Abstract: In terms of collision probability calculation in space debris threat assessment, the traditional calculation methods are divided into whether to consider the uncertainty of spacecraft and space debris velocity, and the variances of the initial status of two space objects considered in both cases are small. It will produce a large error in the case of large uncertainty of two space objects orbit determination. In order to calculate the collision probability accurately with the situation of nonlinear relative motion process of two space objects and large initial uncertainty, a collision probability calculation method based on adaptive Gaussian mixture model is proposed, which uses the adaptive Gaussian mixture model to approximate the uncertainty propagation process of spacecraft and space debris orbits, and calculates the collision probability according to the propagation results. The simulation results in the two body environment show that the method achieves well calculation accuracy when the relative motion process is nonlinear, and the calculation accuracy is better than other collision probability calculation methods when the initial uncertainty of the object is large.

Key words: space debris, error propagation, Gaussian Mixture model, unscented transformation, collision probability