浆液循环泵作为火电厂湿法脱硫系统的关键能耗设备，传统依靠操作人员经验运行，缺乏精细化管理的手段，导致运行能耗偏高。本文针对上述问题，提出了一种基于数据挖掘的浆液循环泵组合运行优化方法。通过模糊C-均值聚类算法对历史运行数据进行特征相似组的聚类，并构建以烟囱出口SO2浓度为约束的能耗目标函数，从而在聚类中筛选出最优泵组合形成历史工况库。利用该库训练了LightGBM分类器，实现了泵组合的智能优化运行。本文所提优化方法实际部署于某650MW火电机组，结果表明，相比传统经验运行方式，脱硫能耗环比降低6.6%，同比降低7.3%，证明了该优化方法的有效性和实用性。

As the key energy consumption equipment of the wet desulfurization system of thermal power plant, the slurry circulating pump traditionally relies on the experience of operators and lacks the means of fine management, resulting in high energy consumption during operation. To solve the above problems, this paper proposes a slurry circulating pump combination operation optimization method based on data mining. The fuzzy C-mean clustering algorithm is used to cluster the historical operation data with similar characteristics, and the energy consumption objective function is constructed with SO2 concentration at the chimney outlet as constraint, so as to screen the optimal pump combination in the clustering and form the historical working condition database. The LightGBM classifier is trained by the library, and the intelligent optimal operation of pump combination is realized. The optimization method proposed in this paper is actually deployed in a 650MW thermal power unit. The results show that compared with the traditional empirical operation mode, the desulfurization energy consumption is reduced by 6.6% on a year-on-year basis and 7.3% on a month-on-month basis, which proves the effectiveness and practicability of the optimization method.

国家能源投资集团有限责任公司科技项目（HJ-22-KJ|2022|4）