An edible oil transverse relaxation signal feature extraction method based on a 2D-CNN

Abstract

The invention provides an edible oil transverse relaxation signal feature extraction method based on a 2D-CNN. The method comprises the following steps: S1, reading CPMG original data, and carrying out inversion to obtain inversion data; S2, respectively preprocessing the CPMG original data and the inversion data; S3, drawing a transverse relaxation attenuation curve and a multi-component relaxation spectrum; S4, constructing a two-dimensional convolutional neural network; S5, extracting data from the self-transverse relaxation attenuation curve and the multi-component relaxation spectrum to form a training set and a test set; S6, inputting the training set into a two-dimensional convolutional neural network; S7, inputting the test set into the trained two-dimensional convolutional neuralnetwork; And S8, obtaining a classification result. 2D-based method of the invention According to the CNN edible oil transverse relaxation signal feature extraction method, edible oil transverse relaxation is directly subjected to feature extraction and classification through the two-dimensional convolutional neural network model, ineffective features can be effectively prevented from being generated, and the accuracy of a classification result is guaranteed.

Description

technical field [0001] The invention relates to the field of deep learning and nuclear magnetic resonance signal processing, in particular to a 2D-CNN-based feature extraction method for lateral relaxation signals of edible oil. Background technique [0002] The safety of edible oil has become a common topic of concern to the common people. However, due to the lack of an accurate, simple and rapid detection method for the authenticity of edible oils, the quality and safety of edible oils still presents a state of repeated prohibitions. Traditional and effective detection methods are not only expensive, but also complicated to operate and maintain. Therefore, the establishment of an accurate, simple and rapid detection method to identify the authenticity of edible oil provides technical support for law enforcement officers, which will help speed up the formulation and improvement of national standards for edible oil, prevent the occurrence of edible oil quality and safety is...

AI Technical Summary

Problems solved by technology

However, in the inversion process, invalid features may be generated due to different inversion parameters and other reasons, which may affect the classification results.

Images