基于样条插值与人工蜂群优化的非线性盲源分离算法

doi:10.11959/j.issn.1000-436x.2017147

通信学报 ›› 2017, Vol. 38 ›› Issue (7): 36-46.doi: 10.11959/j.issn.1000-436x.2017147

基于样条插值与人工蜂群优化的非线性盲源分离算法

陈雷¹,甘士忠²,张立毅¹,王光艳¹

¹ 天津商业大学信息工程学院，天津 300134
² 天津工业大学电子信息工程学院，天津 300387

修回日期:2017-05-06 出版日期:2017-07-01 发布日期:2017-08-25
作者简介:陈雷（1980-），男，河北唐山人，博士后，天津商业大学副教授，主要研究方向为盲信号处理、仿生智能计算等。|甘士忠(1994-)，男，河南信阳人，天津工业大学硕士生，主要研究方向为盲信号处理、仿生智能计算等。|张立毅（1963-），男，山西忻州人，博士，天津商业大学教授、博士生导师，主要研究方向为盲信号处理、信号检测与处理等。|王光艳（1975-），女，河北邯郸人，博士，天津商业大学副教授，主要研究方向为盲信号处理、语音增强、水下语音通信等。
基金资助:
国家自然科学基金资助项目(61401307);中国博士后科学基金资助项目(2014M561184);天津市应用基础与前沿技术研究计划基金资助项目(15JCYBJC17100);天津市应用基础与前沿技术研究计划基金资助项目(14JCZDJC32600);天津市科技特派员基金资助项目(16JCTPJC48400)

Nonlinear blind source separation algorithm based on spline interpolation and artificial bee colony optimization

Lei CHEN¹,Shi-zhong GAN²,Li-yi ZHANG¹,Guang-yan WANG¹

¹ School of Information Engineering,Tianjin University of Commerce,Tianjin 300134,China
² School of Electronics and Information Engineering,Tianjin Polytechnic University,Tianjin 300387,China

Revised:2017-05-06 Online:2017-07-01 Published:2017-08-25
Supported by:
The National Natural Science Foundation of China(61401307);China Postdoctoral Science Foundation(2014M561184);Tianjin Research Program of Application Foundation and Advanced Technology of China(15JCYBJC17100);Tianjin Research Program of Application Foundation and Advanced Technology of China(14JCZDJC32600);Tianjin Research Program of Science and Technology Commissioner of China(16JCTPJC48400)

摘要/Abstract

摘要：

针对更加复杂的非线性混合情况，提出一种基于样条插值拟合与群智能优化的后非线性盲源分离算法。采用样条插值函数拟合去非线性函数，使用负熵作为分离的评价准则，建立分离模型。分离过程采用改进的人工蜂群算法优化求解样条插值节点参数，并在分离的目标函数中引入相关性约束条件进行解空间范围限制，克服分离过程中存在的异常值现象。针对语音数据的分离实验结果表明，所提算法能够有效实现非线性混合信号的盲分离，较传统的基于奇数多项式拟合的分离算法具有更高的分离精度。

关键词: 盲源分离, 后非线性, 样条插值, 群智能优化, 人工蜂群算法

Abstract:

A post-nonlinear blind source separation algorithm based on spline interpolation fitting and artificial bee colony optimization was proposed for the more complicated nonlinear mixture situations.The separation model was constructed by using the spline interpolation to fit the inverse nonlinear distortion function and using entropy as the separation criterion.The spline interpolation node parameters were solved by the modified artificial bee colony optimization algorithm.The correlation constraint was added into the objective function for limiting the solution space and the outliers wuld be restricted in the separation process.The results of speech sounds separation experiment show that the proposed algorithm can effectively realize the signal separation for the nonlinear mixture.Compared with the traditional separation algorithm based on odd polynomial fitting,the proposed algorithm has higher separation accuracy.

Key words: blind source separation, post nonlinear, spline interpolation, swarm intelligence optimization, artificial bee colony algorithm

中图分类号:

TN911.7

陈雷,甘士忠,张立毅,王光艳. 基于样条插值与人工蜂群优化的非线性盲源分离算法[J]. 通信学报, 2017, 38(7): 36-46.

