基于电厂能效监测平台的机组特性在线识别与可视化

Unit Characteristics of On-line Identification and Visualization Based on Power Plant Energy Efficiency Monitoring Platform

基于省级热电厂能效监测平台的运行数据,采用M5′模型树方法在线识别和绘制了较为精确、详细的单抽和双抽机组工况图。该方法采用分段式线性拟合,自动优选分段数并逼近可能非凸或非连续的非线性耗量曲线。利用该方法进行机组定期滚动识别对比,可以更好地察觉机组特性的变化,为机组能效监视提供可视化感知工具,为测量数据提供校验依据,并为电厂热电负荷优化分配和电网节能调度提供精确的分段线性化机组特性函数。

The MS＇ model tree algorithm is used to on-line identify and draw the accurate and detailed single and double extraction unit working condition charts. This method automatically and optimally selects the number of sections and approaches the nonlinear consumption curves might be with non-convex and discontinuous property using piecewise linearization fitting. By comparison of scrolling identified characteristics at a fixed period, the change of unit can be detected and an awareness visualization tool of unit can he achieved. This method can provide the check basis for measured data and provide accurate piecewise linearization function of unit characteristics for load optimal sharing of thermal power plant and energy saving scheduling of the power grid.