基于词嵌入的元组级数据溯源方法

在信息爆炸时代,数据量与日剧增,使用数据挖掘技术可挖掘其内在联系,但前提是所使用的数据正确无误,否则其后的一切工作将毫无意义。数据溯源技术可帮助数据分析人员快速定位到错误数据的来源和加工过程,减少错误数据的分析时间和难度,对数据质量控制与可信管理具有重要价值。现有的元组级数据溯源方法存在存储开销大和溯源效率低的问题,文章使用词嵌入技术改进元组级数据溯源方法。首先,研究元组向量化编码机制,依据元组向量相似度识别元组溯源关系;其次,提出基于属性重要性的优化算法提高溯源的精确率;再次,引入近似最近邻搜索和元组过滤优化机制降低溯源时间复杂度;最后,采用有向无环图展示元组数据的溯源关系。实验结果表明,该方法精确率较高、时间复杂度较低、存储消耗较少,能够有效改进元组级数据溯源方法。

基于出力模式匹配的风电集群点多时间尺度功率预测

风电集群点出力预测是提高大规模风电接入条件下电力系统运行安全性和经济性的有效手段。针对风电集群点出力的模式性,利用元组向量表示其变化趋势,通过元组向量时间扭曲法对当前出力模式在历史出力模式库中进行匹配,以时间偏移量、时间压缩率、幅值压缩率3个修正参数表征相似出力模式的差异性,结合匹配结果对风电集群点出力进行预测。实际应用结果表明,该方法可以取得较为理想的效果,参数修正能够提高预测精度,使预测结果更加合理。

Counter-Rotating Type Tidal Stream Power Unit Boarded on Pillar(Performances and Flow Conditions of Tandem Propellers)

The authors have invented the unique counter-rotating type tidal stream power unit composed of the tandem pro- peUers and the double rotational armature type peculiar generator without the traditional stator. The front and the rear propellers counter-drive the inner and the outer armatures of the peculiar generator, respectively. The unit has the fixftful advantages that not only the output is sufficiently higher without supplementary equipment such as a gearbox, but also the rotational moment hardly act on the pillar because the rotational torque of both propel- lers/armatures are counter-balanced in the unit. This paper discusses experimentally the performances of the power unit and the effects of the propeller rotation on the sea surface. The axial force acting on the pillar in- creases naturally with the increase of not only the stream velocity but also the drag of the tandem propellers. Be- sides, the force vertical to the stream also acts on the pillar, which is induced from the Karman vortex street and the dominant frequencies appear owing to the front and the rear propeller rotations. The propeller rotating in close to the sea surface brings the abnormal wave and the amplitude increases as the stream velocity is faster and/or the drag is stronger.