under different conditions;riskvalues and confidence level can reduce floods economic losses and improve the utilization offlood.