Automated method for cosmic ray data analysis and detection of sporadic effects

An automated method for detecting multiscale sporadic effects in data from ground-based neutron monitors is proposed. The method is based on the wavelet transform and neural networks of learning vector quantization type (LVQ neural networks). The choice of Daubechies wavelets and Coiflets at the data preprocessing stage is justified. An algorithm for choosing the “best” approximating wavelet basis in the class of orthogonal functions is proposed. The effectiveness of the method as applied to the detection of small-scale sporadic effects is shown experimentally. The possibility of a numerical implementation of the method for operational use is demonstrated.