基于Fast ICA算法的2种储粮害虫活动声信号识别

Active acoustic signals recognition of two kinds of stored grain pests based on Fast ICA

利用快速独立分量分析(fast independent component analysis,Fast ICA)算法,对混有高斯噪声的2种储粮害虫玉米象Sitophilus zeamais和赤拟谷盗Tribolium castaneum的活动声信号进行去噪,并使用Fast ICA算法识别和分离了2种储粮害虫爬行与翻身的4种活动声信号,证明了使用Fast ICA算法识别混合信号中每种害虫声信号的有效性和准确性。

Using fast independent component analysis algorithm,active acoustic signals mixed with gaussian noise of Sitophilus zeamais and Tribolium castaneum known as stored grain pests,were de-noised.Then Fast ICA algorithm were used to recognize and separate acoustic signals of four kinds of active acoustic signals,such as creeping and vibratory signals of Sitophilus zeamais and Tribolium castaneum.The results demonstrate the validity and accuracy to recognize each pest′s acoustic signal from mixed signals by Fast ICA algorithm.