A test idea for the stability of the correlation order results of the proximity grey incidence model is proposed for the lack of test methods for model results in the current grey incidence theory. Firstly, the numerical difference of the corresponding observation points between the sequences is used as the simplest element to construct the magnitude change between the sequences to characterize and define the proximity grey incidence degree. Secondly, the basic idea of information entropy is introduced, and the information grey entropy is constructed by using the simplest element of the change of magnitude and size between sequences. The information grey entropy is used to measure the stability degree of the relationship between sequences from the perspective of proximity. Then, the behavior sequence of the research object is divided into different stages on average, and the change value of the proximity grey incidence degree between sequences under the change of different phases is calculated. The size of the tested coefficient finally determines the stability between the models. Finally, several standard grey correlation models are selected to analyze the influencing factors of the main air pollutants in Nanjing from 2016 to 2021. The results of each model are verified and screened according to the stability test coefficient obtained by calculation.