数据挖掘技术已广泛应用于传统火电行业的节能优化运行与管理，但目前基于数据挖掘的机组运行优化方法往往缺乏普遍性和实操性，针对少量典型工况挖掘得到的参数目标值，无法给予实际运行机组全面的指导。本文在分析锅炉和汽轮机等主要子系统性能指标影响因素的前提下，建立基于层次划分的火电机组能耗指标体系，提出了数据挖掘与火电机组能耗机理分析耦合的经济性诊断模型，有效提高了输出的鲁棒性。以供电煤耗为例，基于历史数据挖掘的火电机组经济性诊断模型和流程，准确分析出能耗水平高于报警阈70%是由于实时排烟氧量偏离基准值造成的，并给出符合实际工况的调节建议。

Data mining technology has been widely used in the energysaving optimization operation and management of traditional thermal power industry, but the current unit operation optimization methods based on data mining often lack universality and practical operation. The parameter target values mined for a small number of typical working conditions cannot give comprehensive guidance to the actual operation of units. On the premise of analyzing the factors affecting the performance indexes of main subsystems such as boiler and steam turbine, this paper establishes the energy consumption index system of thermal power units based on hierarchical division, and puts forward the coupled economic diagnosis model of data mining and energy consumption mechanism analysis of thermal power unit, which effectively improves the robustness of output.Taking the coal consumption of power supply as an example, through the economic diagnosis model and process of thermal power units based on historical data mining, it is accurately analyzed that the reason why the energy consumption level is 70% higher than the alarm threshold is due to the deviation of realtime exhaust oxygen content from reference value, and the practical improvement scheme is put forward.

TM621