基于Q-Learning的深度神经网络自适应退避策略

针对无人机自组织网络,结合Q-Learning和深度神经网络,提出一种自适应退避策略,以提高基于竞争的MAC协议通信性能.以Matlab为仿真平台,仿真比较了自适应退避策略与二进制指数退避策略的性能.     

船舶混合动力系统与智能能量管理控制策略研究进展

开发新型动力系统和智能控制策略是解决节能减排的重要技术方向。本文首先对混合动力系统的配置和拓扑结构进行归类与比较，对控制策略中涉及的智能算法进行归纳和总结；其次，对新型动力船舶在动力系统和控制算法使用情况进行阐释；最后，从不同角度对混合动力系统以及能量管理系统的发展趋势进行展望，为绿色船舶混合动力系统开发以及能量管理系统的研发提供了参考依据。     

基于多目标决策的 对岸火力支援火力分配模型

在建立对岸火力支援作战中火力分配思维机制的基础上,根据最优火力分配的基本要素,构建了基于多目标决策分析的火力分配模型。通过改进后遗传算法对模型进行仿真求解,结果验证了多目标决策分析方法在对岸火力分配问题中的适用性,运用改进后的遗传算法对火力分配模型求解具有明显的优势。     

A hybrid chemical modification strategy for monocrystalline silicon micro-grinding: Experimental investigation and synergistic mechanism

For high performance manufacturing of micro parts and features, a hybrid chemical modification strategy is proposed to decrease critical energy barrier of mechanical removal of hard and brittle crystal material by refining localized machining condition. The strategy, namely UV-light and IR-laser hybrid chemical modification (UVIR-CM) strategy, includes two steps, an ultraviolet light (UV-light) catalytic advanced oxidation and an infrared laser (IR-laser) assisted selective modification based on Fenton liquid–solid reaction for monocrystalline silicon. The modification effects of UVIR-CM strategy were investigated by surface morphology micro-observation, cross-section transmission electron microscopy (TEM) observation, Raman spectroscopy analysis and nanoindentation test. Experimental results demonstrated that varied degrees of laser texturing appeared on different strategy samples’ IR-laser scanned area. And the IR-laser thermal damage has been successfully inhibited due to the refraction and reflection of energy by bubbles in liquid medium. But for the UVIR-CM strategy, a uniform and amorphous silicate layer is detected in a certain boundary. The UV-light promotes oxidation cycle ability of the chemical solution and ensures sufficient oxide modified layer for subsequent step. Attributing to synergism of photochemical, photothermal and kinetic effects induced by IR-laser, the modified layer displays layered structure with about 600 nm thickness, (2.7 ± 0.60) GPa nanohardness, and (93.7 ± 22.9) GPa indentation modulus. And the layered structure is amorphous layer, nanocrystal and micro-twins layer from the surface to the interior of sample. Consequently, it reveals that the subsequent mechanical removal will become easy due to decreasing energy barrier of monocrystalline silicon in selective area. Meanwhile, its original excellent mechanical properties also are maintained under a certain depth. The results contribute to develop a novel combined micro-machining technology to achieve collaborative manufacturing of structure shape and surface integrity for micro parts and feature.     

Transition control of a tail-sitter unmanned aerial vehicle with L1 neural network adaptive control

The main task of this work is to design a control system for a small tail-sitter Unmanned Aerial Vehicle(UAV) during the transition process. Although reasonable control performance can be obtained through a well-tuned single PID or cascade PID control architecture under nominal conditions, large or fast time-varying disturbances and a wide range of changes in the equilibrium point bring nonlinear characteristics to the transition control during the transition process, which leads to control prec...     

基于低轨全覆盖卫星星座的反向缝建链策略

低轨卫星星座的拓扑设计是天地一体化网络部署中较为重要的一环。在低轨卫星网络中，如何设计星座拓扑来提升卫星网络的稳定性与传输性能，是亟待解决的问题。提出一种基于轨道覆盖带方法的低轨持续全覆盖卫星星座设计方法，通过该方法构建低轨卫星星座，满足覆盖性能，可有效减少卫星切换的次数。基于构建的星座拓扑，制定邻轨k最近邻星的反向缝建链策略，优化设计链路长度和链路持续时长，获取网络传输性能与稳定性的平衡。仿真结果表明:通过上述方法设计出的星座，在满足对地持续全覆盖性能的情况下，相较于轨道覆盖带方法切换次数减少10%;邻轨k最近邻星策略相较于最短建链策略，拓扑稳定性能提升79.62%，使网络拓扑稳定性与传输性能达到更优的平衡。     

基于多智能体航空公司航班恢复协同决策方法

航空公司进行航班延误恢复时，各种资源之间会通过航班计划产生间接关联，此时各决策单元若独立地考虑本领域内的资源恢复问题，将难以保证恢复方案的整体可行性和全局优化性。为探究航空公司航班恢复过程中各决策部门的决策模型及其协同关系，本文提出了基于多智能体技术的航班恢复协同决策仿真方法。首先，基于航空公司实际组织架构构建了航班恢复多智能体决策系统框架；其次，对部门间协同决策的动态过程进行了分析，将延误恢复的全过程分为了预恢复、可行解协商、均衡解协商3个阶段，构建了三阶段协同决策机制；最后，根据不同资源的恢复特性建立各决策部门的核心决策模型与部门间自动协商模型，并基于多智能体系统进行仿真。仿真结果显示，基于多智能体的协同决策方法能够在3.8 s的极短时间内针对1天中包含3架飞机和12个航班的航班计划做出完整的延误恢复方案，并且能够在保障航空公司整体效益的情况下一定程度地平衡各决策主体的自身利益。     

大气环境监测卫星地方时控制策略优化

针对大气环境监测卫星地方时的控制和优化问题，从太阳同步轨道卫星的地方时漂移规律入手，建立了地方时漂移估计模型，给出了通过改变倾角来调节地方时漂移的控制策略。在此基础上，采用遗传算法对控制策略进行了优化，讨论了相关算子的选择，最终得到了燃料有限情况下的最优控制曲线。优化结果表明:不限燃料情况下，大气环境监测卫星可通过寿命内一次倾角控制实现小于6.6 min的地方时偏差。此外，为了保证地方时偏差小于15 min的指标要求，卫星至少需要进行一次0.07°的倾角控制。     

基于支持向量机的抗干扰决策技术

为了应对日益复杂的电磁环境，认知雷达的概念被引入电子战中。认知电子战系统包括认知侦察、认知对抗和认知效能评估3个重要环节。针对认知电子战系统对抗干扰决策技术的需求，为实现自适应的决策抗干扰措施，将支持向量机(SVM)算法运用于抗干扰决策中。仿真结果表明:基于SVM的抗干扰决策技术可以自适应地决策抗干扰措施，进行干扰对抗，该算法完成了认知电子战中的最后的一步，使认知电子战系统实现闭环。     

A novel adaptive lateral reentry guidance algorithm with complex distributed no-fly zones constraints

Aimed at complex distributed no-fly zones avoidance problems, a novel adaptive lateral reentry guidance algorithm is proposed. Firstly, by introducing the improved attractive and repulsive potential fields, an improved artificial potential field method is developed. Combined with the proposed judgment criterion for whether a no-fly zone has been avoided, the proposed improved artificial potential field method effectively solves the reference heading angle determination problem under the constrai...