提高旋转式电场仪测量电场特性的方法研究

在介绍旋转式电场仪测量电场强度的机理及测量系统构成的基础上,针对在特高压直流输电线路下地面合成电场测量中,各电场仪的输出电压与被测电场强度的特性曲线不一致,外场较恶劣测量环境对测量结果造成干扰影响测量准确度等问题,提出了相应的解决方法（特性曲线归一化处理和双路消噪差动测量法）。所述方法便于用户现场使用,提高了旋转式电场仪抗干扰能力和测量准确度。     

Configurationmodelofpartial repairablesparesunder batch ordering policy based on inventory state

Rational planning of spares configuration project is an effective approach to improve equipment availability as well as reduce life cycle cost （LCC）. With an analysis of various impacts on support system, the spares demand rate forecast model is constructed. According to systemic analysis method, spares support effectiveness evaluation indicators system is built, and then, initial spares configuration and optimization method is researched. To the issue of discarding and con-sumption for incomplete repairable items, its expected backorders function is approximated by Laplace demand distribution. Combining the （s-1, s） and （R, Q） inventory policy, the spares resup-ply model is established under the batch ordering policy based on inventory state, and the optimi-zation analysis flow for spares configuration is proposed. Through application on shipborne equipment spares configuration, the given scenarios are analyzed under two constraint targets：one is the support effectiveness, and the other is the spares cost. Analysis reveals that the result is consistent with practical regulation;therefore, the model＇s correctness, method＇s validity as well as optimization project＇s rationality are proved to a certain extent.     

遥感场景分类的批处理协方差神经网络方法

针对卫星遥感图像场景分类数据集中存在的局部区域特征异常问题，提出一种采用批处理协方差层的神经网络(CovNN)模型进行遥感场景分类的方法。该方法通过计算全输入通道的局部区域均值实现一种3D批处理协方差算法，能够有效消除局部区域均值的影响，从而更好地处理局部光照过强和局部区域存在无关特征的问题。将其应用于存在局部光照异常和局部无关特征问题的卫星采集AID数据集和NWPU RESISC45数据集中，实验表明CovNN在两个数据集上均取得了超过现有卷积神经网络(CNN)的召回率，可有效降低图像局部区域特征异常的不利影响。