基于时序数据融合的乙烯装置能效价值研究及应用

Time-series data fusion and its application to energy efficiency value for ethylene plants

为了获取乙烯装置的能效价值,提出基于k-means聚类的多变量时序数据线性最小方差优化融合算法及其证明。采用改进的k-means聚类算法对乙烯装置能耗相关时序数据进行聚类分析,获取多变量时序数据的分类及其聚类中心,以各聚类中心计算每个时序变量均方差,并用线性最小方差优化融合算法对各聚类中心加权向量融合,提取乙烯装置能效价值的多维数据虚拟标杆。提出采用多维能效价值雷达曲线展示虚拟标杆与能耗相关的多维数据,直观揭示各年度装置的运行状态与操作水平,准确找出产生能耗的主要原因及提高能效的主要方向。提出的方法不但为乙烯装置的能效评估提供了可行方法,也可应用到其他流程装置的能效评估。

In order to obtain energy efficiency value of ethylene plants,multivariate time series data linear minimum variance optimal fusion algorithm based on k-means clustering was proposed.The improved k-means clustering algorithm was adopted to analyze time-series data of energy consumption of ethylene unit with related multi-variables.So the various groups and cluster centers could be obtained,and the calculation of each variable variance in all kinds of groups was provided based on cluster centers.And then comprehensive vector-weighting was used to fuse all clustering centers,and realize multi-dimensional data virtual benchmarking of energy efficiency value of ethylene plants.Virtual benchmarking and multi-dimensional data about energy efficiency were shown by multi-dimensional energy efficiency value Radar curve.The operational status and operational levels were revealed directly,and the main reason for energy consumption was easily identified to improve energy efficiency.The proposed algorithm provided a method of energy consumption evaluation not only for ethylene plants,but also for other process plants.