Research On The Optimization Method Of The Medium And Long Term Power Generation Operation Of Reservoir Group

With the rapid development of Chinese economy,the demand for energy,especially clean water energy,is growing rapidly.The cascade reservoirs in each basin have been built and put into operation one after another.The joint optimal scheduling of hydropower energy system composed of large-scale hybrid cascade reservoirs across regions and basins has taken shape.However,the joint optimal scheduling of large-scale hybrid cascade reservoir group not only obtains the significant comprehensive benefits of flood control,shipping,ecological water supply,but also intensifies the complex,high-dimensional and strong coupling characteristics of the hydropower energy system,which makes the complex phenomena and technical problems in the optimal scheduling and management of the system increasingly prominent,bringing new challenges to the joint optimal scheduling of the cascade reservoir group in river basin.The joint optimal scheduling of cascade reservoirs is affected by hydrologic uncertainty,various profit demand and other factors,and there is a close relationship between hydraulic and electric power between reservoirs.It is a typical complex constrained optimization problem with high dimension,nonlinearity and strong coupling characteristics.The efficient and stable solution method of the problem has always been a hot and difficult issue in the research field of water resource management.Therefore,this paper studies the theory and method of characteristics of of single reservoir,cascade reservoirs and large-scale reservoir group under different scales based on several scientific problems faced by the joint optimal operation of cascade reservoirs in the basin,combined with the theories and methods of hydropower energy,operation research and system dynamics,so as to realize the comprehensive and efficient utilization of hydropower resources in the basin The standard has been studied in depth,and some research results which can provide reference for the actual production and operation of cascade reservoirs have been obtained.The main research contents and innovative achievements of this paper include:(1)Starting from the research on the efficient solution for the medium and long term optimal power generation scheduling of a single reservoir,a two-stage efficient optimization method for optimal of medium and long term optimal power generation scheduling based on a two-stage analysis is proposed.The research work first briefly outlines the medium and long term optimal power generation scheduling of a single reservoir and its dynamic programming optimization method,and then establishes a two-stage model for the medium and long term optimal power generation scheduling of a single reservoir,analyzes the inherent unevenness and monotony of the power generation benefit objective function of the two-stage model,deduces the optimal criterion of the carryover state of the two-stage model;the idea of the dividing and conquering is introduced to design a strategy to transform the multi-stage optimization problem into a progressive two-stage problem,a two-stage optimization method based on the optimal criterion for the two-stage model is proposed.The Three Gorges reservoir is taken as the research object to carry out the case study.The experimental results show that the optimal results of the method proposed in the study are close to the global optimal solution obtained by dynamic programming,and the calculation time is millisecond level,which significantly improves the real-time performance of the medium and long term optimal power generation scheduling of a single reservoir.(2)Aiming at the problem of long,medium and short term nested optimal forecasting power generation scheduling of the reservoir under random inflow conditions,system dynamic modeling method for long,medium and short term nested forecasting and dispatching with response to forecast deviations is established.Based on the existing research of nested dispatching of reservoir,the coupling nesting relationship between dispatching models of different scales of the reservoir is analyzed,and the long,medium and short term nested forecasting power generation model of reservoir is established.The influence of forecast error on the optimal decision of forecasting power generation scheduling is analyzed.Combined with the feedback mechanism of system dynamics and the classical control theory,a reasonable and effective feedback control response mechanism of the nested forecasting power generation scheduling for the forecast deviation is proposed.According to the heuristic information of the water level deviation caused by the deviation between forecast inflow and real-time inflow,the scheme of short-term deviation forecast operation is modified adaptively and dynamically.On this basis,the system dynamic modeling method for long,medium and short term nested forecasting scheduling is proposed.The case study of the Three Gorges reservoir is carried out and compared with the operation chart,stochastic dynamic programming and the traditional nested forecasting scheduling optimization method.The experimental results show that the proposed method can respond to the deviation between forecast inflow and real-time inflow in the actual operation process in time and make more reasonable forecasting scheduling decision,so as to improve the benefit of optimal power generation scheduling for reservoir under the uncertainty of inflow.