基于遗传算法优化神经网络的直升机参数选择

在直升机初步设计阶段估算其基本参数是很重要的.神经网络的通用性和精度比传统的估算方法有更多的优势,但是在应用神经网络时存在如何选择合适的网络结构和隐层节点数目等一些困难.应用遗传算法优化三层神经网络结构和连接权重,并将优化得到的网络应用于直升机参数选择中.该方法不但可以给出一个最优的神经网络结构和连接权重,而且降低了估算误差,具有及时应用最新数据学习的能力.此外,该方法易于在直升机设计系统中得到应用.     

无线电引信海面及目标多普勒回波仿真分析

频谱识别技术利用海面及目标回波的多普勒频差来滤除海杂波信号。为准确分析弹目交会时引信回波的多普勒频谱,采用了三维弹目交会模型对弹目不共面时引信回波进行模拟,利用等多普勒线划分海面区域进而计算海面回波多普勒频率,利用目标不同散射点合成的方法推导了目标回波多普勒频率的表达式,并在弹目飞行速率、弹目交会角及目标脱靶量等因素不同时对引信回波多普勒频率进行仿真,通过仿真分析得出了以上因素对引信回波多普勒频率的影响,并验证了频谱识别技术在不同交会条件下的有效性。     

STUDY OF OPTIMAL SLIDING-MODE GUIDANCE LAW

An optimal guidance law based on missile-target line-of-sight (LOS) angular rate is presented for intercepting a nonmaneuvering target. It is then integrated with sliding-mode control theory by using reaching-law of sliding-mode, in order to derive an optimal sliding-mode guidance law for intercepting a maneuvering target. The new guidance method's robustness against target maneuvers and good miss distance performance are proved by the second method of Lyapunov and simulation results. The presented guidance law is simple to implement in practical applications.     

粒子群优化的粗糙集-神经网络在航空发动机故障诊断中的应用

提出了一种基于粒子群优化算法的邻域粗糙集-神经网络的发动机智能故障诊断方法,首先利用基于邻域粗糙集模型的属性约简方法对样本数据进行属性约简,然后采用粒子群优化算法替代传统BP算法来训练神经网络的权值和阈值,再用训练好的神经网络对航空发动机气路故障进行诊断.仿真结果表明:该方法降低了神经网络结构的复杂性,减少了网络训练时间,提高了诊断精度.     

Efficient and accurate online estimation algorithm for zero-effort-miss and time-to-go based on data driven method

This paper introduces a novel and efficient algorithm for online estimation of zero-effortmiss and time-to-go based on data driven method.Only missile-target separations are utilized to construct the estimation models, and a practical Fisher fusion algorithm is derived to acquire the estimates with high accuracy and computational efficiency.Further, the two parameters can be online estimated at a particular time.Meanwhile, the kinematics equations of the missile-target engagement are independent, and assumptions of the missile guidance system dynamics and behaviors of the missile and target are completely out of consideration.Moreover, the effectiveness and applicability are explicitly verified through various simulation scenarios.     

失效卫星姿态接管的并行学习合作博弈控制

针对多颗微小卫星合作接管失效卫星姿态运动的问题,研究了考虑微小卫星控制约束的多星合作博弈策略学习与协同控制方法。首先,建立了微小卫星合作博弈模型,给出了能够处理微小卫星控制约束的多星合作博弈帕累托最优策略显式表达式。其次,针对微小卫星合作博弈策略学习需求,通过过去与当前时刻数据的并行使用,设计了基于并行学习的策略迭代方法,该方法放松了神经网络（NN）权值矢量学习对持续激励条件的要求。给出了为确保神经网络权值矢量估值收敛,所使用的过去时刻数据所需满足的条件,并通过Lyapunov方法分析了神经网络权值矢量估计误差的一致最终有界性。之后,采用并行学习策略迭代方法进行了微小卫星合作博弈帕累托最优策略数值解的逼近。所获得的合作博弈策略具有反馈控制形式,在进行神经网络权值矢量学习后,各微小卫星能够通过合作博弈策略的独立计算实现失效卫星姿态运动接管过程中的闭环协同控制。所设计方法避免了传统姿态控制方法所需进行的力矩分配,消除了微小卫星数量对其控制计算复杂度的影响。最后,通过数值仿真对所设计方法的有效性进行了验证。     

空地导弹滑模制导方法研究

针对空地导弹的制导规律设计问题,给出了滑模制导方法,建立了纵向平面内的弹-目相对运动方程。基于滑模控制理论,设计了空地导弹的制导规律,以同时保证脱靶量和弹道倾角要求,并针对固定目标,给出了相对距离和相对速度的解算方法,介绍了针对移动目标的相对距离和相对速度估算方法。仿真结果表明,该制导方法具有较高的制导精度,并且满足弹道倾角约束。     

