基于递归模糊神经网络的鲁棒BP算法研究

讨论了一种基于递归模糊神经网络的鲁棒BP算法的实现。该算法采用极大似然法修正传统BP算法的指标函数 ,利用M估计器和自举算法在线估计样本的误差分布 ,适合于动态非线性系统的辨识 ,逼近精度高 ,收敛速度快 ,在小噪声扰动时处于稳定 ,对过失误差有很强的鲁棒性。仿真结果证明了该算法的有效性     

Simulator of pyroshock environment and effect rules of its adjustable parameters

During the launching of spacecraft, the on-board devices will undergo a series of pyroshock environments. In order to verify the reliability of these devices under these pyroshock environments, all of them are needed to take the shock test before launching. This paper has carried out an in-depth research on the simulation method of the pyroshock based on the true explosive excitation. In this study, a simulator containing multiple adjustment parameters is presented and the safety is considered by the design of the protective cover. And the working process of this setup is simulated with the explicit dynamic codes LS-DYNA. What’s more, the effects of the adjustment parameters on the three factors of shock Response Spectrum (SRS) of the resonant board are explored carefully. The rules achieved in this paper are verified by a typical example. The results indicate that the improved simulator can avoid the danger of explosive and make full use of the advantage of actual explosive excitation. And the test condition can be quickly realized at the simulator according to the effect rules of the three adjustable parameters.     

基于ACMPE、ISSL-Isomap和GWO-SVM的行星齿轮箱故障诊断

针对从行星齿轮箱非线性、非平稳振动信号特征提取困难的问题，提出了一种基于自适应复合多尺度排列熵(ACMPE)、改进监督型自组织增量学习神经网络界标点等度规映射(ISSL-Isomap)和灰狼群优化支持向量机(GWO-SVM)相结合的行星齿轮箱故障诊断方法。利用ACMPE从复杂域提取振动信号的故障特征，构建高维故障特征集；采用ISSL-Isomap方法对高维故障特征集进行维数约简，提取出低维、敏感故障特征；应用GWO -SVM分类器对低维故障特征进行模式识别，判断故障类型。行星齿轮箱故障诊断实验结果分析表明:与多尺度排列熵(MPE)、复合多尺度排列熵(CMPE)等特征提取方法相比，ACMPE方法在分类效果和识别精度上更具优势；与局部切空间排列(LTSA)、等度规映射(Isomap)、加权Isomap(W-Isomap)、监督Isomap(S-Isomap)和监督型自组织增量学习神经网络界标点Isomap(SSL-Isomap)等降维方法进行比较，ISSL-Isomap方法降维效果最佳；所提方法的故障识别率达到100%，具有一定优越性。      

立方星电源系统最大功率点跟踪优化控制方法

针对立方星在能量来源严重受限条件下如何提高太阳能利用率的难题，提出一种适用于立方星的集中供电式空间微电源架构(EPS)，并设计基于改进粒子群优化算法的最大功率点跟踪(MPPT)控制策略来提升能量转换效率。首先，推导太阳电池阵列的数学模型，并根据太阳电池阵列的工作特性，提出电源系统最大功率点跟踪控制的物理系统实现结构。其次，设计基于改进粒子群优化(PSO)的最大功率点跟踪控制算法，并进行了数学仿真校验。最后，对所设计的电源系统架构进行了硬件实现和试验验证。地面试验结果表明，电源系统的太阳能最大转换效率可达95.5%。该电源系统成功应用于世界首颗12U立方星“翱翔之星”的飞行试验，在轨数据表明电源系统工作状态良好，为微纳卫星电源系统的设计提供了有益参考。     

基于MED和LMD的自动倾斜器轴承广义Shannon熵谱分析

针对轴承信号微弱故障特征易被强背景噪声淹没的问题,提出采用最小熵反褶积,通过逆滤波器最优化设计,对目标信号进行降噪处理,其峭度值提高了约3.8倍,增强了信号的微弱故障特征;针对非平稳非线性信号频率成分复杂难以解调的问题,提出采用局部均值分解(LMD)和峭度-相关系数筛选准则,其可对非平稳非线性信号进行自适应分解和最优重构,提高了信号的信噪比;针对信号耦合调制及边频突出的问题,通过引入广义Shannon熵进行包络谱带内降噪处理,信号一阶故障特征调制频率与故障特征频率的幅度比降低了24%~43%。通过实验室信号及某型直升机自动倾斜器轴承故障诊断地面试验的分析结果验证了该方法的合理性和可行性。      

Hybrid partition- and network-level scheduling design for distributed integrated modular avionics systems

