On the basis of third-order Kolmogorov-Gabor polynomial

extract principal component variables structural linear test of auxiliary regression and F statistic

collinearity problems of auxiliary regression is well solved and improve flexibility and freedom of test. Simulation experiment shows that compared with the test for some specific nonlinear function form

the test does not need to make any assumption

and has better reliability. Under different experimental environment

empirical research shows that in wireless sensor networks（ WSNs）

data sending rate and packet loss rate have equilibrium relationship

and the relationship has obvious nonlinear characteristics. Different experimental environment will affect the result

among them

data transmission reliability after routing nodes transfer is shown to be reduced.