Drift suppression method for electronic nose of domain self-adaptive extreme learning machine (ELM) based on domain correction

Abstract

The invention discloses a drift suppression method for an electronic nose of a domain self-adaptive extreme learning machine (ELM) based on domain correction. The drift suppression method comprises the following steps: mapping source domain data and target domain data which are inconsistent in data distribution into a high-dimensional Hilbert space from the point of view of data distribution, andminimizing domain distance between a source domain and a target domain in the space; meanwhile, preserving the data properties of an original source domain and the target domain to the maximum extent,and obtaining the source domain data and the target domain data after domain correction, thus suppressing drift from the data level; and incorporating migration samples and unlabeled samples in the target domain into ELM for learning to obtain the domain self-adaptive ELM and improve the robustness of a predictive model from the decision level. The drift suppression method disclosed by the invention has the advantages that data distribution is adjusted without samples being added; and in addition, the unlabeled samples in the target domain are incorporated into the learning of a classifier, so that the drift is suppressed from two levels, i.e., the data level and the decision level, and the robustness of the model is improved.

Description

technical field [0001] The invention belongs to the field of odor recognition of electronic noses, and relates to a domain correction-based domain adaptive extreme learning machine electronic nose drift suppression method. Background technique [0002] The electronic nose is a system composed of a set of sensor arrays combined with corresponding pattern recognition algorithms, which can identify gases. When the gas to be detected enters the detection chamber, the gas sensor will generate a transient response signal to it, and use the pattern recognition algorithm to identify the gas according to the sensor response. Electronic nose has experienced rapid development in the past two decades and has been used in the detection of perfume, fruit, wine, tea and coffee. [0003] In odor recognition, many pattern recognition algorithms for electronic nose classification and regression have been proposed. Among them, the neural network is a very important way, including two learnin...

AI Technical Summary

Problems solved by technology

In actual application scenarios, due to the aging of sensor sensitive materials, the pattern recognition model trained with data without sensor drift cannot be directly applied to the electronic nose system that has experienced sensor drift.

Images