Research On Improved Genetic Algorithm And Optimal Operation Of Cascade Reservoirs

Energy is the material basis for people to survive,and an indispensable production factor for social development."The 13th Five-Year Plan for the Electricity Development" points out that the entire society is expected to use 6.8-7.2 trillion kWh of electricity by 2020,and the installed power generation capacity in the country will reach 2 billion kilowatts.With the continuous increase of people's demand for energy,the limited nature of coal resources and the impact on the environment cannot be ignored,and thus,vigorous development of renewable energy has become the requirements of the present era.According to statistics,as of the end of 2017,the installed capacity of renewable energy power generation in China reached approximately 656 million kilowatts;the installed capacity of hydropower,wind power and solar power ranked first in the world.Due to the abundant hydropower resources in China,the advantages of tens of thousands of rivers and hydropower energy make the hydropower industry have broad prospects for development and occupy an important position in many renewable energy and clean energy sources.With the gradual formation of cascade reservoirs in river basins of China,it is of great theoretical and scientific value to study the theories and methods for the joint optimal operation of cascade reservoirs to improve the utilization of water resources.In addition,due to the various functions of reservoirs such as flood control,power generation and irrigation,a reasonable operation plan of cascade reservoirs can not only effectively ensure the personal and property safety,but also create great wealth for the society.To this end,starting from the theoretical method and practical application,this paper studies the joint optimal operation of cascade reservoirs.In-depth research is conducted on improved intelligent algorithms and applications,optimization of cascade energy storage scheduling graphs,mid-term and long-term power generation scheduling optimization of cascade reservoirs,parallel design and application of improved algorithms,and short-term uncertain optimal operation of cascade reservoirs.The main research contents and corresponding results are as follows:(1)Research on the improvement of Genetic Algorithm(GA).In light of the problems of prematurity and easy convergence in traditional genetic algorithms,the transformation mechanism of high-low dimension space mapping in Self Organizing Maps(SOM)is introduced.Individuals in GA are learned and updated through SOM,so that renewal individuals can inherit the topology preservation and distribution preservation property of parent individuals,the local search ability and computational efficiency of the GA in the later period are further improved,and a Self Organizing Map-Genetic Algorithm(SOM-GA)is proposed.The improved algorithm is tested by several standard test functions such as non-restraint and restraint,and the rationality and feasibility of the algorithm are verified from the solution result,convergence efficiency and stability;(2)Research on optimal operation model of cascade reservoirs based on Self Organizing Map-Genetic Algorithm.Cascade reservoirs operation is a multi-stage nonlinear complex mathematics problem.To verify the practicality of SOM-GA algorithm in solving this problem,firstly,starting from the conventional operation method,based on the discriminant method,the cascade reservoirs energy storage scheduling graph meeting the maximum guaranteed power under the requirements of the power generation guarantee rate is plotted.On the basis of storage energy scheduling graph,each operation line in the graph is optimized using SOM-GA,specific to the characteristics of the operation graph,the optimization step of SOM-GA is proposed based on progressive optimality algorithm(POA),and finally through the multi-year runoff simulation scheduling,the rationality and effectiveness of the optimized energy storage operation graph is verified.Secondly,from the optimization operation approach,with the maximum generating capacity of the cascade reservoirs as the operation target,the mid-term and long-term optimization operation calculation step of cascade reservoirs based on the SOM-GA algorithm is given,and the algorithm initialization strategy under cascade reservoirs scheduling is proposed.The optimizing calculation is performed under various constraint conditions,and comparative analysis of the calculation results are analyzed by comparing with the dynamic programming and traditional genetic algorithm to summarize the application effect and advantage of the SOM-GA in the handling of cascade reservoirs optimal operation problem;(3)Research on optimal operation of cascade reservoirs based on parallel uniform self organizing mapping genetic algorithm.When solving complicated optimization problems such as cascade reservoirs operation,SOM-GA often suffers problems such as slow convergence and slow solution performance due to uneven distribution of the initial solution when the dimension of the solution problem increases to a certain number.To this end,further improvements are made to SOM-GA from two aspects:first,from the point of view of design tests,the concepts such as number of levels and number of factors in uniform design are introduced to characterize the genes structure of individuals,and a Uniform Self Organizing Map-Genetic Algorithm(USOM-GA)is proposed.The initial population individual generation strategy of the SOM-GA algorithm is improved so that it could purposefully generate individual genes in the feasible region,thereby improving the population diversity and the convergence rate of the algorithm;Secondly,based on the good parallel performance of the USOM-GA algorithm,a parallel multi-core algorithm is designed for USOM-GA,a multi-core parallel USOM-GA algorithm is constructed,in combination with the case applicaiton of optimal scheduling of cascade reservoirs,the corresponding solving calculation steps are given,and finally,the rationality and effectiveness of the algorithm is verified through result analysis and evaluation index;(4)Research on short-term optimal operation of cascade reservoirs considering the output error of hydropower stations.In order to effectively reduce the operation risk with formulation of power generation plans for cascade hydropower stations due to the inflow forecasting error,the concept of risk value is introduced to describe the potential maximum loss of power generation plan,quantify output error of cascade hydropower stations caused by the inflow forecasting error,and a short-term optimal operation model for cascade reservoirs considering output errors is constructed.Furthermore,for the defect that USOM-GA could not solve the uncertain optimization problem,the USOM-GA algorithm based on the extreme value theory is proposed to further improve the generality of the algorithm.Moreover,the calculation steps of the model solution are put forward,and the optimal power generation strategy under different risk preferences is solved to make its operation decision more flexible.Through the example application,the practicability and feasibility of the model and algorithm are verified.