Adsorption kinetics based electronic nose data feature extraction method

Abstract

The invention discloses an adsorption kinetics based electronic nose data feature extraction method. The method utilizes an adsorption kinetics equation to fit an electronic nose response curve by corresponding curve fitting algorithm, and an adsorption kinetics equation usually has only two to three undetermined coefficients to be determined by fitting, therefore after fitting, each electronic nose sensor is corresponding to two to three characteristic values, thus reducing the dimension of original electronic nose data. At the same time, a multilayer perceptron neural network is introduced to process the feature extracted data and verify the advantage of the feature extraction method, and a good classification forecast effect can be achieved. The method provided by the invention utilizes the adsorption kinetics equation to analyze the relationship between electronic nose detection time and corresponding signals, reveals the adsorption mechanism of electronic nose sensors, can comprehensively and accurately extract the features of the electronic nose response curve, and maximumly represents the overall feature of the response curve.

Description

technical field [0001] The invention relates to the field of detection of crop diseases and insect pests, in particular to an electronic nose data feature extraction method based on adsorption kinetics. Background technique [0002] So far, the feature extraction methods of odor signals in electronic noses can be mainly divided into three types: 1. Extracting the geometric features of the curve; 2. Extracting the frequency information of the sample data as features; 3. Fitting the signal curve, extracting The parameters of the combined model are used as features. [0003] The first category can be divided into two subcategories: extracting the geometric features of the original curve and extracting the geometric features of the derivative curve of the first order or above. Extracting the geometric features of the sample curve from the original curve is a traditional curve feature extraction method, and the extracted features include the peak value of the curve, the position...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide an electronic nose data feature extraction method based on adsorption kinetics for crop disease and insect pest detection, to solve the problem that the existing feature extraction method cannot reflect the steady-state response and transient response information of the entire response curve, and when the characteristics of the change domain are lost domain information, cannot fully reflect the characteristics of the entire response process, and cannot reflect the technical problems of the actual response model of the detected gas and the electronic nose array signal

Images