The Research Of Scheduling Rules And Benefit Evaluation For Lishui River Basin Cascade Reservoirs

The research of reservoir operation optimization has a long history at home and abroad.China is a country with huge water resources and large scale of hydropower construction,which provides the conditions and samples for reservoir optimization.Implicit Stochastic Optimal Operation method is one of the most widely used methods for reservoir optimal operation.It has plenty theoretical methods and good generation benefit.For so many theories,methods,models to choose,,the systematic research is not that much for which method has a better accuracy for the reservoir optimal operation.Based on this,this paper takes the cascade reservoirs of the Lishui River Basin as the research object,selects the linear stepwise regression,BP neural network and cloud model to extract the scheduling rules,and studies the feasibility and adaptability of the three methods to the simulation operation of the reservoir group Analysis,and compare which method has better performance.In addition to the simulation results will be affected by the method of acquiring scheduling rules,and influenced by the accuracy of runoff forecast.Therefore,the text used the single reservoir and cascade reservoirs of the Lishui River Basin as examples,The influence of the accuracy of runoff prediction on the simulation results of dispatching rules is further studied.It provides quantitative and reliable basis for reservoir power generation dispatching plan.The main contents include:For the cascade reservoirs,a deterministic optimization model is established,and the SA-POA method is used to obtain the optimal solution of the reservoir,which is the optimal sample for the implicit stochastic optimization of cascaded reservoirs.The simulation results of the scheduling rules are obtained by linear stepwise regression,BP neural network and cloud model.The comparison between the total power generation and the error,the scheduling process and the system average power generation and system guarantee rate are compared with the deterministic optimal scheduling.Then this paper analyzes the scheduling rules to fit the effect and evaluates the performance of the scheduling rules.Based on the results analysis,the results of the three scheduling rules are less than and close to the deterministic optimization algorithm,which shows the feasibility and practicability of the three scheduling rules.Compared to the BP neural network and the cloud model method for extracting the scheduling rules to get simulation results,the results of linear stepwise regression are closer to the deterministic optimization results and have better performance.Due to the error of runoff forecasting,this paper has a certain deviation on the basis of the actual runoff,and forms the error runoff series to quantitatively study the effect of runoff error on the simulation results of the scheduling rule.The results show that the power generation capacity of single reservoir and cascade reservoirs is more decreased,as the error rate of runoff forecast is increasing.Compared with the annual average generating capacity of cascade reservoirs,the average annual generating capacity of single reservoirs obviously decreases.It is shown that the joint scheduling of the cascade reservoirs is more advantageous than the single reservoir.