[关键词]
[摘要]
针对目前厂级负荷分配中煤耗模型难以准确建立等问题,提出了一种基于历史数据挖掘的厂级自适应经济负荷分配的建模方法。首先,根据历史数据中性能参数相对于边界参数的条件概率分布,利用分位数回归筛选运行经济性优良的历史数据为目标样本,建立不同工况下总负荷与各机组所分配负荷之间的映射关系。其次,考虑到煤耗特性容易时变的特点,设计了一种基于当日运行样本的煤耗特性自适应机制;最后,分析了分位数回归对调度结果的影响,以及最新采集的数据流的分布对自适应能力的影响。算例仿真和工程实例分析的结果表明,所提方法在保证较高优化精度的同时,可以自动适应煤耗特性的突变或漂移,保证对理论最优解的跟踪,具有较高工程应用价值。
[Key word]
[Abstract]
In view of the current difficulties in establishing the precise coal consumption rate model for plantlevel load distribution,a plantlevel adaptive and economic load distribution modeling method based on historical data mining is proposed.Firstly,according to the conditional probability distribution of the performance parameters relative to the boundary parameters in the historical data,the historical data with better operation economy are screened out as target samples by quantile regression,to establish the mapping relation between the total load and the dispatched load of each unit under different working conditions.Secondly,an adaptive mechanism of coal consumption characteristics based on dayrunning sample is designed,considering the timevarying characteristics of coal consumption.Finally,the effect of quantile regression on the dispatch results and the influence of the distribution of newly collected data streams on the adaptive ability are analyzed.The analysis results of numerical simulation and engineering examples show that the modeling method can automatically adapt to the mutation or drift of the coal consumption rate,and ensure tracking the theoretical optimal solution while ensuring higher optimization accuracy,which has higher engineering application value.
[中图分类号]
TM621
[基金项目]