多核贝叶斯优化的模型决策树算法

构造模型决策树时超参数较多,参数组合复杂,利用网格搜索等调参方法将会消耗大量的时间,影响模型性能的提升。提出了一种多核贝叶斯优化的模型决策树算法,该算法为应对不同分类数据特性,采用三种高斯过程建模寻优,利用贝叶斯优化技术,选出最优的参数组合。实验结果表明,所提算法在参数寻优上要优于传统的模型决策树寻优方法,并且能够在迭代次数不多的情况下找到全局最优参数值,在一定程度上提升了算法的分类性能,节省了大量的调参时间。     

电传动装甲车辆负载功率预测实时能量管理策略

针对电传动装甲车辆负载功率预测功能缺失导致控制作用滞后的问题,提出一种具有较高负载功率预测精度的实时能量管理策略。在分析整车结构的基础上,采用理论分析和数据拟合方法,建立各动力源数学模型。将差分自回归移动平均模型和自适应马尔可夫链两种预测方法相结合,设计非平稳趋势性负载功率组合预测方法。在非线性模型预测控制框架下,构建多目标优化函数,采用序列二次规划法在有限时域内实时求解最优控制指令,优化多动力源协调控制过程。依托硬件在环仿真平台进行多路面行驶实验,对比有无功率预测的能量管理控制效果。结果表明,改进的实时能量管理策略对未来负载功率具有较好的预见性,能够显著优化多动力源协调控制过程,提升车辆燃油经济性,稳定母线电压和电池荷电状态,对传统模型预测控制下的工程应用场景具有一定借鉴意义。     

Shock initiation performance of NTO-based polymer bonded explosive

3-nitro-1,2,4-tri-azol-5-one (NTO) is a high energy insensitive explosive. To study the shock initiation process of NTO-based polymer bonded explosive JEOL-1 (32%octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX), 32% NTO, 28% Al and 8% binder system), the cylinder test, the gap experiments and numerical simulation were carried out. Firstly, we got the detonation velocity (7746 m/s) and the parameters of Jones-Wilkins-Lee (JWL) equation of state (EOS) for detonation product by cylinder test and numerical simulation. Secondly, the Hugoniot curve of unreacted explosive for JEOL-1 was obtained calculating the data of pressure and time at different Lagrangian positions. Then the JWL EOS of unreacted explosive was obtained by utilizing the Hugoniot curve as the reference curve. Finally, we got the pressure growth history of JEOL-1 under shock wave stimulation and the parameters of the ignition and growth reaction rate equation were obtained by the pressure-time curves measured by the shock-initiation gap experiment and numerical simulation. The determined trinomial ignition and growth model (IG model) parameters can be applied to subsequently simulation analysis and design of insensitive ammunition with NTO-based polymer bonded explosive.     

A quasi-isentropic model of a cylinder driven by aluminized explosives based on characteristic line analysis

A quasi-isentropic study on the process of driving a cylinder with aluminized explosives was carried out to examine the influence of the aluminum (Al) reaction rate on cylinder expansion and the physical parameters of the detonation products. Based on the proposed quasi-isentropic hypothesis and relevant isentropic theories, the characteristic lines of aluminized explosives driving a cylinder were analyzed, and a quasi-isentropic model was established. This model includes the variation of the cylinder wall velocity and the physical parameters of the detonation products with the Al reaction degree. Using previously reported experimental results, the quasi-isentropic model was verified to be applicative and accurate. This model was used to calculate the physical parameters for cylinder experiments with aluminized cyclotrimethylenetrinitramine explosives with 15.0 % and 30.0 % Al content. The results show that this quasi-isentropic model can be used not only to calculate the cylinder expansion rule or Al reaction degree, but also to calculate the physical parameters of the detonation products in the process of cylinder expansion. For explosives with 15.0 % and 30.0 % Al, 24.3 % and 18.5 % of the Al was found to have reacted at 33.9 μs and 34.0 μs, respectively. The difference in Al content results in different reaction intensity, occurrence time, and duration of two forms of reaction (diffusion and kinetic) between the Al powder and the detonation products; the post-detonation burning reaction between the Al powder and the detonation products prolongs the positive pressure action time, resulting in a continuous rise in temperature after detonation.     

