基于中心损失的条件生成式对抗网络的冷水机组故障诊断

针对冷水机组产生的故障数据不足，数据集中正常数据和故障数据数量不平衡，进而导致故障诊断精度下降的问题，提出一种基于中心损失的条件生成式对抗网络(central loss conditional generative adversarial network，CLCGAN)和支持向量机(support vector machine，SVM)的故障诊断方法。首先，CLCGAN利用少量真实故障数据生成新的故障数据；然后，将生成的故障数据与初始数据集混合，使正常数据与故障数据的数量达到平衡；最后，利用平衡数据集构建SVM模型进行故障诊断。在GAN生成冷水机组故障数据时，构建动态中心损失项并加入到目标函数中，利用动态的中心损失减少冷水机组生成的各种故障数据的类内距离，从而降低各个故障生成数据之间的重叠程度，增加生成数据的可靠性。在生成故障数据之前配置相应的故障标签，并输入到CLCGAN中指导数据生成过程，使生成的故障数据可以均衡地分布于各个故障类别。在ASHRAE 1043-RP数据集上对所提方法进行了验证，结果表明，相较于其他解决数据不平衡问题的故障诊断方法，所提方法具有更高的故障诊断准确率。

Advances in engineering electroactive biofilms by synthetic biology approaches

Electroactive biofilm is a conductive polymer formed by the aggregation of extracellular polysaccharides, proteins, extracellular DNA (etracellular DNA, eDNA), fimbria and other components secreted by electrical energy cells and cross-linking with the cells themselves. Biofilms, which are different from single cells in the form of population aggregation, and play a crucial role in electrocatalytic systems including microbial fuel cells, microbial bioelectrosynthesis for the production of value-added chemicals, metal wastes treatment, and biomedicine. In the natural state, the thickness of electroactive biofilm is relatively thin, the cell quantity is small and the structure is not stable. In this review, we summarize the progress of the research on the modification of electroactive biofilm by synthetic biology in the past five years. We systematically describe the construction control of biofilm, the synthesis of structural components and the transformation of electrical conductivity, so as to realize the electroactive biofilm with high efficiency of electron transfer, and lay a foundation for the further realization of high efficiency of electrocatalysis in the future.