Research On Multi-objective Optimal Operation And Decision Methods For Cascaded Reserviors

Under the background of energy shortage, energy conservation and emissions reduction, strength of hydropower endevelopment in China enhances and pace speeds up, meanwhile the scale of reservoir groups gets bigger and bigger, and the joint operation problem is becoming more and more complicated. On the one hand, there are complex hydraulic and electrical connections between upstream and downstream for cascaded reservoirs(complex constraints), and as well as the influence of incoming runoff uncertainty there is certain gap between theoretical research and practical application of reservoirs optimal operation; on the other hand, with the improvement of human living standards and knowledge, the comprehensive utilization demand of large water control peojects is growing and a certain competition and conflict relationship exists in different operation goals, thus there is certain limitation to use the traditional theories and methods of reservoir optimal operation to solve problems of reservoirs optimal operation in the new period, unable to meet the requirements of comprehensive benefits for reservoirs multi-objective operation, and new theories and methods of reservoirs joint operation are necessitated in present. Around multi-objective joint operation of cascade reservoirs and and programs decision problem between different targets under the the background of rapid development of hypowerelectric energy, aiming at maximizing reservoirs long-period comprehensive benefits, combined with theory of systems engineering, intelligent optimization methods, and multi-objective optimization and decision theory, we deeply study the joint operation problem of cascade reservoir from several aspects, as mid-long term runoff forecast, single objective optimal operation, water-sediment multi-objective optimal operation and decision joint operation and a series of research results with theoretical and practical value are achieved. The main research results in this paper are listed as follows:(1)Mid-long term inflow runoff forecast method of reservoir has been researched. Firstly autoregressive moving average model, nearest neighbor bootstrapping regressive model, artificial neural network model are introduced as reservoir runoff forecast models, the related model parameters are calibrated, and the rationality and reliability of those models are verified by relevant indexes; Based on those models, combined with the cyclical and years non-uniformity characteristics of month inflow runoff, monthly combination forecast model based on support vector machine (SVM) is established, then indicators of the average relative error and determination coefficient are adopted to set up runoff forecast precision-evaluation model; Take inflow runoff forecast of Longpan hypdropower station for example, results show that combination forecast model can reduce the uncertainty of single forecast model and obtain higher forecast accuracy, thus the rationality and superiority of the proposed model are verified.(2)Optimal operation methods of cascade reservoirs have been researched. Combined with good global optimization performance of shuffled frog leaping algorithm and good local search ability of immune clonal selection algorithm, this paper immune leapfrog algorithm is proposed in this paper and the corresponding process is given; Meanwhile for hydraulic and electrical link between cascaded reservoirs, in order to improve optimization efficiency of intelligent optimization algoriths while solving reservoirs optimal operation problems, methods of building the feasible region of reservoirs optimal operation problems and generating initial solutions are raised. Taking mid-long term optimal operation and short-term optimal operation of cascade reservoirs as an example respectively, together with corresponding conventional optimization methods, the feasibility and superiority of proposed shuffled frog leaping algorithm in solving optimal operation problems of cascaded reservoirs are verified.(3)Optimal operation of cascade reservoirs based on data mining has been researched. Taking optimal operation process obtained by deterministic optimal operation of cascade reservoirs as basic data, operation sample sets including time factors, space factors, energy factors and decision variables is firstly built, then rough set method is introduced to select decision variables and influence factors of monthly operation function, finally, support vector machine (SVM) is employed as fitting method of operation function, and combined the analysis of key problems in long-series operation operation simulation the application process of implicit stochastic optimization operation of cascade reservoirs based on data mining is given. Taking the operation function extracting of cascaded reservoirs located in Yangtze river middle reaches for an example, the rationality and superiority of the proposed method is validated, which provides new ideas to look for scheduling rule and extract joint scheduling rules of cascaded reservoirs.(4)Water-sediment multi-objective optimal operation of cascaded reservoirs has been researched. Aiming at contradictions between long-term benefits and sand sedimentation reduction for cascaded reservoirs on alluvial rivers with heavy sediment load, based on one dimension model of sediment deposition calculation, multi-objective optimal operation model for water-sediment of cascaded reservoirs is firstly bulit; And considered that traditional water-sediment joint operation usually adopt loose coupling calculation method (reservoir operation and sediment deposition calculation seperately), combined with multi-objective dynamic programming iterative algorithm, the dimension reduction algorithm combined with dynamic programming successive approximation and multi-objective dynamic programming iteration algorithm is put forward and used to solve the problem of water-sediment joint operation for cascaded reservoirs, and then considered that water-sediment calculation does not meet no aftereffect of dynamic planning and its calculation need a long time, thought transformation that takes sand sediment as basic goal and electricity as constraints respectively is adopted, and the corresponding solving process is presented. Finally taking Three Gorges reservoir as an example, based on multi-objective dynamic programming iteration algorithm water-sediment non inferior solution set is obtained, so the rationality and validity of the proposed method is verified.(5)Scheme evaluation of water-sediment multi-objective optimal operation for the reservoir has been researched. Considered that traditional TOPSIS decision-making method doesn't take the correlation between indicators into count and only adopt subjective weight, advanced TOPSIS multiple attribute decision-making method based on markov distance is introduced; Then evaluation system of water-sediment coordinative operation program for the reservoir is set up, and the calculation method of relevant indexes and the correlation between indicators are analyzed; Finally takeing scheme sorting of water sand dispatching non inferior solution set in Three Gorges reservoir for an example, evaluation schemes ranking results and the equilibrium solution are obtained, so the rationality of the proposed evaluation system model of water-sediment coordinative operation program and decision-making method is verified.