Study On γ Spectral Data Analysis Methods

NaI (Tl)γspectrometer has many advantages, such as high efficiency of the detection, low cost device, easy to use and easy maintenance, It is has been widely used in the radiation monitor. But it also has the shortcoming, such as low resolution, especially in the low radionuclide content, many kinds of backgrounds interfere measurement conditions,many peaks in low energy areas cause overlaps, all of which lead to the uncertain results ofγspectral analysis.In this paper, through comparing the current algorithm ofγspectrum solution based on spectrum for NaI (Tl)γspectrum of the decomposition, the spectrum smoothing, peak searching, peak region determination, background reduction, energy peak fitting function, net peak area solving and other related problem are studied and verified.1. The spectral data smoothing: B-spline smoothing method proposed to reduce the statistical fluctuations, and compared with other smoothing methods. B-spline smoothing method that has local invariance, continuity, convexity of the other properties, can effectively eliminate most of the statistical fluctuation, optimal spectral shape, and strong adaptability, to better reflect the primaryγspectral important features of the peak, to a large extent reduce the human error.2. Finding peak and peak region determination: With peak shape calibration, energy calibration function, the use of joint finding peak method (differential method, slope-width finding peak method, covariance method) to find the peak of the spectrum, can improve the ability to distinguish peaks, recognition rate of radionuclide. Changing the traditional method of determining peak region by setting the probability of peak interval (99.7%) to determine peak region boundary for ascertain the effective range of counts, to improve the accuracy of calculating peak area.3. Peak area determination: Based on the principles of the formation mechanism ofγspectral peaks, fitting the peaks with Gaussian functions and the SNIP curve background function, and the overlapping peaks were decomposed through changing the same traditional background model, each sub-peaks are fitted with a separate function to decompose. This approach will have a better fitting effect. The analysis of the measured spectral data showed that the spectrum data processing method used in this paper, to better complete the qualitative and quantitative analysis of spectral data. B-spline smoothing, the use of joint finding peak method can effectively reduce the impact of fluctuations in statistical data, high resolution finding peak, the weak peak and false peak finding leakage probability is small, and the adjustment of each independent peak region boundary and the fitting function correction, making the processing precision of peak area improved to a certain extent.