基于近似技术的高亚声速运输机机翼气动/结构优化设计

探索基于近似技术的高亚声速运输机机翼气动/结构多学科设计优化方法,建立了基于近似技术的多学科设计优化框架。气动学科采用全速势方程加黏性修正进行翼身组合体跨声速流动的气动计算,结构学科采用有限元分析方法进行应力与变形计算。采用均匀设计法给出若干样本点,分别采用二次响应面、Kriging模型和径向基神经网络等多种近似技术,构造气动学科和结构学科的近似分析模型,并对几种近似模型精度进行了分析和比较。研究发现,Kriging模型和二次响应面具有几乎等同的较高的近似精度,神经网络的近似精度则较差,由于二次响应面计算量更小,故最终选定为机翼设计优化的近似方法。以升阻比和结构重量为目标,考虑升力、机翼面积以及应力和应变约束条件,对运输机机翼4个外形参数和4个结构参数进行多目标、多约束优化设计。优化后的机翼具有较好的气动/结构综合性能,表明本文方法是可行的。     

基于神经网络及专家系统的混合诊断法

本文将神经网络与人工智能专家系统结合起来,利用神经网络的联想记忆能力对发生过的故障进行直接记忆和联想。利用专家系统的反向推理能力对联想产生的侯选故障进行验证。据此原理开发了一种更接近人类思维过程的诊断系统。      

Finite time L1 Approach for Missile Overload Requirement Analysis in Terminal Guidance

This article analyzes the problem about the missile overload requirement in a homing terminal guidance under various engagement scenarios. An augmented proportional navigation guidance (APNG) model is introduced on the basis of linear kinematics. To analyze the peak-to-peak performance of the terminal guidance system, a new finite time L₁ performance measure for a linear time-varying (LTV) continuous system is proposed. Then, according to the idea of the adjoint system, a novel method for computing the L₁ norm of a linear continuous system is first derived. Within the finite time L₁ framework, the quantitative relation between the guidance loop dynamics and the maximum missile-target maneuver ratio is offered. This relation is expressed in the form of graphs and formulas that can be used to synthesize some of the major subsystem specifications for the missile guidance system. The illustrative examples show that a significant performance improvement is achieved with the proposed guidance loop dynamics.     

Neural Network aided PMSM multi-objective design and optimization for more-electric aircraft applications

This study uses the Neural Network(NN) technique to optimize design of surfacemounted Permanent Magnet Synchronous Motors(PMSMs) for More-Electric Aircraft(MEA)applications. The key role of NN is to provide dedicated correction factors for the analytical PMSM mass and loss estimation within the entire design space. Based on that, a globally optimal design can be quickly obtained. Matching the analytical estimation with Finite-Element Analysis(FEA) is the main research target of training the NN. ...     

Tracking in clutter with nearest neighbor filters: analysis andperformance

The measurement that is “closest” to the predicted target measurement is known as the “nearest neighbor” (NN) measurement in tracking. A common method currently in wide use for tracking in clutter is the so-called NN filter, which uses only the NN measurement as if it were the true one. The purpose of this work is two fold. First, the following theoretical results are derived: the a priori probabilities of all three data association events (updates with correct measurement, with incorrect measurement, and no update), the probability density functions (pdfs) of the NN measurement conditioned on the association events, and the one-step-ahead prediction of the matrix mean square error (MSE) conditioned on the association events. Secondly, a technique for prediction without recourse to expensive Monte Carlo simulations of the performance of tracking in clutter with the NN filter is presented. It can quantify the dynamic process of tracking divergence as well as the steady-state performance. The technique is a new development along the line of the recently developed general approach to the performance prediction of algorithm with both continuous and discrete uncertainties     

带空间协同的多导弹时间协同制导律

针对多导弹在平面内从期望的弹目视线（LOS）相对方向同时击中固定目标问题,提出了一种带空间协同的多导弹时间协同制导律。基于平面内的导弹-目标相对运动方程,建立了带空间协同的多导弹时间协同制导模型;基于多智能体协同控制理论,在视线方向设计了分布式时间协同制导律,可保证所有导弹的打击时刻在有限时间内达到一致,在视线法向方向设计了分布式空间协同制导律,可保证所有导弹的相对视线角在有限时间内收敛到期望值;最后,通过仿真验证了所设计的协同制导律可使多导弹从期望的弹目视线相对方向同时击中目标。     

