基于逻辑斯蒂回归的智能机器人语音指令识别方法  

Speech Instruction Recognition Method of Intelligent Robot Based on Logistic Regression

作　　者：冯玉雷 FENG Yulei(Heilongjiang Vocational College(Heilongjiang Economic Management Cadre College),Harbin 150080,China)

机构地区：[1]黑龙江职业学院(黑龙江省经济管理干部学院),黑龙江哈尔滨150080

出　　处：《科技创新与生产力》2024年第11期142-144,共3页Sci-tech Innovation and Productivity

摘　　要：为了提高智能机器人语音指令识别精度,本文提出了基于逻辑斯蒂回归的智能机器人语音指令识别方法。采用多维语音传感器,对语音指令信号进行了采集,利用子空间降噪方法对语音指令信号进行了降噪处理。通过频谱特征参数估计方法进行了特征参数估计,构建了智能机器人语音指令的协方差融合模型,融合语音指令参数。采用逻辑斯蒂回归分析方法,结合稀疏性估计,实现了智能机器人语音指令识别。仿真结果表明,本文所提出方法的智能机器人语音指令识别精度较高,能够有效提升智能机器人的控制能力。In order to improve the accuracy of intelligent robot speech instruction recognition,a method of intelligent robot speech instruction recognition based on logistic regression is proposed.The multi-dimensional speech sensor is used to collect the speech instruction signal,and the subspace noise reduction method is used to denoise the speech instruction signal.The feature parameters are estimated by the spectral feature parameter estimation method,and the covariance fusion model of intelligent robot speech instructions is constructed to fuse the speech instruction parameters.Logistic regression analysis method and sparsity estimation are used to realize intelligent robot speech instruction recognition.The simulation results show that the speech instruction recognition accuracy of the intelligent robot based on the proposed method is high and can effectively improve the control ability of the intelligent robot.

关 键 词：智能机器人 逻辑斯蒂回归 语音指令 指令识别 降噪处理 子空间降噪 稀疏性估计 参数融合 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]