(3)In order to overcome the shortcomings of local convergence and poor convergence stability for the intelligent optimization method in solving the problem of medium and long term optimal power generation scheduling of cascade reservoirs,and to deal with multiple complex constraints such as time and space coupling,magnitude and dimension can not be unified in the process of optimization,this paper proposes a differential evolution algorithm with self-adaptive updating strategy of partition parameters and a constraint handling strategy considering the different priority between equality,rigidity and flexibility constraints of cascade reservoirs.Based on the analysis of the advantages and limitations of the existing constraint handling strategies of intelligent optimization algorithm,this paper puts forward the evaluation criteria of the unified flow constraint and constraint violation degree of cascade reservoirs,and designs the constraint handling strategy considering the different priority between equality,rigidity and flexibility constraints of cascade reservoirs.At the same time,an improved differential evolution algorithm with self-adaptive updating strategy is proposed to improve the convergence performance of the algorithm.Same examples are carried out on the lower reaches of Jinsha River and Three Gorges cascade reservoirs.The results show that the proposed partition parameters adaptive differential evolution algorithm has achieved better performance in the test function experiment and the case study of medium and long term optimal power generation scheduling of cascade reservoirs.Combined with the proposed constraint handling strategy,it can improve the convergence performance of the intelligent optimization method in solving the problem of optimal operation,on the premise of ensuring that the operation results meet the rigid constraints related to the safe operation of the reservoir and violating the flexible constraints as little as possible.(4)Focusing on the problem that the multi-objective benefits of optimal operation of cascade reservoirs are not commensurable and the mutual feed relationship is difficult to be quantitatively analyzed in multi-objective optimal scheduling of cascade reservoirs,a gradient analysis method of mutual feed relationship of multi-objective optimization is proposed,which combines the multi-objective optimization theory and the concept of gradient.The joint optimal scheduling model of water supply,power generation and environment of cascade reservoirs is established,and the non-dominated solution set of Pareto front of the multi-objective optimal scheduling is obtained by constraint method and dynamic programming.The spatial distribution characteristics of Pareto front solution set of the multi-objective optimal scheduling of cascade reservoirs is analyzed.Combined with multi-objective optimization theory and gradient analysis method,the Pareto frontier solution space function of multi-objective optimization results is constructed,and the partial derivative function values and gradients among objectives in the normalized objective space are calculated,so as to quantify the mutual feed relationship among multiple objectives,which provides a new research idea for the benefit conversion and mutual feed relationship analysis of multi-objective optimal scheduling of cascade reservoirs.(5)In view of the characteristics of large-scale reservoir group,such as high dimension,strong coupling and nonlinearity,an improved successive approximation dynamic programming based on the monotonicity of two-stage resource allocation problem and relaxation strategy is proposed.Firstly,the monotonicity of two-stage resource allocation problem with the characteristics of concave function is summarized.Then,with the assumption of no deserted outflow and neglecting the coupling relationship between reservoir capacity and discharge,it is proved that the recursive function of power generation of single reservoir optimized by dynamic programming is a concave function in successive approximation dynamic programming modeling and solving the problem of medium and long term joint optimal power generation scheduling of hybrid reservoir group.Combined with the monotonicity of the two-stage resource allocation problem based on the concave function,an improved successive approximation dynamic programming method with relaxation strategy of the two-stage resource allocation problem is proposed.Taking the large-scale hybrid reservoir group in the upper reaches of the Yangtze River with 61 reservoirs as the research object,the case study of medium and long term joint optimal power generation scheduling of large-scale hybrid reservoir group is carried out.The relevant experimental results show that the proposed improved successive approximation dynamic programming significantly improves the calculation speed while ensuring the superiority of the results,so as to provide an efficient and stable optimization method for the medium and long-term joint power generation operation of large-scale hybrid reservoirs,which is helpful to improve the comprehensive utilization efficiency of regional water resources and operation and management level of hydropower energy system composed of large-scale reservoir group.