Distributed Integrated Modular Avionics (DIMA) develops from Integrated Modular Avionics (IMA) and realizes distributed integration of multiple sub-function areas. Time-triggered network provides effective support for time synchronization and information coordination in DIMA systems. However, inconsistency between processing resources and communication network destroys the time determinism benefiting from partitions and time-triggered mechanism. To ensure such time determinism and achieve guaranteed real-time performance, system design should collectively provide a global communication scheme for messages in network domain and a corresponding execution scheme for partitions in processing domain. This paper firstly establishes a general DIMA model which coordinates partitioned processing and time-triggered communication, and then proposes a hybrid scheduling algorithm using Mixed Integer Programming to produce feasible system schemes. Furthermore, incrementally integrating new functions causes upgrades or reconfigurations of DIMA systems and will generate integration cost. To control such cost, this paper further develops an optimization algorithm based on Maximum Satisfiability Problem and guarantees that the scheduling design for upgraded DIMA systems inherit their original schemes as much as possible. Finally, two typical cases, including a simple fully connected DIMA system case and an industrial DIMA system case, are constructed to illustrate our DIMA model and validate the effectiveness of our hybrid scheduling algorithms.     

Guidepost-based autonomous orbit determination method for GEO satellite

Guidepost-based navigation system is a novel autonomous orbit determination method for the GEO satellite. The system is achieved by using the camera imaging function to obtain the guidepost images and the GNSS signal receiver to obtain the pseudoranges between the GEO and the navigation satellites. Due to the high altitude of GEO satellite and the time-varying sunlight condition in the space environment, it may be difficult to obtain object image points and the distance measurements of GNSS because of the weak visibility of the guideposts. To deal with the problem, a novel integrated orbit determination system is presented. The Earth landmarks, the in-orbit spacecraft and GNSS navigation satellites whose line-of-sights and the distance can be easily obtained are used at the same time as information for the GEO satellite navigation based on the observability conditions analysis. The observability of the GEO satellite navigation system is analyzed through the physical observability, the mathematical observability and the engineering observability through the observing geometry, the rank of observability matrix and the Cramer-Rao lower bound (CRLB) respectively. Besides, the maximum correntropy unscented Kalman filter (MCUKF) algorithm is applied to improve the estimation stability of the system in the presence of non-Gaussian noises. The simulation indicates the feasibility of the proposed scheme.     

面向再制造设计方案的去主观混合多属性决策方法

面向再制造设计(Design for remanufacturing,DfRem)是将再制造特征考虑到产品设计过程,因设计者缺乏明确定义的再制造知识,面向再制造方案决策受设计者主观偏好、经验等主观因素的影响,导致不合理决策设计方案。为了客观地得到面向再制造最优设计方案,提出一种去主观的混合多属性决策方法克服设计过程主观因素的影响。首先,通过物元理论表征设计特征与需求信息,结合设计准则得到初步设计方案,接着建立考虑技术、经济和环境因素的评价指标体系,采用熵权法和模糊集通过多属性决策得到最优设计方案。最后,以面向再制造的电机拆解机设计为案例验证上述方法的可行性与实践性,结果表明上述方法对面向再制造设计的方案决策有效可行。     

复杂微通道内对流传热的场协同及熵产

利用场协同和熵产原理研究了针肋宽度、凹穴宽度及雷诺数(Re)对凹穴和针肋组合式微通道内对流传热特性的影响,分析了微结构强化传热的本质原因,并对微通道的综合性能进行了评价。结果表明,增大针肋和凹穴宽度能够显著减小传热协同角,提高流场和温度场的协同程度,有利于强化对流传热,但局部漩涡会使流动协同角减小,增大微通道压降;增大针肋宽度能够提高能量利用效率,从而强化传热,但同时导致流动熵产率增大;适当增大凹穴宽度能够减小传热熵产率,但凹穴宽度过大会导致传热不可逆性和流动摩擦均增大;综合考虑泵功、相对针肋宽度和相对凹穴宽度,提出了预测热阻的经验关联式;当相对针肋宽度为0.2,相对凹穴宽度为2时,微通道的热阻最小,综合性能最好。      

基于贝叶斯—熵权法的岸舰导弹作战效能研究

通过系统分析岸舰导弹作战流程,确定了影响岸舰导弹作战效能的 4个主要因素。选取了 17个指标,构建了岸舰导弹作战效能评估模型,运用贝叶斯网络原理以及熵权法对模型进行仿真,并针对仿真结果,提出岸舰导弹战斗使用和装备发展对策建议。