| 摘要: |
| 针对海面小目标因体积小、移速慢而导致的检测难问题，提出了一种改进WTD-SVD-WOA-LSTM检测方法。首先，利用改进小波阈值法(Wavelet Threshold Denoising，WTD)结合优化奇异值分解(Singular Value Decomposition，SVD)法对海杂波去噪；然后，通过改进鲸鱼优化算法(Whale Optimization Algorithm，WOA)对长短期记忆神经网络(Long Short-term Memory，LSTM)的超参数选优，获得最佳预测模型；最后，根据预测误差均方根值进行小目标检测。利用冰区多参数成像X频段雷达(Ice Multiparameter Imaging X-band Radar，IPIX)实测海杂波数据进行验证，所提方法相较于单一LSTM检测方法，检测阈值区间更广，检测能力至少提高了16%。 |
| 关键词: 小目标检测 海杂波去噪 改进鲸鱼优化算法 长短期记忆神经网络 |
| DOI：10.20079/j.issn.1001-893x.230405002 |
|
| 基金项目:国家自然科学基金资助项目（61801196）；国防基础科研计划稳定支持专题项目（JCKYS2020604SSJS010）；江苏省研究生科研与实践创新计划资助项目(SJCX22_1889) |
|
| Small Target Detection in Sea Clutter Background Based on Improved WTD-SVD-WOA-LSTM Method |
| ZHU Jian,SHANG Shang,SHI Yishan,QIAO Tiezhu,LIU Qiang |
| (Ocean College,Jiangsu University of Science and Technology,Zhenjiang 212100，China) |
| Abstract: |
| For the problem of difficult detection of small targets on the sea surface due to small size and slow moving speed,an improved WTD-SVD-WOA-LSTM detection method is proposed.First,the improved wavelet threshold denoising (WTD) combined with the optimized singular value decomposition (SVD) method is used to denoise the sea clutter.Then,the improved whale optimization algorithm (WOA) is used to optimize the hyperparameters of long short-term memory (LSTM) neural network,and obtain the best prediction model.Finally,the root mean square value of prediction error is used to detect small targets.The sea clutter data measured by Ice Multiparameter Imaging X-Band Radar (IPIX) are used for verification.Compared with the single LSTM detection method,the proposed method has a wider detection threshold range and the detection ability is improved by at least 16%. |
| Key words: small target detection sea clutter denoising improved whale optimization algorithm long short-term memory neural network |
