| 引用本文: | 边靖伟,寇立伟,项基.应用PSO和SVM的水下航行器黑箱建模[J].哈尔滨工业大学学报,2019,51(10):55.DOI:10.11918/j.issn.0367-6234.201807078 |
| BIAN Jingwei,KOU Liwei,XIANG Ji.Black-box modeling based on PSO and SVM for underwater vehicles[J].Journal of Harbin Institute of Technology,2019,51(10):55.DOI:10.11918/j.issn.0367-6234.201807078 |
|
| 摘要: |
| 随着新型水下航行器不断涌现,现有水下航行器数学模型已难以与实际模型吻合.为更好了解新型水下航行器实际模型以及预测新型水下航行器运动,提出应用粒子群（particle swarm optimization,PSO）参数寻优和支持向量机（support vector machine,SVM）的水下航行器黑箱建模方法.首先根据水下航行器的运动状态信息和推进器力,应用支持向量机构造出之间的非线性映射关系,然后通过粒子群智能优化算法获得支持向量机的最佳参数组合,进而实现水下航行器的黑箱建模,最后根据推进器力是否时变,分别以新型四旋翼水下航行器的两种空间运动进行实验验证,并以均方根误差作为空间运动预测结果的评价标准. 试验结果表明, 基于粒子群参数寻优和支持向量机所构建的水下航行器黑箱模型对空间运动预测具有较小的均方根误差,空间运动预测结果与实际运动基本一致,所建黑箱模型与实际模型基本吻合,能有效预测水下航行器运动状态. |
| 关键词: 水下航行器 黑箱建模 支持向量机 粒子群 空间运动 |
| DOI：10.11918/j.issn.0367-6234.201807078 |
| 分类号:U661 |
| 文献标识码:A |
| 基金项目:国家自然科学基金(61773339); 浙江省重点研发项目(2019C02002) |
|
| Black-box modeling based on PSO and SVM for underwater vehicles |
|
BIAN Jingwei,KOU Liwei,XIANG Ji
|
|
（College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China）
|
| Abstract: |
| The existing mathematical model of underwater vehicles is difficult to match the actual model with new emerging of underwater vehicles. In order to deal with modeling problem and predict space motion for new underwater vehicles, a black-box modeling method based on particle swarm optimization (PSO) and support vector machine (SVM) was proposed. Nonlinear mapping relationship between state of motion and thrusters for underwater vehicles was constructed by SVM. Optimal parameters of SVM were obtained through PSO algorithm. Then, a black-box model was established for underwater vehicles. Finally, by judging whether thrusters vary with time, space motion of a new kind of quadrotor underwater vehicle was adopted to verify the effectiveness of the proposed method. Space motion prediction results were evaluated by the root mean square error. The experimental results demonstrated that root mean square errors of the space motion prediction results were small. Space motion prediction results were in accordance with the actual space motion. The black-box model constructed by PSO and SUM was basically identical with the actual model and could effectively predict the space motion of underwater vehicles. |
| Key words: underwater vehicle black-box modeling support vector machine (SVM) particle swarm optimization (PSO) space motion |