Lei CHEN,Shi-zhong GAN,Li-yi ZHANG,Guang-yan WANG. Nonlinear blind source separation algorithm based on spline interpolation and artificial bee colony optimization[J]. Journal on Communications, 2017, 38(7): 36-46.

图/表 13

图1

图2

图3

图4

图5

图6

图7

表1

图8

图9

表2

表3

表4

参考文献 33

[1]	HYVARINEN A , KARHUNEN J , OJA E . Independent component analysis[M]. JohnWiley ＆ Sons, 2001.
[2]	KURAYA M , UCHIDA A , YOSHIMORI S ,et al. Blind source separation of chaotic laser signals by independent component analysis[J]. Optics Express, 2008,16(2): 725-730.
[3]	MATILAINEN M , NORDHAUSEN K , OJA H . New independent component analysis tools for time series[J]. Statistics ＆ Probability Letters, 2015,105: 80-87.
[4]	DIAMANTARAS K I , PAPADIMITRIOU T . Applying PCA neural models for the blind separation of signals[J]. Neurocomputing, 2009,73(1-3): 3-9.
[5]	STONE J V . Blind source separation using temporal predictability[J]. Neural Computation, 2001,13(7): 1559-1574.
[6]	陈雷, 张立毅, 郭艳菊 ,等. 基于时间可预测性的差分搜索盲信号分离算法[J]. 通信学报, 2014,35(6): 117-125.
	CHEN L , ZHANG L Y , GUO Y J ,et al. Blind signal separation algorithm based on temporal predictability and differential search algorithm[J]. Journal on Communications, 2014,35(6): 117-125.
[7]	GUAN L , KEARNEY R , ZHU C Y A ,et al. High-performance digital predistortion test platform development for wideband RF power amplifiers[J]. International Journal of Microwave and Wireless Technologies, 2013,5(5): 149-162.
[8]	ACCARDO A , CUSENZA M , MONTI F . Linear and non-linear parameterization of EEG during monitoring of carotid endarterectomy[J]. Computers in Biology and Medicine, 2009,39(6): 512-518.
[9]	HYVARINEN A , PAJUNEN P . Nonlinear independent component analysis:existence and uniqueness results[J]. Neural Networks, 1999,12(3): 429-439.
[10]	TALEB A , JUTTEN C . Source separation in post-nonlinear mixtures[J]. IEEE Transactions on Signal Processing, 1999,47(10): 2807-2802.
[11]	FILHO E F S , SEIXAS J M , CALOBA L P . Modified post-nonlinear ICA model for online neural discrimination[J]. Neurocomputing, 2010,73(6-8): 820-2828.
[12]	DUARTE L T , SUYAMA R , RIVET B ,et al. Blind compensation of nonlinear distortions:application to source separation of post-nonlinear mixtures[J]. IEEE Transactions on Signal Processing, 2012,60(11): 5832-5844.
[13]	AZIZ N B A , ABDULLAH W F H , TAHIR N M . Implementation of nonlinear blind source separation for CHEMFET sensor arrays[C]// The 2014 IEEE 10th International Colloquium on Signal Processing and its Applications. 2014: 238-241.
[14]	LEE T W , KOEHLER B U , ORGLMEISTER R . Blind source separation of nonlinear mixing models[C]// The 1997 IEEE Signal Processing Society Workshop:Neural Networks for Signal Processing VII. 1997: 406-415.
[15]	YANG H H , AMARI S , CICHOCKI A . Information-theoretic approach to blind separation of source in non-linear mixture[J]. Signal Processing Litters, 2000,7(7): 197-200.
[16]	TAN Y , WANG J , ZURADA J . Nonlinear blind source separation using a radial basis function network[J]. IEEE Transactions on Neural Networks, 2001,12(1): 124-134.
[17]	KARABOGA N . A new design method based on artificial bee colony algorithm for digital IIR filters[J]. Journal of the Franklin Institute, 2009,346(4): 328-348.
[18]	ATYABI A , LUERSSEN M H , POWERS D M W . PSO-based dimension reduction of EEG recordings:implications for subject transfer in BCIO[J]. Neurocomputing, 2013,119: 319-331.
[19]	KUMAR E V , RAAJA G S , JEROME J . Adaptive PSO for optimal LQR tracking control of 2 DoF laboratory helicopter[J]. Applied Soft Computing, 2016,41: 77-90.
[20]	陈雷, 张立毅, 郭艳菊 ,等. 基于细菌群体趋药性的有序音信号分离算法[J]. 通信学报, 2011,32(4): 77-85.
	CHEN L , ZHANG L Y , GUO Y J ,et al. Sequential blind signal algorithm based on bacterial colony chemo taxis[J]. Journal on Communications, 2011,32(4): 77-85.
[21]	张银雪, 田学民, 邓晓刚 . 基于改进人工蜂群算法的盲源分离方法[J]. 电子学报, 2012,40(10): 2026-2030.
	ZHANG Y X , TIAN X M , DENG X G . Blind source separation based on modified bee colony algorithm[J]. Acta Electronica Sinica, 2012,40(10): 2026-2030.
[22]	MAVADDATY S , EBRAHIMZADEH A . A comparative study of bees colony algorithm for blind source separation[C]// The 20th Iranian Conference on Electrical Engineering. 2012: 1172-1177.
[23]	CHEN L , ZHANG L Y , GUO Y J ,et al. Blind source separation based on covariance ratio and artificial bee colony algorithm[J]. Mathematical Problems in Engineering, 2014,484327.
[24]	GORRIZ J M , PUNTONET C G , ROJAS F . Optimizing blind source separation with guided genetic algorithms[J]. Neurocomputing, 2006,69(13-15): 1442-1457.
[25]	TAN Y , WANG J . Nonlinear blind source separation using higher order statistics and a genetic algorithm[J]. IEEE Transactions on Evolutionary Computation, 2002,5(6): 600-612.
[26]	KARABOGA D . An idea based on honey bee swarm for numerical optimization[R]. Technical Report-TR06, 2005.
[27]	SINGH A . An artificial bee colony algorithm for the leaf-constrained minimum spanning tree problem[J]. Applied Soft Computing, 2009,9(2): 625-631.
[28]	SABAT S L , UDGATA S K , ABRAHAM A . Artificial bee colony algorithm for small signal model parameter extraction of MESFET[J]. Engineering Applications of Artificial Intelligence, 2010,23(5): 689-694.
[29]	MANOJ V J , ELIAS E . Artificial bee colony algorithm for the design of multiplier-less nonuniform filter bank transmultiplexer[J]. Information Sciences, 2012,192: 193-203.
[30]	GAO W F , LIU S Y . Modified artificial bee colony algorithm[J]. Computers ＆ Operations Research, 2012,39(3): 687-697.
[31]	HYVARINEN A . Fast and robust fixed-point algorithms for independent component analysis[J]. IEEE Transactions on Neural Networks, 1999,10(3): 626-634.
[32]	FREEDMAN D , PISANI R , PURVES R . Statistics[M]. W.W.Norton＆ Company. 2007.
[33]	BOFILL P , ZIBULEVSKY M . Underdetermined blind source separation using sparse representations[J]. Signal Processing, 2001,81(11): 2353-2362.

