论文部分内容阅读
提出了一种基于导数分析与典型特征选取的信号识别新算法。首先,分析信号导数与频率的关系,得到信号的时频特性;其次,根据时频分布的特点,选取具有代表性的典型特征,以降低特征维度,减少识别时间;最后,应用概率神经网络(PNN)对典型特征进行学习和分类。采用4种扰动信号对本文算法进行实验验证,平均正确识别率达95.7%,且识别时间小于0.23s。实验结果表明,本文算法能够快速准确地识别扰动类型,为Mech-Zehnder(M-Z)干涉型周界系统的模式识别提供一种科学可靠的方法。
A new signal recognition algorithm based on derivative analysis and typical feature selection is proposed. Firstly, the relationship between signal derivative and frequency is analyzed, and the time-frequency characteristics of the signal are obtained. Secondly, according to the characteristics of time-frequency distribution, typical representative features are selected to reduce the feature dimension and reduce the recognition time. Finally, using the probabilistic neural network PNN) Learn and classify typical features. The proposed algorithm is validated by four kinds of disturbing signals. The average correct recognition rate is 95.7% and the recognition time is less than 0.23s. Experimental results show that the proposed algorithm can quickly and accurately identify the types of perturbations and provide a scientific and reliable method for pattern recognition of the Mech-Zehnder (M-Z) interferometric perimeter system.