| Under the comprehensive influence of climate change and human activities,the global underlying surface conditions and the spatiotemporal evolution of hydrological cycle elements have undergone profound changes,the characteristics of extreme climate and hydrological events have changed,the instability of hydrological rhythm has intensified,and the uncertainty has increased.Since the 21 st century,China's flood disasters have entered a period of frequent occurrence.From 1991 to 2017,the direct economic losses caused by floods in China increased at an average annual rate of 64%.Flood disasters have brought great threat to the sustainable development of social economy.In order to effectively cope with river basin floods,China has planned a large number of flood control and control pivot projects in different river basins.After nearly 60 years of water conservancy construction,China has initially formed a relatively complete combination of engineering measures and non-engineering measures for floods and droughts.The defense system and the river basin disaster prevention system have undergone a fundamental improvement.However,there are still obvious shortcomings and deficiencies in river basin flood response measures such as hydrological monitoring,forecasting and early warning,joint operation scheme formulation,and dynamic risk assessment.Therefore,there is an urgent need to carry out research on scientific,refined,and intelligent operation of flood control systems for river basins and reservoirs,in order to effectively promote the modernization of water conservancy and enhance the national flood and drought disaster defense capabilities under the new situation.However,large-scale mixed-connected reservoirs have complex topological structures,numerous flood control points,different flood control standards,difficulty in coordination between regional flood control and overall flood control,serious flooding in river channels,poor flood synchronization,and the randomness and complexity of the incoming water process.The flood control operation is a kind of multi-stage,multi-objective,multi-constraint,multi-variable panoramic optimization problem.In particular,traditional reservoir operation theories and methods are difficult to universally determine the size of the reserved flood control storage capacity,the timing of storage and the order of use of each reservoir under different inflow scenarios,which makes it difficult for the comprehensive benefits of water conservancy projects to continue to play,and it is extremely difficult to coordinate and control flood control risks.Therefore,aiming at the key scientific and technological problems faced by the panoramic operation and risk decision-making of reservoir group flood control system,combined with the relevant theories and methods of stochastic hydrology,artificial intelligence,optimal operation of reservoir group,system risk analysis and decision-making,this paper carried out the research on the stochastic simulation of uncertainty in multi station streamflow series,panoramic optimal allocation of reservoir group flood control capacity under stochastic inflow conditions,and coordination The whole flood control demand panoramic operation rule extraction,flood control panoramic operation scheme risk assessment and decision-making under the influence of uncertainty.The extraction of flood control operation rules and risk decision-making for reservoir groups under the panoramic perspectives of multiple inflow scenarios,multiple dispatch targets,multiple uncertain disturbances,and multiple risk constraints are realized.The research has achieved some valuable theoretical results,which can provide theoretical and technical support for further improving the flood control system of the reservoirs in the Yangtze River Basin and enhancing the flood disaster prevention capabilities.The main research contents of this paper are as follows:Aiming at the non-stationary,multi-dimensional,strong correlation,nonlinear and other multidimensional coupling characteristics of the inflow process of the mixed reservoir group in the basin,taking the self-and cross-correlation of the multi-station streamflow series of the main and tributary of the basin as the starting point,a spectral decomposition technique based latin hypercube sampling multiStation streamflow simulation(ILHS)method was proposed to simulate multi-site streamflow series.The simulation results were compared with the simulation results of the multi-station seasonal autoregressive model(MSAR)for verification and analysis.Furthermore,according to the design flood data of major downstream stations,by setting different sampling targets,the simultaneous simulation streamfow process of multiple stations was resampled,and the overall design flood process of the basin with different frequencies and encounter combinations was simulated and generated.Finally,adaptively adjust the inter-annual variation of the resampling parameters to generate a multi-station and multiscenario streamflow process with different long-term streamflow characteristics.The streamflow process of nine key control stations and reservoir control sections in the upper reaches of the Yangtze River was simulated by the proposed methods.The results show that,compared with the traditional model,the improved Latin hypercube sampling method can not only meet the basic statistical characteristics of the streamflow sequence,but also more effectively describe the high-order autocorrelation structure of the streamflow sequence of each station in the year;and the model can consider historical design flood data and Different possible high and low change trends of streamflow generate a multi-site,multi-scenario simulation streamflow process that meets the needs of different engineering applications.The research results can provide new modeling ideas for solving the random simulation of high-dimensional variables in hydrological time series,and can also provide random incoming water background fields for panoramic dispatch storage capacity allocation and risk analysis.Focusing on the problem of optimal allocation of flood control storage capacity of the reservoir group,taking full account