基于动态时间弯曲算法的相似洪水识别方法

相似洪水动态识别是在大数据背景下弥补实时洪水预报预见期不足的有效途径,对于支撑防洪调度具有重要作用。由此,结合产汇流理论建立反映洪水形成发展的复杂多元动态事件集;采用动态时间弯曲算法改进了欧氏距离算法对相位的适应性;基于多元时间序列相似性原理构建了相似洪水动态识别方法。以池潭水库30场雨洪资料相对完备的洪水为例,对所提方法进行验证,结果表明该方法具有较好的洪水识别效果。

Similar Flood Recognition Method Based on Dynamic Time Warping Algorithm

With the background of big data, dynamic identification of similar flood appears to be an effective way to overcome the drawback of short forecast lead time in real-time flood forecast, which may provide useful information for flood control operation. Therefore, based on the runoff generation and routing theory, we established a complex multivariate dynamic event set which can reflect the formation and development of flood events. Dynamic time warping (DTW) algorithm was used to improve the phase adaptability of Euclidean distance algorithm. Then, a similar flood dynamic identification method was established based on the similarity analysis theory of multiple time series. Taking the 30 relatively complete flood and rainfall data in Chitan reservoir as an example, the proposed method was verified. The results show that the proposed method has a good effect on flood dynamic identification.