ID	原文	译文
8224	利用一元线性回归与残差分析,快速、自动地选出聚类中心点,去除了人为选择的主观性。	By using monadic linear regression and residual analysis, quickly and automatically choose the clustering center, in addition to the subjectivity of human choice.
8225	通过理论分析以及在人工数据集和真实数据集的对比实验表明,提出的基于线性回归分析的快速搜索聚类中心算法能够克服原有算法的缺陷,并且在聚类效果和计算时间上优于原有算法、基于密度的带有噪声的空间聚类算法(density based spatial clustering of applications with noise,DBSCAN)以及K-means算法。	Through theoretical analysis and the experiments of artificial data and real data sets show that the clustering center based on linear regression analysis of fast search algorithm can overcome the defects of original algorithm, and on the clustering effect and the computation time is better than the original algorithm, with noise of spatial clustering algorithm based on density (density -based spatial clustering of applications with noise, DBSCAN) and K - means algorithm.
8226	针对同时受自然退化和随机冲击影响的单部件系统可靠性评估问题,首先,建立了退化型与突发型竞争失效过程的可靠性模型,该模型考虑了冲击对退化的影响,即冲击会增加系统的退化量;	Also affected by the natural degradation and stochastic shocks of single component system reliability evaluation problem, first of all, the degradation model is established with incident type competition failure process reliability model, the model takes into account the impact effect on the degradation, the impact will increase the degradation of system;
8227	其次,提出了基于状态的维修策略,构建了一个成本率函数;	Secondly, put forward the maintenance strategy based on state, built a cost function;
8228	最后,以一个由4个零件串联而成的单部件系统为研究对象,确定了该系统的可靠度函数和最优的预防更换阈值。	Finally, in a single parts composed of four parts in series system as the research object, to determine the reliability of the system function and the optimal preventive replacement threshold.
8229	该结果不仅验证了可靠性模型的合理性和有效性,而且体现了基于状态的维修策略的可行性与经济性。	The results not only verify the rationality and effectiveness of the reliability model, and embodies the maintenance strategy based on state the feasibility and economy.
8230	作为解决序贯决策的机器学习方法,强化学习采取持续的"交互-试错"机制,实现智能体(Agent)与环境的不断交互,从而学得完成任务的最优策略,契合了人类提升智能的行为决策方式。	As solving sequential decision making method of machine learning, reinforcement learning to continue "interaction - trial and error" mechanism, implementing Agent (Agent) and the environment interaction, so as to learn the optimal strategies to complete the task, the way the human ascension intelligent behavior decision-making.
8231	知识作为一种包含了经验、价值观、认知规律以及专家见解等要素的结构化信息,应用于强化学习可以有效提高Agent的学习效率,降低学习难度。	Knowledge as a kind of contains experience, values, cognitive law and structured information expert opinions and so on elements, used in reinforcement learning can effectively improve the learning efficiency of Agent, reduce the difficulty of learning.
8232	鉴于此,本文以强化学习的基本理论为起点,对深度强化学习以及基于知识的深度强化学习研究成果进行了系统性的总结与梳理。	In view of this, this paper takes the basic theory of reinforcement learning as a starting point, the depth of the depth of the reinforcement learning and knowledge based reinforcement learning research has carried on the systematic summary and comb.
8233	针对新息自适应滤波算法噪声跟踪精度和跟踪灵敏度相互矛盾导致的窗口宽度选取困难问题,提出了一种基于滑动窗口的新息自适应组合导航算法。	For new rate adaptive noise filtering algorithm tracking accuracy and tracking sensitivity caused by conflicting window width is difficult problem, this paper proposes a new rate adaptive integrated navigation based on sliding window algorithm.