【作者】 孙昌佑；

【导师】 马震岳；

【作者基本信息】 大连理工大学 ， 水利水电工程， 2004， 硕士

【摘要】 水电站厂内经济运行是在满足各种约束条件下，合理选择水电机组台数与组合，经济分配系统负荷和旋转备用容量，以期获得最大的经济效益。国内外学者研究厂内经济运行时，一般采用忽略多种约束的简化模型，与实际运行相差较大，难以满足实时发电调度的要求。为此，本文针对水电站厂内经济运行中需要考虑的关键问题，建立了强调水电机组启停成本、避开汽蚀振动等限制运行区间和考虑强制性开停机要求以及最小开停机时间约束的水电站厂内经济运行的数学模型。根据该模型的特点，分解成两个子模型—最优机组组合和最优负荷分配模型，提出了一种基于遗传算法求解的新思路。本文的主要研究工作及成果概况如下： 1．对于最优负荷分配模型，在以往的优化模型中，一般只考虑水电机组最小、最大技术出力限制条件。但是，在实际负荷分配过程中，存在机组的汽蚀和振动区等所谓阈区间限制问题，这类问题由于影响水轮机效率、使用寿命、甚至可能危及机组运行安全，因此求解机组最优负荷分配问题时必须考虑这些约束条件。为此，本文在分析机组出力运行工况的基础上，把机组出力范围(最小、最大技术出力)细分为安全运行区和非安全运行区。这样，机组出力范围被非安全运行区间断为一系列离散的子区间，较以往的模型中只考虑最小、最大技术出力限制有了很大的改进。针对上述最优负荷分配模型中水电机组安全运行区是一系列离散子区间构成的特点，把最优负荷分配模型抽象为等式约束下的离散可行域约束优化数学模型，改进了遗传算法，采用十进制编码技术，并设计了合适的遗传操作算子有效地处理模型中的离散可行域约束，使得算法在遗传操作迭代过程中的所有个体都是可行解。与常用的惩罚函数法处理约束的方法进行了数值试验比较分析表明，算法可行、有效。 2．对于最优机组组合模型，由于水电机组启停成本低，运行限制少的缘故，在以往的优化模型中，很少考虑研究周期内的机组开停机次数，但长时间范围内来看，机组开停机不仅有水量的消耗，还对机组造成一定磨损，若短时间内频繁开停机，将造成机组运行寿命减少和机组效率降低。为此，本文为模型增加了最小开停机时间约束和强调水电机组启停成本来避免频繁开停机现象。另外，机组在实际调度中，还需考虑强制开停机、机组固定出力等有效性约束。鉴此，本文提出了强调水电机组启停成本、考虑最小开停机时间约束以及强制开停机等有效性约束的最优机组组合模型。针对最优机组组合模型特点，设计了一些启发式技术，指导遗传算法初始种群的生成，使得所有个体均是可行解，而且大大节省了随机方法产生可行解浪费的时间，同时，采用二进制、十进制混合编码策略有效地节省了种群的存储空间。 3．本文提出的基于遗传算法的水电站厂内经济运行模型，在丰满水电站应用研究表明，模型合理，算法可行、有效，为制定水电站厂内经济运行方式开辟了一条新的途径。 最后对全文进行了总结，并对有待进一步研究的问题进行了展望。更多还原

【Abstract】 The innerplant economical operation in hydroelectric power station is a problem of selecting hydraulic units to be in service during a scheduling period, and to optimally allocate the load demand and the spinning reserve among the running units while satisfying a variety of contains. This calculated schedule is proposed in order to get maximum economical benefit. The simplification model of innerplant economical operation was proposed commonly while neglecting a great deal of constraints when scholars at home and abroad have been studying the problem. So this thesis presents a model, aiming at the key questions for innerplant economical operation. In this model, besides traditional constraints such as power balance and capacity limits constraint etc., many constraints such as the cost for start-up/shut-down status of units, compulsory constraints for unit start/shut, minimum up/down time limits and spinning reserve requirements are also considered. According to the characteristics of the model, this model can be divided into two sub-models ,namely the "unit commitment"(UC) decision and the "economic dispatch"(ED) decision. An advanced genetic algorithm approach to the problem is discussed. The major research works are as follows:1. ED problem. The units minimum and maximum output operating limits were only consider commonly in the past optimization models. But in the actual ED problem, there are a large number of units operating limits (e.g. vapor eroding, or vibration etc,) which units must avoid running. If not considering the units operating limits above, the efficiency and the life are reduction; moreover, it could lead to danger for units. So this thesis divides the units operating limits into safe and unsafe range. The safe range of units is a set of sub-ranges and interrupted by unsafe one. There is greater improvement on the model, which only considering minimum and maximum output operating limits of units previous. According to the character of the above ED model, this thesis abstracts a discrete feasible constraints model with a constraint of equation. In the proposed algorithm, genetic algorithm solution is coded as decimal representation. Genetic operators are designed to coincide with discrete feasible constraints that every individual is a feasible solution to the operating processes. Case study also showed that the model is reasonable and the proposed algorithm is feasible and efficient, comparing with the most common penalty function method handling constraints.2. UC problem. Owing to the cheap cost for start-up/shut-down of hydraulic units and few operating limits, the time for start-up/shut-down of units was not considered in the previous optimal models during a scheduling period. But in a long time range, start-up/shut-down of hydraulic units leads to not only consume water, but also wear and tear. If the status of units were changed frequently in short time, the life and the efficiency of units were reduced. So the thesis presents the optimal model, adding minimum up-time and minimum down-time constraints and insisting on the cost for start-up/shut-down of hydraulic units in order to avoid the phenomenon above.Furthermore, in the actual operation, it is believed that unit’s availability must be considered (e.g. must run, unavailable, available, or fixed output etc.). To summarize above, this thesis present a UC optimal model, insisting on the cost of start-up/shut-down of hydraulic units and considering compulsory constrains for units start/shut, minimum up/down time limits, or other constrains for unit’s availability. According to the characteristics of above unit commitment model, some heurist techniques are designed to guide producing initial population, which every individual is a feasible solution and it spends less time on producing initial population than the random method. Furthermore, in the proposed algorithm, genetic algorithm solution is coded as a mix of binary and decimal representation to save the storage size for population.3. The proposed model is更多还原