基于进化算法低信噪比环境的基音频率检测  

Evolutionary algorithm based fundamental tone frequency detection in low SNR environment

作　　者：张小恒[1,2] 李勇明[2] 谢文宾 

机构地区：[1]重庆广播电视大学,重庆400052 [2]重庆大学通信工程学院,重庆400030

出　　处：《现代电子技术》2017年第11期46-52,共7页Modern Electronics Technique

基　　金：国家自然科学基金(61108086;61171089;91438104;61471073);CSTC(cstc2012jj A40015);中央高校基金(CDJZR12160011;CDJZR155507;CDJZR13160008;CDJZ165501);国家博士后基金(2013M532153);重庆博士后特别资助;重庆市教委科学技术研究项目资助(KJ1603805)

摘　　要：构造频域空间的检测模型,将基音频率作为特征值进行提取,然后为检测模型引入模型参数即优化因子,通过进化算法对该因子进行全局优化,从而获取基音频率的全局最优值,在优化精度和时间代价上取得了较好的平衡。采用两种具有代表性的进化算法进行算法设计,包括遗传算法(GA算法)和粒子群算法(PSO算法)。将所提算法与相关有代表性的算法进行比较,结果表明,所提算法在不同类型不同程度的噪声环境下,能显著提升检测识别率,尤其是在极低信噪比下,优势更为明显。A frequency-domain detection model was constructed. The fundamental tone frequency is extracted as characteristic value. The model parameter(optimization factor)is introduced into the detection model,for which the global optimization is carried out with evolutionary algorithm(EA)to get the global optimum of the fundamental tone frequency,and obtain a better balance in optimization accuracy and time cost. Two representative EAs(genetic algorithm(GA)and particle swarm optimization(PSO)algorithm)are used to perform the algorithm design. The proposed algorithm is compared with the other representative algorithms. The comparison results show that the proposed algorithm can improve the detection recognition rate greatly in the noise environments of different types and different degrees,especially in the very-low SNR environment.

关 键 词：极低信噪比环境 基音频率 进化算法 遗传算法 粒子群算法 

分 类 号：TN912.3[电子电信—通信与信息系统]