摘要
根据粉尘衍射光的角谱特征,提出可适用于井下实时粉尘检测的模式识别算法。为了提高测量精度,对误差进行分析,提出加入修补函数的模式识别反演算法,利用修补函数求出36个修补模式的特征向量,将其预存于单片机中,测量时选择合适的修补模式对最贴近的模式进行局域修补,使修补后的模式与待测尘样更加贴近,同时由于传感器的信号较弱,运用谐波分析的方法对测量信号的组成进行了分析,针对分析结果,采用了切实可行的数字滤波方案,使信号的信噪比得到了较大的提高。
Mine dust has a great effect on miners health. And above all it can cause the dust explosion, So it is radical for mine to detect the real time dust, The dust pattern recognition algorithm was presented according to its diffraction angular distribution, To improve the dust detect precision,the error analysis was made and dust amendment function had been brought up . The eigenvectors of thirty six amendment patterns were worked out and stored in singlechip computer. Suitable amendment pattern improved the optimum pattern in local scope, The variance between the pattern eigenvectors mended and the real time dust signals becomes less. As to the weak signals of the detector, the author used harmonic analysis to analyze the composition of the detected signal. According to the analyzed result, the author adopted digital complex filtration scheme to increase the Signal - to - Noise, Finally the simulation indicates the detection precision and speed have been improved.
出处
《煤矿安全》
CAS
北大核心
2006年第5期7-9,共3页
Safety in Coal Mines
关键词
粉尘传感器
衍射光角谱
模式识别
模式修补
数字滤波
dust sensor
angular distribution
pattern recognition
pattern amendment
digital filter
