中文摘要：

传统厂级负荷优化分配以火电机组煤耗曲线为依据，以供电煤耗率最低为目标。考虑到火电机组结构日益复杂，多变的边界条件和运行工况加剧了机组能耗特性的不确定性，给厂级负荷优化分配带来新问题。该文基于火电机组的海量运行数据，引入大数据分析方法，通过模糊粗糙集计算方法提高数据处理的效率，利用决策相关函数评价能耗决策的置信度，获得机组不同边界和运行工况下的能耗特性。将得到的机组供电煤耗率作为厂级负荷动态规划的依据，进而预测负荷优化分配的节煤潜力。结果表明，基于大数据分析方法的厂级负荷分配可有效降低火电厂的供电煤耗率，对火电机组的节能发电调度具有参考意义。

英文摘要：

On the basis of coal rate curves, the traditional plant-level load dispatching is fulfilled with the minimum coal rate. Considering the increasing complexity in power unit structure, it is of great uncertainties to describe the energy consumption and difficult for the plant-level load dispatching especially under the varying operation boundary and conditions. With a great volume of operation data of thermal power units, big data-based analytics were introduced. The fuzzy rough set （FRS） method was used to improve the efficiency of data processing, and the decision correlation function was introduced to measure the confidence, so as to derive the coal rate under specific working conditions. Taking such coal rate as the basis of plant-level dynamic load planning, the energy-saving potential was predicted for the optimized load dispatching. The result shows that the load dispatching based on big data analytics is effective to reduce the gross coal rate, making great reference for the energy-saving power generation dispatching.