俯仰跟踪任务中的驾驶员神经网络模型辨识

 在飞机设计中,应用驾驶员数学模型预测飞机飞行品质是避免人机系统出现不良耦合的重要途径之一。为了提高飞行品质的预测精度,采用人工神经网络(NN)方法进行驾驶员模型辨识,着重研究该模型对不同飞机被控对象的适应能力。首先,详细分析了驾驶员完成俯仰跟踪任务的操纵行为特点,提出适用于该驾驶员行为描述的神经网络模型结构形式。然后,根据对不同被控对象进行俯仰跟踪实时仿真实验的结果,对神经网络模型参数进行识别。最后,对模型辨识结果进行了精度评价。研究结果表明,该辨识方法适用于研究具有不同增益、不同短周期振荡频率飞机被控对象的驾驶员操纵行为特性。     

基于神经网络的无线通信系统RF电路设计方法

典型的无线通信系统RF电路由大量的线性和非线性元器件组成。建立这些微波电路元器件的模型在现代RF电路的设计中至关重要。采用一种神经网络来学习非线性电路的动态特性,并建立相应的模型。神经网络模型可以对原始电路的模拟特性进行整体预测,可用于高层电路的仿真和优化。以功率放大器为例验证了神经网络模型的实用性,表明神经网络模型可以极大地提高无线通信系统RF电路的建模精度和速度。     

弹载合成孔径雷达自适应重频分析与设计

弹载合成孔径雷达(Synthetic Aperture Radar,SAR)的目标距离、视线角由于高速逼近目标而快速变化,这导致传统的固定脉冲重复频率(Pulse Repetition Frequency,PRF)(简称重频)波形难以兼顾弹载SAR雷达在成像各方面的约束条件,故需要根据当前弹体运动和弹目关系变化情况实时计算重频。详细分析了影响重频选择的各项因素,包括避免距离模糊、方位模糊、高度杂波、发射遮挡影响及SAR成像分辨率、系统相参性要求等影响因素,并设计了自适应重频计算的工作流程。某SAR雷达系统实验表明,该设计能够在实际飞行弹道条件下根据实际弹目关系自适应调整脉冲重复频率,从而更好地实现SAR雷达系统的工作性能,有效解决了固定重频波形不能适应弹载SAR工作条件的难题。     

Application of sliding-mode control to air-air interception problem

A sliding-mode control (SMC) law is applied to a nonlinear system representing an air-air missile-target interception process. Promising results are obtained for a simple switching surface based on proportional navigation (PN). Partial robustness of the controller is demonstrated, i.e. the control is robust only with respect to uncertainties present in the control dynamics. This is illustrated by a breakdown in the control actuator for which SMC was found to be superior to PN     

基于改进 AHP-EWM的导弹装备保障能力模糊综合评估

针对导弹装备保障活动技术复杂、保障能力评估难度大的问题,提出将层次分析法(Analytic Hierarchy Pro. cess,AHP)与改进的熵权法(Entropy Weight Method,EWM)结合,修正主观赋权导致的可信度较低的实际情况,综合确定各要素权重,运用模糊综合评价法评定出各保障要素优劣等级及导弹装备保障能力水平。通过实例分析与评估,得出了该单位导弹装备保障能力与实际相符合的结论,验证了该方法的有效性,并指出了该单位保障能力短板及提升建议。     

基于自适应模拟退火遗传算法的月球软着陆轨道优化

 将自适应遗传算法与模拟退火算法相结合,形成一种自适应模拟退火遗传算法。该算法不但具备了自适应遗传算法的强大全局搜索能力,也拥有模拟退火算法的强大局部搜索能力。针对月球软着陆轨道优化的特点,利用一种新的参数化方法将轨道优化问题转换为非线性规划问题,并应用提出的自适应模拟退火遗传算法进行优化。数值结果表明：该算法的收敛速度快,优化精度高,且避免了初值敏感、病态梯度和局部收敛等问题,能够搜索到全局最优轨道。     

Critique of some neural network architectures and claims forcontrol and estimation

While there is great potential for successful use of neural network (NN) algorithms in automatic target recognition (ATR) and other pattern identification/classification applications, significant barriers have been encountered that, to date, defy rigorous use of NNs within feedback control designs. The status of several problems and contradictions involving NNs relating to control and estimation theory applications (and to practical failure detection within INS/GPS navigation systems) are summarized here. To give a positive spin and for a balanced perspective, we also mention many novel laudable NN results obtained by invoking the techniques and results of control and estimation theory