of the availability of forecast information,a flood control optimal operation model based on forecast operation rules is established.First of all,constructing a collection of the storage capacity of the upstream reservoir group in use and the peak cutting capacity of the downstream flood control poin by optimiziation of different typical years,and fitting the flood control storage capacity benefit curve of each reservoir.on this basis,the optimal storage capacity allocation plan is obtained by optimizing the distribution of flood control storage capacity of the reservoir group under different typical incoming water conditions with the goal of maximizing the benefit of flood control storage capacity.Furthermore,considering the uncertainty of design flood area composition in the process of simulation and optimal operation,the flood control capacity benefit curve cluster is optimized.Through feature extraction and curve correction,the sample set covering the key flood feature vector,the upstream reservoir capacity in use and the peak shaving of downstream flood control points is constructed.A variety of machine learning algorithms are introduced to fit the nonlinear mapping relationship between the sample features.On this basis,a panoramic optimal allocation strategy of flood control storage capacity with multiple inflow scenarios and multiple standards is proposed.Taking the flood control systems of Xiluodu,Xiangjiaba and Pubugou as the research objects,the rationality and effectiveness of the proposed panoramic optimal allocation method for flood control capacity of reservoir groups are fully verified,and the compensation mechanism of flood control capacity of reservoir groups in the process of panoramic operation is clarified.The optimized flood control storage capacity allocation plan can provide theoretical guidance for the reservoir group flood control planning and design,and the reservation of flood control storage capacity decision allocation in real-time dispatch.Aiming at the extraction of large-scale reservoir group flood control panoramic operation rules,based on the existing reservoir group joint flood control operation system,reserved storage capacity is set to ensure the safety of flood control points in upstream areas.On this basis,using the preset method of operation experience and the random search method of decision tree,the decision variables such as storage timing,storage method and storage volume of each reservoir in the framework of joint flood control panoramic operation of reservoir groups are parametrically modeled.A model for extracting flood control panoramic operation rules for reservoir groups considering different complex typical inflow processes and coordinating the flood control safety of upstream and downstream areas is established,and a multi-objective intelligent algorithm is used to solve the flood control panoramic operation plan set.The research takes the flood control and dispatch system of the upper reaches of the Yangtze River as the research object.Based on the method of system decomposition and coordination,the dynamic relationship between upstream and downstream regional flood control and overall flood control and its optimization strategy are studied.Taking the flood control requirements of various regions as the optimal control boundary conditions,a multi-objective joint flood control panoramic optimization operation model of the twelve reservoirs in the upper reaches of the Yangtze River was established,and the frontiers of the optimal operation schemes obtained based on different operation frameworks were compared and analyzed.The results show that the frontier convergence of the panoramic dispatch scheme under the empirical preset dispatch framework is stronger,and the joint dispatch of upstream reservoirs can effectively reduce the pressure of the Three Gorges flood control and balance the overall flood control risk of the basin.The research results are helpful to realize the digital and logical expression of flood control operation rules for reservoir groups in the river basin,and meet the needs of the times for accurate flood control operation and refined management of reservoir groups in the new era.Aiming at the risk assessment and decision-making problems of the flood control panoramic operation plan under the influence of uncertainty.On the basis of analyzing the flood control capacity of reservoirs and undertaking flood control tasks,a risk analysis model of reservoir group flood control system based on uncertainty stochastic simulation is established.A dispatch risk assessment index system covering risk rate,conditional value-at-risk,and risk entropy has been constructed to quantitatively assess the risks of water level violations and flow exceeding standards under the coupling effect of multiple uncertainties,revealing the law of risk response under the influence of panoramic dispatch on the upper reaches of the Yangtze River.Furthermore,the priority discriminant matrix of the flood control operation of the reservoir considering the flood control storage capacity and multi-year average streamflow index of each reservoir is derived,and the weight coefficient of each reservoir is determined.Based on this,the Topsis method with improved weighted generalized Mahalanobis distance is introduced to make risk decisions for different operation schemes,and the rational decisionmaking and optimization of flood control panoramic operation schemes under multiple risk constraints are realized.Finally,the risk level of the flood control operation plan of the Xiluodu and Xiangjiaba Three Gorges cascade reservoirs in the lower reaches of the Jinsha River under the influence of uncertainty was evaluated from multiple angles.According to the proposed method,risk decisionmaking is carriedo out on the set of panoramic operation schemes for the joint flood control of the twelve reservoirs in the Yangtze River Basin obtained in Chapter 4.The research results can provide theoretical basis and technical support for risk quantification,assessment and decision-making of realtime joint flood control panoramic operation of reservoir groups affected by multiple uncertainties. |
PDF Full Text Request