量子保密通信技术前沿述评

量子保密通信是利用量子力学的基本原理来实现信息无条件安全传递的一种全新密码方法,其核心是量子密钥分发。2014年正好是量子保密通信的第一个协议—BB84协议提出30周年。经过30年的发展,量子保密通信在理论和实验上都得到了快速发展,实际工程应用也初具规模。文章对量子保密通信的历史发展及研究现状进行了简要回顾,特别介绍了基于实际量子保密通信系统的量子黑客和防御。最后,对未来的几个研究方向进行了简要讨论。     

基于随机拟合的悬浮式深弹弹阵特征建模

通过分析悬浮式深弹发射后在空中的弹道特性,建立了深弹弹着点坐标的计算模型。采取蒙特卡洛方法,分别对舰艇六自由度状态下,单管和六联装火箭深弹的弹着点坐标进行了仿真计算,对弹着点的分布规律进行了研究,并得出弹着点的联合密度函数。结果表明,在发射参数存在正态扰动下,弹着点散布区域均呈椭圆形分布,弹着点坐标均仍服从正态分布。     

水雾扩散模型仿真

建立水雾红外衰减模型时,通常把水雾空间看作单一均匀整体处理,然而雾滴的扩散现象使水雾的雾滴密度无法保证处处相等,所以仿真结果与实际情况相比普遍存在较大偏差。针对这一问题,基于高斯扩散理论与Mie散射理论,提出了一种新的建模方案。通过将水雾空间划分为多个子空间,并计算每个子空间内水雾消光性能的建模方法,提高建模的准确性。同时结合红外制导导弹来袭的方向变化,计算出不同方向上的水雾消光性能。从仿真结果可以看出,这种建模方式能够很好地克服以往建模方法的缺陷,为探索水雾的精确建模提供了一种新思路。     

Numerical simulation of the natural fragmentation of explosively loaded thick walled cylinders

The ability to predict the natural fragmentation of an explosively loaded metal casing would represent a significant achievement. Physically-based material models permit the use of small scale laboratory tests to characterise and validate their parameters. The model can then be directly employed to understand and design the system of interest and identify the experiments required for validation of the predictions across a wide area of the performance space. This is fundamentally different to the use of phenomenologically based material algorithms which require a much wider range of characterisation and validation tests to be able to predict a reduced area of the performance space. Eulerians numerical simulation methods are used to describe the fragmentation of thick walled EN24 steel cylinders filled with PBXN-109 explosive. The methodology to characterise the constitutive response of the material using the physically based Armstrong–Zerilli constitutive model and the Goldthorpe path dependent fracture model is described, and the results are presented. The ability of an Eulerian hydrocode to describe the fragmentation process and reproduce the experimentally observed fragment mass and velocity distributions is presented and discussed. Finally the suitability of the current experimental analysis methodology for simulation validation is addressed.     

Neural network modeling to evaluate the dynamic flow stress of high strength armor steels under high strain rate compression

An artificial neural network (ANN) constitutive model is developed for high strength armor steel tempered at 500 °C, 600 °C and 650 °C based on high strain rate data generated from split Hopkinson pressure bar (SHPB) experiments. A new neural network configuration consisting of both training and validation is effectively employed to predict flow stress. Tempering temperature, strain rate and strain are considered as inputs, whereas flow stress is taken as output of the neural network. A comparative study on Johnson–Cook (J–C) model and neural network model is performed. It was observed that the developed neural network model could predict flow stress under various strain rates and tempering temperatures. The experimental stress–strain data obtained from high strain rate compression tests using SHPB, over a range of tempering temperatures (500–650 °C), strains (0.05–0.2) and strain rates (1000–5500/s) are employed to formulate J–C model to predict the high strain rate deformation behavior of high strength armor steels. The J-C model and the back-propagation ANN model were developed to predict the high strain rate deformation behavior of high strength armor steel and their predictability is evaluated in terms of correlation coefficient (R) and average absolute relative error (AARE). R and AARE for the J–C model are found to be 0.7461 and 27.624%, respectively, while R and AARE for the ANN model are 0.9995 and 2.58%, respectively. It was observed that the predictions by ANN model are in consistence with the experimental data for all tempering temperatures.     

云模型在雷达干扰资源多目标优化配置中的应用

鉴于干扰效果/效益评估在雷达干扰资源优化分配中的重要地位,提出了一种基于云模型和匈牙利算法的雷达干扰资源优化分配模型。利用一维云模型单条件单规则发生器和逆向云发生器构建了评估指标干扰效果推理器,利用基于风险态度因子的云模型映射函数定义了云模型之间距离的度量方法。提出了一种基于云模型与逼近理想解法的雷达干扰效益矩阵求解方法,最后运用匈牙利算法实现了干扰资源优化分配。将所建模型应用到干扰资源分配中,结果表明该模型能够较好的处理评估中的不确定性知识,是合理、可行的。