%A %T Laser Induced Fluorescence Spectrum Analysis of Water Inrush in Coal Mine Based on FCM %0 Journal Article %D 2018 %J SPECTROSCOPY AND SPECTRAL ANALYSIS %R 10.3964/j.issn.1000-0593(2018)05-1572-05 %P 1572-1576 %V 38 %N 05 %U {https://www.gpxygpfx.com/CN/abstract/article_9831.shtml} %8 2018-05-01 %X Rapid identification of mine water inrush types in coal mine is of great significance for prevention and control. In view of the fact that traditional chemical method of water source identification is time-consuming and other problems, we put forward the fuzzy C mean clustering (FCM) algorithm and multidimensional scaling analysis (MDS) for laser induced fluorescence spectrum identification of mine water inrush and the new ideas.Because the FCM algorithm has been successfully used in spectral analysis and pattern recognition, and laser spectroscopy with fast response time, high sensitivity, less interference, the fluorescence spectra of the real-time data acquisition of water, the use of FCM and MDS on the spectral data analysis can identify sample types. A mine in east area of goaf water and Ordovician limestone water were mixed in proportion to get a total of 7 samples (each sample and 20 samples) as experimental materials, we used laser of 405nm to send laser into the measured water body, collected a total of 140 groups of fluorescence spectral data, and then selected the appropriate wavelength interval analysis. 105 sets of spectral data of each group were used as the training set, and the other 35 groups were used as the test set. We Used MDS to establish the model of five kinds of different water samples, and then used the FCM algorithm in cluster analysis to get the cluster center of the five kinds of water samples, finally useed the cluster center to test the test set. The experimental results show that there are dramatic difference between the spectra of different samples, we selected the appropriate wavelength range of spectral data, the dimension at 2 under MDS, and classfied the water samples by using FCM algorithm, finally the accuracy rate of all 140 samples reaches 100%.