摘要： 提出了一种新颖的可在线计算的时间序列启发式算法。算法具有多边形约简算法相同的优良的近似质量，并可在固定数据缓冲区空间内在线运算。用启发式搜索方法自动获取最佳分段数。在随机时间序列上仿真试验证明算法有很高的逼近质量和较低的计算复杂性。

关键词: 时间序列, 多边形边界约简, 数据挖掘, 启发式搜索

Abstract: The novel online heuristic algorithm of representation based on polygonal boundary reduction is presented in detail. The quality of approximation is as good as polygonal boundary reduction, and it can process the data online in a constant buffer. The number of segments is automatically computed due to the inherent heuristic searching. Higher quality of approximation is accomplished in simulation on stochastic time-series shows that, and also lower computing complexity does.

中图分类号: 

• TP311.13