Self-gravity calculation method based on polynomial regression

The invention relates to a self-gravity calculation method based on polynomial regression, and the method comprises the following steps: S1, selecting the displacement of two test masses and the change of an included angle of a mobile optical assembly (MOSA) as feature input, and selecting the three-axis acceleration of the two test masses as target output; s2, building a spacecraft finite element model, and changing feature input by using a finite element modeling analysis method based on a gravitation formula to calculate and generate sample data; s3, adopting a polynomial regression algorithm to perform self-gravity function fitting calculation on feature input and target output on the basis of sample data, performing linear normalization on input data of a polynomial corresponding dimension, performing reverse normalization on output data, and adopting cross validation to evaluate model performance in a regression process; and S4, quickly and effectively calculating the self-gravity changed due to the cha