首页|链式回转弹仓区间不确定性动力学模型

链式回转弹仓区间不确定性动力学模型

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
针对具有区间不确定性参数的辨识问题，提出一种基于区间可能度转换模型的区间不确定性参数的双层嵌套辨识(Double-layer Nested Identification，DNI)方法。通过将待辨识参数分为两类，利用DNI方法辨识出第1 类确定性参数，再通过基于DNI思想的区间优化方法优化第 2 类区间不确定性参数的区间范围;面向嵌套策略类型方法计算量庞大且效率低的问题，选用贝叶斯优化-粒子群优化(Bayesian Optimization-Particle Swarm Optimization，BO-PSO)方法作为内层算法以提高求解效率。DNI方法的内层利用 BO-PSO 方法计算区间上下界，外层利用改进型布谷鸟搜索(Improved Cuckoo Search，ICS)方法辨识特定参数。为进一步缩短求解时间，提出一种ICS多核极限学习机(ICS-Multiple Kernel-Extreme Learning Machine，ICS-MK-ELM)代理模型，ICS-MK-ELM代理模型克服了人工调节每个核函数超参数的困难，并且模型预测精度明显高于核 ELM(Kernel ELM，KELM)和MK-ELM;将DNI方法应用于链式回转弹仓的参数辨识，解决了链式弹仓具有区间不确定性参数的辨识困难的问题，参数辨识结果表明所提DNI方法以及基于DNI思想的区间优化方法具有更高的精度和稳定性。
A Dynamic Model of Interval Uncertainty of Rotational Chain Shell Magazine
For the identification problem with interval uncertain parameters,a double-layer nested identification(DNI)method based on an interval possibility degree transformation model is proposed.By dividing the parameters to be identified into two categories,the first type of deterministic parameters are identified by DNI method,and the interval range of the second type of interval uncertainty parameters is optimized by the DNI-based interval optimization method.The BO-PSO algorithm is chosen as the inner-layer algorithm to improve the efficiency of the nested strategy type method.For the inner layer of DNI method,BO-PSO method is used to calculate the upper and lower bounds of interval,and for the outer layer,ICS method is used to identify the specific parameters.In order to shorten the solving time,an ICS-MK-ELM agent model is proposed.The ICS-MK-ELM agent model overcomes the difficulty of manually adjusting the hyper-parameters of each kernel function,and the prediction precision of the model is obviously higher than those of KELM and MK-ELM.Finally,the DNI method is applied to the parameter identification of the rotational chain shell magazine,which solves the problem of the parameter identification of the chain-type magazine with interval uncertainty.The results of parameter identification show that the DNI method and the interval optimization method based on DNI have higher accuracy and stability.

uncertaintyinterval possibility degreeshell magazineparameter identificationmultiple kernel extreme learning machineBayesian optimizationcuckoo search algorithm

赵伟、侯保林、闫少军、鲍丹、林瑜斌

展开 >

南京理工大学 机械工程学院,江苏 南京 210094

南京理工大学 瞬态物理国家重点实验室,江苏 南京 210094

内蒙古第一机械集团有限公司,内蒙古 包头 014000

不确定性 区间可能度 弹仓 参数辨识 多核极限学习机 贝叶斯优化 布谷鸟搜索方法

南京理工大学瞬态物理国家重点实验室项目

2022-JCJQ-LB-061-02

2024

10.12382/bgxb.2023.0379

兵工学报
中国兵工学会

兵工学报

CSTPCD北大核心
影响因子：0.735
ISSN：1000-1093
年,卷(期)：2024.45(6)
  • 12