【作者】 马超；

【导师】 练继建；

【作者基本信息】 天津大学 ， 水利水电工程， 2005， 硕士

【摘要】 本文对梯级水电站竞价上网和短期优化调度以及电价预测三方面的理论、方法、应用进行了全面深入的研究,主要工作可归为以下几个方面:第一部分,详细介绍了蚂蚁算法的基本理论和发展状况。在保留蚂蚁算法核心思想的条件下,针对梯级水电站短期优化调度问题的特点,通过引入遗传算法的交叉和变异思想,结合自适应搜索步长方法,提出了改进蚂蚁算法。并以最小耗水率模型为例,构建了基于改进蚂蚁算法的梯级水电站短期优化调度模型。为了使机组启停优化和耗水量最小两个目标同时达到最优,基于改进蚂蚁算法的梯级水电站短期优化调度模型分为机组启停组合优化子模型和最优负荷分配子模型两个部分,其中在机组启停组合优化子模型中引入了可行性检查和冗余检查方法,从而降低了模型寻优的难度,并提高了搜索的速度和效率。最后通过龙羊峡-李家峡梯级和硕多岗梯级两个实例验证了所建模型的可行性和实用性。得出了设定工况下较优的梯级调度运行方案。第二部分,分析电价的重要性和电价预测的特点,系统地阐述了国内外电价预测的理论和方法。结合已有的数据资料,构建了电价预测人工神经网络BP模型,并利用Levenberg-marquart方法对模型进行了改进,以提高模型的预测精度。研究模糊数学理论,利用模糊聚类法和模糊神经网络进行电价预测,构建模糊神经网络电价预测模型,以弥补BP神经网络对突变点数据预测误差较大的缺陷。最后通过实例对所建模型的可行性和有效性进行了验证,取得较好的预测结果。第三部分,详细阐述了电力市场的运行规则、竞价模式等方面的理论知识。通过对电力市场中发电厂商竞价策略的分析,提出了市场竞价下梯级水电站的竞价策略,构建了市场竞价下梯级水电站短期优化调度模型。模型采用概率分布来模拟发电厂商报价策略,并利用蒙特卡洛方法对电力市场出清过程进行随机模拟,而且在最大效益子模型中考虑了梯级各电站发电成本不同的因素。最后通过实际算例,对梯级水电站在市场竞价条件下的整体运行生产过程进行了模拟优化求解,得出了设定工况下较优的电价模拟结果和梯级运行方案,并通过对优化结果的详细分析,指出了模型的优点和缺陷,为下一步研究或实际运行奠定了基础。更多还原

【Abstract】 This paper is concerning with the research on theory, method and application ofthree parts, which are electricity price bidding and short-term optimal dispatching ofcascaded hydropower stations and electricity price forecasting. The major work of thisdissertation is outline as following:In the first part, the basic theory and development of ant algorithm are introduced.Based on the main theory of ant algorithm and aiming at the characteristics ofshort-term optimal dispatching of cascaded hydropower stations, improved antalgorithm is put forward by using the crossover and mutation of genetic algorithm andthe technique of adaptive search radius. Then a minimum water consumption modelbased on the improved ant algorithm is constructed, which includes start-up &shut-down optimization sub-model and load distribution optimization sub-model inorder to realize the two aims together. In start-up & shut-down optimizationsub-model, feasibility-checking method and redundancy-checking method areintroduced, so the optimization difficulty is reduced, the search speed and efficiencyare enhanced. At last, by the application to the Longyangxia-Lijiaxia cascadedhydropower stations and the Shuoduogang cascaded hydropower stations, thefeasibility and practicability of the constructed models is validated, and optimaldispatching programs are also attained.In the second part, this paper analyzes the characteristics and importance ofelectricity price forecasting, and also expounds the theory and method in this area indetail. On the basis of existing data, BP model in ANN for electricity price forecastingis constructed, and the forecasting precision is also improved by usingLevenberg-marquart method. Based on fuzzy theory, this dissertation takes advantageof fuzzy clustering and fuzzy neuron network to do electricity price forecasting, inorder to remedy the shortage of low precision by the BP model in break points. At last,examples are used to demonstrate the feasibility and availability of the constructedmodels, which gains good forecasting results.In the third part, this paper introduces the basic theory of operation rules andbidding modes of electricity market. By the analysis to the bidding strategies ofgenerators, bidding strategies for cascaded hydropower stations in electricity marketare put forward, and a whole short-term optimal dispatching model is also constructed.In this model, this dissertation choose probability distribution to simulate pricebidding of generators, and take advantage of the Monte Carlo method to simulate thecourse of price sorting randomly. It also considers different generation cost ofdifferent stations in maximum profit model. At last, the whole dispatching course ofcascaded hydropower stations in electricity market is simulated, which attains goodexpected price results and dispatching programs. Then by the analysis to the optimalresults, the advantage and disadvantage of the constructed models are pointed out,which establishes well base for future research or real dispatching.更多还原