【摘要】 应用光电传感器和瞬时波形记录系统记录了5种蚊虫的翅振波形。结果显示,每种蚊虫的翅振波形为相似的正弦波。蚊虫的翅振频率虽然彼此间存在交叉,但差异明显。因此,通过建立人工神经网络对蚊虫的种类进行分类识别是可行的。研究中分别以蚊虫翅振频率和翅振波形建立人工神经网络,结果发现以翅振频率为特征值的神经网络的识别准确率高。该网络识别的平均准确率为72.67%,最高为89%。更多还原

【Abstract】 The wingbeat waveforms of five species of mosquitoes were recorded by a photosensor and aWfRer system.Wingbeat waveforms of the result showthat the mosquitoesis analogical sine wave.Al-thoughtheir wingbeat frequencies are overlapped with each other,the diversities of their mean wing-beat frequencies are obvious.Thus,it is possible to construct artificial neural network for classifyingthe wingbeat frequencies of five species of mosquitoes.Artificial neural network classifiers are respec-tively built by wingbeat waveformti me series and wingbeat frequencies.The most accurate classifiertestedis an artificial neural network by using variable of wingbeat frequency.The accuracy is average72.67 %and highest 89 %.更多还原