ID 原文 译文
5834 其次,采取基于正切函数的惯性权重调整结合线性调整学习因子的方法来提高粒子群优化算法的全局搜索能力,同时,在适应度函数中加入转艏角控制来提高路径的平滑性; Secondly, based on tangent function of inertia weight adjustment combined with linear learning factor method to improve the global search ability of particle swarm optimization algorithm, at the same time, joining in the fitness function to turn bow Angle control to improve the smoothness of the path;
5835 最后,利用改进后的粒子群优化算法规划出每个综合视域内的路径。 Finally, using the improved particle swarm optimization algorithm is used to plan a path of each integrated horizon.
5836 仿真实验结果表明,该优化算法能实现USV集群的实时避碰,并快速为USV集群规划出平滑、安全的全局最优无避碰路径。 Simulation experiment results show that the optimized algorithm can realize real-time collision avoidance of USV clusters, and quickly for USV cluster planning global optimal smoothing, security is no collision avoidance path.
5837 针对传统合成孔径雷达(synthetic aperture radar,SAR)图像舰船目标检测算法检测精度易受斑点噪声影响,且只能提取底层特征及其泛化性较差的问题,提出了一种基于深度卷积神经网络的SAR图像舰船目标检测算法。 In view of the traditional synthetic aperture radar (synthetic aperture radar, SAR) image of ship target detection algorithm accuracy is affected by speckle noise, and can extract the underlying characteristics of the poor and its generalization, this paper proposes a convolutional neural network based on depth of SAR images of ship target detection algorithm.
5838 首先将目前先进的单次多盒检测器(single shot multibox detector,SSD)检测算法应用到SAR图像舰船目标检测领域,指出了其在该领域存在的局限性。 Will first at present, a lot of advanced single box detector (single shot multibox detector, SSD) detection algorithm is applied to the SAR images of ship target detection field, points out its limitations.
5839 在此基础上提出了基于SSD的新的检测方法,包括融合上下文信息,迁移模型学习,在公开的SSDD数据集上进行了训练和测试,对实验结果进行了对比分析。 In the field on the basis of this puts forward new detection method based on SSD, context information, including fusion study migration model, in open SSDD data set on the training and testing, has carried on the comparative analysis with the result of the experiment.
5840 实验结果表明,相比于原始的SSD检测算法,所提出的方法不仅提高了目标检测精度,同时也保证了算法的检测效率。 The experimental results show that compared with the original SSD detection algorithm, the proposed method not only improved the precision of target detection, but also ensure the detection efficiency of the algorithm.
5841 针对复杂电磁环境下利用人工提取特征识别雷达信号存在的主观性强、特征冗余的问题,提出了一种基于深层卷积神经网络的识别方法。 For using artificial extraction under complex electromagnetic environment existed in the recognition of the radar signal characteristics characteristics of strong subjectivity, redundancy problem, proposes a recognition method based on deep convolution neural network.
5842 该方法首先提取雷达信号的双谱信息作为深层卷积神经网络模型的输入,然后利用模型的自学习能力提取深层特征,实现对不同调制样式雷达信号的识别,最后对不同结构网络模型的识别结果进行对比。 The double spectrum information of radar signal is firstly extracted as deep convolution the input of the neural network model, and then deep self learning ability of the model is utilized to extract features, implementation of different styles of radar signal modulation recognition, finally comparing the identification results of different network structure model.
5843 仿真实验结果表明,相比传统雷达信号识别方法,该方法对于不同调制类型信号的识别效果优异,并且在识别率、抗噪性上都有所提升。 Simulation experiment results show that compared with the traditional radar signal recognition method, the method to the identification of different types of modulation signal effect is excellent, and the recognition rate and improve the noise resistance.