A Fast Gas Identification Algorithm Based on Frequency Domain Feature Extraction

Abstract

The invention provides a frequency domain feature extraction based rapid gas identification algorithm which is a novel gas identification algorithm. The rapid gas identification algorithm includes that passing gas is subjected to data acquisition and interception through sensors, intercepted data are subjected to Fourier transform to acquire distribution of intercept sequences in a frequency domain, first 20 data values from the 0Hz position are removed to serve as features of the data acquired through single sensors, features of n sensor data are spliced into a 20*n vector according to a certain order, a large number of the vectors are acquired by repeating the previous steps and are given labels of corresponding gas types, the vectors and the labels are mapped to a three-dimensional space by means of an LDA (linear discriminant analysis) algorithm to acquire a mapping matrix, the vectors and labor data acquired through dimensionality reduction are input in an SVM (support vector machine) classifier to perform modeling of the classifier, vectors of the data under the three-dimensional space of the LDA are acquired by multiplying unknown feature vectors with the mapping matrix, and the vectors are input into the established classifier model to perform prediction of the gas types measured currently.

Description

[technical field] [0001] The invention relates to an electronic nose gas detection technology, in particular to a fast gas identification algorithm based on frequency domain feature extraction. [Background technique] [0002] The electronic nose is a new type of instrument that imitates the noses of humans and animals and is used to analyze, identify, and detect complex odors and volatile components. Quickly evaluate the odor, and the characteristics of good repeatability, its application is more and more widely, it uses the gas sensor to identify the volatile odor of the sample. In some harsh environments, it is not suitable for manual monitoring. In order to realize real-time monitoring of gases, it is necessary to set up an electronic nose device. [0003] At present, gas sensors have different extraction methods for gas features. Steady-state feature extraction and transient feature extraction are commonly used, but they are mainly aimed at how to remove redundant infor...

AI Technical Summary

Problems solved by technology

[0010] The steady-state feature extraction method takes a lot of time, and the accuracy of recognition is not as good as that of the present invention

Cause: Steady-state feature extraction takes a lot of time to wait for the reaction between the gas and the sensor to come to an end

Images