Study On Short-term Operation Scheduling Of “One-reservoir And Two-cascade” Hydropower Stations

With the rapid development of China's hydropower,the hy dropower system has generally formed a " cascaded hydropower stations with one reservoir " situation in the southwest large river basin.The common feature is to adjust the good performance of large reservoirs as a leader,with a lower level or multilevel high head,small hydropower station,adjusting ability of the poor capacity,in order to make full use of the storage function of leading reservoir,improve the overall efficiency of the hydropower cascade.Because of the large difference of storage capacity,sensitive head,complex diversion mode,special demand of power plant and power grid,the short-term operation of such reservoirs are more difficult.In this paper,based on the Tian Shengqiao cascade as an example.We use data mining technology based a large number of historical data and get the method of the matching of cascade electricity,the formulation of power generation plan and the distribution of the network.The main work in this paper is as follows:(1)To calculate the Electricity matching between the upper and lower power plant,the paper presents a new method.That is,through collecting and cleaning up a large number of historical data,selecting the key factors affecting the power generation volume,controlling single variables,using li near regression method to find out the numerical relationship between daily power generation from historical data,and further finding out the matching relationship between cascade daily power consumption.Through example simulation,this method can match upstream and downstream power well,and further reduce the difficulty of generation plan generation,which has certain practical application value.(2)Calculating the hydropower generation plan is very difficult.Because there are some problems in the daily operation of hydropower stations.First,it is hard to match upstream and downstream power plants.Second,The power station units have multiple vibration zones.And the last,daily operation of hydropower stations need to treat many complex needs,such as power output ramping constraint and running interval ideal point.The paper presents a new method to solv e those problems.This method integrates the cluster analyst and decision tree algorithm.First,energy production relationship among cascaded hydropower plants is determined by the linear regression method.Second,typical generation curves of all hydropo wer plants are identified using cluster analysis method based on massive practical data.These curves are classified into several generation decision processes by considering daily energy production demand,forebay water level and grid characteristic.Thus,the decision library of generation operation can be built.This method is verified to determine the day-ahead generation schedule for Tianshengqiao cascaded hydropower plants in Hongshui River.Results show that the method can quickly obtain generation s chedule according to the given conditions and constraint boundaries.(3)To calculate the Power distribution between power grids,this paper presents a method to distribute electricity considering the characteristics of different power grids.The NILP model is to minimize the variance of surplus charge,by introducing the ideal point to a unified power grid scale,with the rapid solution of the lingo solver,finally introduced the theory of marginal analysis in economics,to determine the optimal weight.Through the example analysis,the method proposed in this paper is superior to the practical engineering methods in regulating the peak and valley difference of the power grid and the fluctuation of residual load.Finally,the content of the whole paper is summarized and suggestions of further study are given.