Abstract: As an important detection method, active fluorescence detection uses fluorescence lifetime as the characteristic parameter of fluorescence detection, which can solve the problem that the fluorescence intensity is easily affected by external environmental factors to a certain extent. Based on the principle of the gated-detection method for measuring fluorescence life, the nonlinear least square regress combined with fluorescence lifetime decay curve is used to fit the fluorescence lifetime decay function to extract the average fluorescence lifetime parameters, and the two-dimensional spatial distribution of fluorescence substances is drawn from the fluorescence life map. Furthermore, a method of oil types recognition using fluorescence lifetime parameter as feature vector is proposed and experimentally verified. The experimental results show that the probability of the pixel point fluorescence lifetime falling into the confidence interval in the excitation region is more than 68% by using the fluorescence average life as a parameter, and the recognition probability is over 77% by using the support vector machine for oil types recognition. It seems that it is feasible and has a good recognition rate to identify oil species by using fluorescence lifetime parameter, and at the same time, less training samples are required for the method combined with support vector machine. Therefore, the oil recognition method based on fluorescence lifetime decay curve combined with support vector machine will provide another reference for oil types recognition research in the field of environmental pollution.

Key words: active fluorescence detection, fluorescence lifetime, time-gated detection, support vector machine