性能指标	y_i		非线性畸变较弱情况			非线性畸变较强情况
性能指标	y_i	信号1	信号2	信号3	信号1	信号2	信号3
	y₁	0.962 8	0.753 7	0.855 3	0.857 7	0.579 4	0.485 8
AVCC	y₂	0.919 4	0.992 1	0.972 9	0.847 9	0.915 7	0.727 3
	y₃	0.824 2	0.785 2	0.989 0	0.778 4	0.759 1	0.594 9
	均值	0.902 1	0.843 7	0.939 1	0.828 0	0.751 4	0.602 7
	y₁	33.608 8	98.262 8	39.250 2	14.090 9	59.797 7	39.337 4
MSE(×10^-3)	y₂	6.217 9	0.799 2	16.613 2	32.167 8	6.012 8	35.192 7
	y₃	50.984 4	39.436 5	3.328 8	26.718 9	69.537 0	103.053 0
	均值	30.270 4	46.166 2	19.730 7	24.325 9	45.115 8	59.194 4
	y₁	2.528 0	3.727 2	3.666 3	6.292 2	-0.695 6	4.029 1
RSNR/dB	y₂	6.945 2	16.508 6	6.120 0	-0.192 9	7.472 5	2.855 9
	y₃	-1.238 0	5.140 2	14.831 7	1.568 1	1.253 9	0.118 0
	均值	2.745 1	8.458 7	8.206 0	2.555 8	2.676 9	2.334 3

