摘 要：为了解决石灰石-石膏湿法烟气脱硫系统（Wet Flue Gas Desulfurization System，WFGD）的运行优化问题，基于某电厂600MW机组的脱硫系统历史运行数据，采用关联规则挖掘方法中的FP-tree（Freq-uent Pattern Tree）算法对历史数据进行关联规则的挖掘。在数据挖掘过程中，首先利用3法则去除原始数据中的粗大误差；其次利用聚类方法中的竞争凝聚（Competitive Agglomeration，CA）算法，对处理后数据的各个属性分别进行聚类，使其划分为不同的区间；然后选取不同锅炉负荷工况下的数据，利用FP-tree算法挖掘关联规则；最后挑选出有意义的关联规则。研究结果表明：将脱硫系统中的可调参数如浆液PH值、吸收塔液位、吸收塔给浆流量、吸收塔给浆密度等，按照挖掘出的结果进行设置，可以在满足脱硫效率的前提下使得脱硫的成本最低，从而提高电厂的经济效益。

Abstract: In order to solve the operation optimization problem of the Wet Flue Gas Desulfurization System (WFGD), based on the historical operation data of the desulfurization system of a 600MW unit of a power plant, the FP-tree (Frequent Pattern Tree) in the association rule mining method is adopted. The algorithm mines the association rules of historical data. In the data mining process, the 3rules are used to remove the coarse errors in the original data; secondly, the Competitive Agglomeration (CA) algorithm in the clustering method is used to cluster the attributes of the processed data to divide them for different intervals; then select the data under different boiler load conditions, use FP-tree algorithm to mine association rules; finally select meaningful association rules. The research results show that the adjustable parameters in the desulfurization system, such as the pH value of the slurry, the liquid level of the absorption tower, the flow rate of the absorption tower, the slurry density of the absorption tower, etc., can be set according to the results of the excavation, which can meet the desulfurization efficiency,the lowest cost of desulfurization is achieved, thereby increasing the economic efficiency of the power plant.

国家自然科学基金项目（61573239）；上海发电过程智能管控工程技术研究中心资助项目（14DZ2251100）