ID 原文 译文
9334 ‭通过构建统一的学习框架,在保证每一次训练迭代中学习过程稀疏化的同时,实现了核权重系数与正则化因子的同步更新。 By building a unified framework of learning and training in every iteration learning process of thinning at the same time, realize the kernel weight coefficient and regularization factor of synchronous updating. ‭
9335 实验结果表明,提出的方法相比与其他基于KELM的在线序贯学习方法,在有无噪声的情况下,均可以有效提升辨识精度,并且具有更好的稳定性。 The experimental results show that the proposed method compared with other online sequential learning method based on KELM, under the condition of the presence of noise, can effectively improve identification precision, and has better stability.
9336 针对现有城市系统作战能力评估方法较少的问题,利用反向传播(back propagation,BP)神经网络在能力评估方面所具有的自适应、自学习、强容错性和泛化映射等优势,建立了评估指标体系并给出了指标的隶属函数。 In view of the existing city less operational capability evaluation system, using the back propagation (BP) back propagation, the neural network in terms of capability assessment of adaptive, self-learning and strong fault tolerance and the advantages of the generalized mapping, established the evaluation index system and index of membership function is presented.
9337 通过模拟退火遗传算法(simulated annealing and genetic algorithm,SAGA)优化BP神经网络的连接权重和阀值,弱化了指标评价中的人为因素,提高了评价结果的准确性、客观性和权威性,有效解决了传统遗传算法和BP神经网络易陷入局部极小值、收敛速度慢和抗干扰能力差等问题。 Through the simulated annealing genetic algorithm (simulated annealing and based algorithm, SAGA) optimizing the connection weights and threshold of BP neural network, weaken the human factors in the evaluation of indicators, improve the accuracy of the evaluation results, objectivity and authoritativeness, effectively solve the traditional genetic algorithm and BP neural network easy to fall into local minimum value, slow convergence speed and poor anti-interference ability.
9338 仿真实例验证了该方法对城市系统作战能力评估的可行性和有效性。 Simulation examples show that this method is feasibility and effectiveness of operational capability evaluation of urban system.
9339 为提高导弹拦截高速机动目标的精度,基于自抗扰控制理论的估计补偿思想,设计了考虑自动驾驶仪动态特性和目标机动的三维动态面导引律。 In order to improve the precision of the missile to intercept the maneuvering target, based on the immunity control theory of the estimated compensation of ideas, design with the consideration of autopilot dynamic characteristics and the surface of the 3 d dynamic target maneuver guidance law.
9340 首先,建立了考虑自动驾驶仪动态特性的三维耦合制导模型; ‭First of all, set up considering dynamic characteristics of autopilot, three-dimensional coupling guidance model;
9341 其次,针对制导模型中所存在的目标机动和测量噪声的干扰,设计扩张观测器估计目标机动和视线角速率,并将其应用到导引律的设计中; Secondly, in view of the guidance model of target maneuver and the interference of measurement noise, the design of expansion observer to estimate target maneuver and the line of sight angular rate, and apply it to the design of guidance law;
9342 再次,基于动态面控制方法设计了三维空间导引律,避免了传统反演控制方法中的"微分膨胀"问题; Again, based on the dynamic surface control method to design the guidance law in three dimensional space, avoid the traditional inversion control method of "differential expansion" problem;
9343 最后,在目标作不同机动情况下,所设计的导引律与比例导引律、动态面导引律进行比较,仿真结果表明所设计的导引律具有更好的制导性能。 Finally, the target for different mobile case, the designed guidance law and the proportional guidance law, dynamic guidance law are compared, the result of simulation shows that the designed guidance law has better guidance performance.