The Study Of Digital S-K Filter Smoothing Method

The raw spectral data have irregular glitch when measured by equipment because the radioactivity measurements have the statistical fluctuation characteristics. Even if the nuclear pulse signal which is input have the good smoothness, the pulse amplitude spectrum which is measured also can not overcome this disadvantage. These glitches will affect the calculation of the overall system and also form the system noise. In order to overcome the whole system noise and improve the accuracy, the spectral data must be further smoothed.Early, the multi-point smoothing is generally used to manage the nuclear spectral data smoothing. And in recent years, the emergence of some new smoothing methods, such as FFT transform smoothing, Kalman filter and wavelet transform, etc put the nuclear spectral data smoothing technology to a new stage. However, the three kinds of smoothing method which are mentioned above require the help of other tools (such as Mat lab) for analysis. Under the limitation of existing mathematical description, measurement and processing equipment when the above methods are used to smooth the nuclear spectrum data, there are some disadvantages in theory proved cumbersome, complex calculation methods and heavy workload.In view of this, the paper presents a digital Sallen-Key filter (hereinafter referred to as SK filter) for nuclear spectroscopy data smoothing processing. As early as1955, SK filter which was designed based on discrete components was first proposed by R.P.Sallen and E.L.Key. And it was successfully realized the Gaussian pulse shaping.SK filter can be widely applied to the pulse signal shaping and filter, and it can be obtained the good quality Gaussian waveform and parameters at lower stages. But SK filter which is designed by discrete components have some deficiencies in shaping parameter adjustment and system stability. Therefore, the S-K filter is imperative to be digitized. There are some aspects in numerical analyzing the digital low-pass S-K circuit. First, the voltage transfer formula was created based on KCL (Kirchhoff Current Law) Law, and then the y=f(x) mathematical function was obtained. Second, this mathematical function was solved based on differential numerical method and then the Gaussian shape model was established. Finally, the standard negative exponential signal and the actual nuclear pulse signal were Gaussian shaped in different parameters. The VBA in EXCEL was the software platform. The simulation and the actual circuit forming results are consistent. It also verified the paper presents the theoretical Gaussian shape model is an indisputable right.In actual sample analysis and testing, mixed sample SiSCaTiFe object and multiple samples were be measured by an energy dispersive X-ray fluorescence analyzer. First, the spectral line of spectral raw data (the data which was not processed) was processed by5-point smoothing filter,11-point smoothing filter, FFT filtering, wavelet filtering and digital S-K filtering. The1000-2030address of spectrum line which has not the peak portion was studied in smoothness. The four times polynomial was used to fitting curve. The R2of fitting curve was used to present the smoothness of spectrum line. Finally, there are some study results between S-K filter and other energy resolution in the smoothness from four aspects, that is, energy resolution and calculation workload, relevance and smoothness. Alternatively, paper also gives the testing results for cement, iron ore, copper, and other various samples in different types of detectors. The digital filter S-K have obvious advantages in software algorithm, filtering parameter adjustment, computing workload and smoothness of line spectral line processing performance parameters.