性能指标	y_i		信号1			信号2			信号3
性能指标	y_i	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法
	y₁	0.996 1	0.997 9	0.998 3	0.998 6	0.999 3	0.998 0	0.998 9	0.996 1	0.997 0
AVCC	y₂	0.999 5	0.999 0	0.999 4	0.998 1	0.999 6	0.999 2	0.999 8	0.995 3	0.998 3
	y₃	0.994 3	0.999 2	0.998 8	0.999 2	0.998 8	0.999 5	0.998 2	0.994 5	0.999 4
	均值	0.996 6	0.998 7	0.998 8	0.998 6	0.999 2	0.998 9	0.999 0	0.995 3	0.998 2
	y₁	0.118 6	0.355 3	0.118 1	0.676 1	2.742 8	0.420 0	2.385 0	0.073 8	0.253 5
MSE(×10^-3)	y₂	0.034 2	0.180 2	0.029 0	0.427 2	0.140 4	0.087 2	0.049 9	0.144 8	0.260 5
	y₃	0.163 3	0.115 0	0.199 0	0.370 7	1.005 3	1.024 6	2.779 1	0.130 6	0.049 3
	均值	0.105 4	0.216 8	0.115 4	0.491 3	1.296 2	0.510 6	1.738 0	0.116 4	0.187 8
	y₁	21.567 8	15.222 7	23.371 9	25.335 0	28.387 3	23.030 6	27.059 8	20.582 4	22.275 6
RSNR/dB	y₂	32.812 9	26.846 4	28.915 6	22.655 3	23.653 2	24.104 1	30.263 2	20.614 9	22.322 9
	y₃	19.221 2	19.957 4	19.874 8	26.965 0	28.515 1	33.748 7	22.848 2	21.460 5	25.229 7
	均值	24.534 0	20.675 5	24.054 1	24.985 1	26.851 9	26.961 1	26.723 7	20.885 9	23.276 1

性能指标	y_i		信号1			信号2			信号3
性能指标	y_i	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法
	y₁	0.411 8	0.991 0	0.996 7	0.382 6	0.982 1	0.995 8	0.315 9	0.982 8	0.991 8
AVCC	y₂	0.952 5	0.992 1	0.996 8	0.842 4	0.991 7	0.996 4	0.718 0	0.997 1	0.998 6
	y₃	0.837 9	0.996 8	0.996 2	0.954 2	0.997 9	0.990 3	0.780 4	0.995 7	0.998 0
	均值	0.734 1	0.993 3	0.996 6	0.726 4	0.990 6	0.994 2	0.604 8	0.991 9	0.996 1
	y₁	75.308 0	1.236 6	0.401 9	80.494 9	3.830 6	0.354 7	165.088 6	3.929 3	3.107 5
MSE(×10^-3)	y₂	10.978 5	3.250 4	0.637 9	22.323 7	1.136 3	0.687 2	71.341 1	7.248 6	1.240 6
	y₃	20.888 5	1.920 7	0.671 1	52.725 2	2.255 1	2.070 1	119.473 7	4.537 4	1.452 5
	均值	35.725 0	2.135 9	0.570 3	51.847 9	2.407 3	1.037 3	118.634 5	5.238 4	1.933 5
	y₁	-0.986 8	21.071 9	24.746 3	-1.987 6	13.120 7	24.540 0	-2.572 5	14.098 0	15.297 4
RSNR/dB	y₂	4.475 9	15.874 9	21.312 7	1.774 2	16.951 7	20.389 6	-0.213 0	12.697 4	23.152 8
	y₃	2.637 3	22.228 5	21.877 3	2.454 5	20.951 5	17.846 7	-0.792 4	15.071 8	18.175 6
	均值	2.042 1	19.725 1	22.645 4	0.747 0	17.008 0	20.925 4	-1.192 6	13.955 7	18.875 3

