摘要
针对机载动力系统测试数据的不确定性,求解参数实时性差的问题,提出了基于快速寻找密度极点聚类与蚁群极限学习机的机载动力系统的参数估计方法。首先利用基于寻找密度极点的聚类算法对全工况范围内的测试数据进行聚类,然后在每一个子类中用极限学习机设计了子参数估计器,并用蚁群算法寻找极限学习机的最优隐层神经元数目。训练与测试表明,参数估计测试相对误差明显优于传统的RBF神经网络方法,且参数估计时间能够满足机载在线实时状态评估的需求,该方法可应用到其他不可测参数的估计。
For the uncertainty of the test data of the airborne power system and the problem of solving parameters with poor real-time performance, a parameter estimation method of the airborne power system based on clustering by fast search and find of density peaks （CFSFDP） and ant colony optimization ex- treme learning machine （ACO-ELM） is proposed. Firstly, the CFSFDP method is utilized to cluster the test bench data in the whole behavior range, and then, a sub-estimator is designed in each cluster using ACO-ELM. In the process of designing the sub-estimator with ACO-ELM, the particle swarm optimization algorithm is utilized to search the best hidden node number of extreme learning machine for getting the best topological structure. Finally, the training and testing results show that the maximum mean relative error is better than the RBF network, which meets the demand of thrust control and onboard real time state assessment. The method can be used for estimating other immeasurable parameters.
作者
孟蕾
许爱强
牛景华
MENG Leila XU Ai-qianglb NIU Jing-hua(Naval Aeronautical and Astronautical University, a. Scientific Research Department b. Department of Airborne Vehicle Engineering, Shandong Yantai 264001, China 2. PLA, No. 92212 Troop, Shandong Qingdao 266000, Chin)
出处
《现代防御技术》
北大核心
2017年第2期172-176,216,共6页
Modern Defence Technology
基金
“十二五”国防技术基础科研项目(Z052013B004)
关键词
飞行器
推力
参数估计
蚁群
快速寻找密度极点聚类
蚁群极限学习机
vehicle
thrust
parameter estimation
ant colony
clustering by fast search and find ofdensity peaks （CFSFDP）
ant colony optimization extreme learning machine（ACO-ELM）
