基于深度学习的区块链蜜罐陷阱合约检测

针对当前检测方法准确率不高以及模型泛化性较差的问题,提出了基于KOLSTM深度学习模型的蜜罐陷阱合约检测方法。首先,通过分析蜜罐陷阱合约的特点,提出了关键操作码的概念,并设计了可用于选取智能合约中关键操作码的关键词提取方法;其次,在传统的LSTM模型中加入关键操作码权重机制,构建了可以同时捕获蜜罐陷阱合约中隐藏的序列特征以及关键操作码特征的KOLSTM模型。最后,通过实验表明,该模型具有较高的识别精确率,在二分类和多分类检测场景下的F值较LightGBM模型分别提升2.39%与19.54%。

Experimental investigation and modeling of flotation column for treatment of oily wastewater

A unique cyclonic static microbubble flotation column was developed for oily wastewater separation.The separation effciency was found to be highly dependent on gas holdup and bubble size distribution.By changing the circulation pressure,gas flow rate,frother concentration,the effect of operation parameters on gas holdup and oil removal effciency were attained.A mathematical modeling between the kinetic constant and the gas holdup was established for oily water separation process.The results show that higher gas holdup and smaller microbubble sizes are benefcial to improve oil removal effciency.

烧结Nd-Fe-B磁体的晶界精细结构研究进展

烧结Nd-Fe-B磁体的矫顽力不仅与其各向异性场(HA)有关,同时严重依赖磁体的微观组织结构。只有几纳米厚的富Nd晶界薄层,以及存在于晶界角隅处的尺度较大的富Nd晶界相,对Nd-Fe-B磁体的矫顽力有重要影响。长期以来,Nd-Fe-B磁体中晶界相被含糊地称为“富钕相”。首先简要回顾了相关领域早期的研究结果,包括:烧结钕铁硼磁体的相组成、微结构特点及其与磁体矫顽力的关系;然后结合最新的研究进展,重点对磁体的晶界相组成和晶界成分、晶界磁性质、添加元素的影响、以及回火以后晶界显微结构的演变和微磁学模拟结果进行综合分析。最后,结合Nd-Fe-B相图研究结果,对目前研究中存在的问题进行探讨,指出今后需要努力的方向。