性能指标	y_i		信号1			信号2			信号3
性能指标	y_i	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法	文献[23]算法	文献[24]算法	本文算法
	y₁	0.375 9	0.769 4	0.989 0	0.343 9	0.875 8	0.989 3	0.259 6	0.836 3	0.941 6
AVCC	y₂	0.920 0	0.912 0	0.991 8	0.898 7	0.927 1	0.982 5	0.832 6	0.959 1	0.987 5
	y₃	0.855 7	0.877 6	0.991 4	0.933 5	0.961 5	0.978 7	0.811 9	0.973 1	0.978 0
	均值	0.717 2	0.853 0	0.990 7	0.725 4	0.921 5	0.983 5	0.634 7	0.922 8	0.969 0
	y₁	106.752 4	28.858 0	1.441 9	110.929 6	15.997 7	1.198 2	204.193 5	97.798 2	18.350 2
MSE(×10^-3)	y₂	27.951 7	15.063 3	3.504 7	34.870 4	7.371 1	2.896 0	49.049 0	22.938 0	4.836 9
	y₃	34.214 6	46.920 1	1.323 0	62.250 2	8.458 1	3.491 6	61.544 5	11.916 4	12.571 0
	均值	56.306 2	30.280 5	2.089 9	69.350 1	10.609 0	2.528 6	104.929 0	44.217 5	11.919 4
	y₁	-2.502 2	9.062 5	17.471 7	-3.380 4	6.763 3	18.535 1	-3.495 7	3.109 9	11.159 2
RSNR/dB	y₂	0.429 5	10.548 4	13.306 8	-0.162 7	8.155 4	13.984 1	1.414 1	9.502 1	16.741 4
	y₃	0.472 2	9.873 7	17.554 6	1.733 6	15.748 6	15.978 3	2.088 4	10.772 7	14.604 6
	均值	-0.533 5	9.828 2	16.111 0	-0.603 2	10.222 4	16.165 8	0.002 3	7.794 9	14.168 4

基于样条插值与人工蜂群优化的非线性盲源分离算法

Nonlinear blind source separation algorithm based on spline interpolation and artificial bee colony optimization

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 33

相关文章 12

Metrics

推荐阅读 0

[1]	解元, 邹涛, 孙为军, 谢胜利. 面向高混响环境的欠定卷积盲源分离算法[J]. 通信学报, 2023, 44(2): 82-93.
[2]	马宝泽, 张天骐, 安泽亮, 邓盼. 基于张量分解的卷积盲源分离方法[J]. 通信学报, 2021, 42(8): 52-60.
[3]	张天骐,马宝泽,强幸子,全盛荣. 带自适应动量因子的变步长盲源分离方法[J]. 通信学报, 2017, 38(3): 16-24.
[4]	叶苗,王宇平,代才,王晓丽. 无线传感器网络中新的最小暴露路径问题及其求解算法[J]. 通信学报, 2016, 37(1): 49-60.
[5]	栗华,贾智平,王洪君,刘琚. 基于动态位隙分组盲分离的UHF RFID防碰撞算法[J]. 通信学报, 2012, 33(4): 47-53.
[6]	王荣杰,詹宜巨,周海峰. 基于零空间表示和最大似然的欠定盲源分离方法[J]. 通信学报, 2012, 33(3): 183-190.
[7]	杨小牛,姚俊良,李建东,李钊,张琰. M-PSK/QAM系统中基于相位预编码和盲源分离的共道干扰消除[J]. 通信学报, 2010, 31(8A): 154-160.
[8]	姚俊良,杨小牛,李建东,付卫红,李钊,张琰. 基于最大比合并的超定盲源分离[J]. 通信学报, 2010, 31(7): 9-17.
[9]	张晋东,秦贵和,陈涛,金健. 基于FastICA-TDS的DS-CDMA盲源信号分离系统[J]. 通信学报, 2008, 29(8): 126-130.
[10]	林秋华,党杰,殷福亮. 盲源分离图像加密的相关运算解密法[J]. 通信学报, 2008, 29(1): 109-114.
[11]	郭文强,邱天爽,查代奉. 基于Borel测度峰值判定的欠定混合盲语音信号分离[J]. 通信学报, 2007, 28(9): 22-26.
[12]	谷波,刘琚,许宏吉. 基于独立分量分析的正交空时分组编码盲检测方案[J]. 通信学报, 2006, 27(12): 127-131.