To address the issues of ambiguous cluster numbers and random cluster center initialization in traditional clustering algorithms for underdetermined mixing matrix estimation, a novel estimation method based on multimodal fusion recognition is proposed. By optimizing the clustering process through multimodal feature fusion in this method, the limitations of conventional clustering algorithms are overcome. Experimental validation under a two-channel four-signal configuration demonstrates that the proposed method achieves a normalized mean square error of -66.034 5 dB for the estimated matrix. Compared with traditional clustering methods, this technique significantly improves mixing matrix estimation accuracy, providing a robust solution for practical applications.
混合信号经过基于多模态特征融合的时频混叠信号识别方法(Time-Frequency Aliasing Signal Recognition Method based on Multimodal Fusion, TRMM)处理,该方法采用时频图与波频图双模态进行特征提取,然后由波频图模态的输出对时频图模态进行加权融合,以多模态特征分值作为分类依据,实现时频混叠信号的分离识别,同时能够在输出结果上标注出混叠区域。TRMM流程图如图3所示。
BOUSSEM, DEBALSO, DE LATHAUWERL. A tensor-based method for large-scale blind source separation using segmentation[J]. IEEE Transactions on Signal Processing, 2017,65(2):346-358.
[2]
BOFILLP, ZIBULEVSKYM. Underdetermined blind source separation using sparse representations[J]. Signal Processing, 2001,81(11):2353-2362.
[3]
LEED D, SEUNGH S. Learning the parts of objects by non-negative matrix factorization[J]. Nature, 1999,401(6755):788-791.
KUMARV A, RAOC V R, LEEMAN. Audio source separation by estimating the mixing matrix in underdetermined condition using successive projection and volume minimization[J]. International Journal of Information Technology, 2023,15(4):1831-1844.
GUANW, DONGL L, ZHOUJ M, et al. Tensor-based approach for underdetermined operational modal identification[J]. Mechanical Systems and Signal Processing, 2021,160:No.107891.
HASSANN, RAMLID A. Underdetermined blind source separation of bioacoustic signals[J]. Pertanika Journal of Science and Technology, 2023,31(5):2257-2272.
[12]
XUD K, TIANY J. A comprehensive survey of clustering algorithms[J]. Annals of Data Science,2015,2(2):165-193.
[13]
潘宁. 复杂电磁环境下时频混叠信号的识别与分离[D].郑州:郑州大学,2021:1-70.
[14]
杨筱菡,王勇智.概率论与数理统计[M].北京:人民邮电出版社,2017:190-211.
[15]
LIUZ P, LIL C, ZHENGZ R. Mixing matrix estimation method for dual-channel time-frequency overlapped signals based on interval probability[J]. ETRI Journal, 2019,41(5):658-669.