模糊专家系统在线谱检测中的应用研究

Application of Fuzzy Expert System in Detection of Line-spectrum

本文给出了一种线谱检测的组合方法。首先利用双门限线谱检测算法进行初始线谱检测 ,同时提取频率等线谱特征 ,在此阶段通过选择低门限 ,可以保证在低信噪比下无论信号和背景噪声是否平稳都可获得高的检测概率 ;然后研制了一个简单高效的线谱识别模糊专家系统 ,根据提取出的线谱特征 ,进一步剔除噪声点。分析验证表明 ,由于该算法将线谱检测和特征提取融合在一起 ,不但提取出了用于目标识别的线谱特征 ,而且非常宽容 。

In the paper, an integrated method for detecting line-spectrum is proposed. In the method, firstly, an initial line-spectrum is detected by a double-threshold line-spectrum detec tion algorithm, and at this stage, in order to ensure high detection probabi1ity at a low signal-to-noise ratio an d no mater signal and noise is stable or not, low thresholds are set and features of line-spectrum, such as frequency, signal-to-noise ratio, number of line-spectrum, etc, are extracted. And then, the above features are used by a simple and efficient fuzzy expert system for line-spectrum recognition, which is newly developed for picking out noise points. The analysis shows that the proposed method combines line-spectrum detection and feature extracti on together, it can not only extract many features of line-spectrum used in target recognition, but also is very ro bust, and can automatically detect line-spectrum with high detection probability and low false probability at a l ow signal-to-noise ratio.