
Aiming at the problems of unreachable search target, weak path finding and obstacle avoidance ability, and slow algorithm convergence speed when dealing with the 3D path planning of autonomous underwater vehicles (AUV) in the traditional cuckoo algorithm in complex waters, an AUV path planning algorithm PSO-ASCS (Particle Swarm Optimization-Adaptive Step-size Cuckoo Search Algorithm) is proposed, which combines the improved Adaptive Step-size Cuckoo Search and Particle Swarm Optimization. This research uses the idea of spatial layering to establish a three-dimensional model of complex waters to conduct path planning and obstacle avoidance experiments on the PSO-ASCS algorithm; The PSO-ASCS algorithm is tested and compared with the adaptive step size cuckoo search algorithm, standard cuckoo algorithm and particle swarm optimization by constructing a fitness function considering the three factors of path length, path smoothness and path hazard. Experiments show that the improved algorithm has strong global search ability and optimization performance, and the algorithm converges well, so that the AUV has the ability of efficient obstacle avoidance and path planning.
在传统的图像描述生成任务中,已有方法对图像的描述仅仅停留在浅层,并缺乏真实世界知识的指导,难以挖掘出对象在特定背景下的逻辑语义关系。新闻文本的引入为图像描述带来了新的可能,同时对模型的学习能力有了更高要求;此外,新闻图集中往往存在多幅图像,且相互之间联系紧密,导致现有单图描述生成方法不适用于新闻图集描述生成。针对上述问题,本文提出了一种基于图文双向引导注意力(image and text bidirectional guidance attention,ITBGA)的新闻图集描述方法,以图集作为研究对象,并辅以对应的新闻文本作为背景知识,基于ITBGA分别实现粗、细两个粒度的跨模态信息交互,并通过指针网络辅助命名实体词生成。在本文构建的新闻图集数据集上进行了实验验证,结果表明ITBGA能有效提升描述文本的质量,在关键的CIDEr指标上达到了最优。
Aiming at the problems of weak keys,low security and many loopholes in the lightweight cryptography and authentica-tion protocols used in most authentication models of the Internet of Things,a multi-factor enhanced authentication model was pro-posed based on zero-knowledge proof.The model was composed of one-way one-time computing type tasks,and combines vari-ous consensus verification technologies to form a multi-factor consensus verification method.And through the distributed consen-sus mechanism of the blockchain architecture,the enhanced processing function of multi-factor one-time difficult tasks had been re-alized.At the same time,flexible access control and privacy protection were realized by dynamically setting the type of evidence in the token.Finally,a simulation experiment was carried out to verify,and a prototype scene was constructed for the offline smart lock scene.Experimental results show that the model implements concise peer-to-peer verification of dynamic difficulty policies,exhibiting millisecond-level high performance and flexibility.Compared with other IoT authentication models,it has better securi-ty and practicality.
Aiming at the problem that the whole network topology information of the new power system Internet of Things is man aged in a decentralized manner by taking power plants and stations(including power plants,substations,renewable energy power plants,virtual power plant,etc).Firstly,a topology analysis method for power plants based on switch path function was proposed.Secondly,the rules for decentralized coordination of plant and station information were proposed,and an improved switch path function method for adjacent plant and station topology analysis was proposed,as well as a method for achieving full network topology analysis through plant and station decentralization.Finally,the feasibility of the proposed method was demonstrated through simulation of an actual new power system.
To address the vulnerability of the CAN(controller area network)bus in Telematics,this paper proposes a hybrid attack intrusion detection model,DGAO-IDS,which uses an unsupervised DBN to learn the basic features of normal CAN message data and a GRU network with an attention mechanism to learn its temporal features,and finally classifies them using a single classification support vector machine.The final detection result is obtained by combining the filter with the classification results of the aforementioned model.Experimental results demonstrate that the proposed rule-based filters have a false positive rate of zero for different attacks.Moreover,the DGAO-IDS model outperforms the comparative models HyDL-IDS and MD-LSTM on the public dataset HCRL.In addition,the model achieves an accuracy of 91.05%on the mixed-attack dataset MixAt,which is 6.55%and 7.93%higher than the HyDL-IDS model and the MD-LSTM model,respectively.