Mechanical arm error compensation method based on improved RBF neural network

The invention belongs to the technical field of mechanical arm precision control, and relates to a mechanical arm error compensation method based on an improved RBF neural network, which mainly comprises two parts of geometric error identification based on a preprocessing sequence quadratic programming algorithm and non-geometric error identification based on a particle swarm optimization RBF neural network. In a geometric error identification compensation stage, a mechanical arm high-dimensional solution space is reduced to a non-convex low-dimensional subspace through preprocessing, and then sequential quadratic programming is designed to solve geometric compensation data; in a non-geometric error identification compensation stage, designing an RBF neural network algorithm based on particle swarm optimization to adaptively predict non-geometric error compensation data; and the positioning error of the mechanical arm is comprehensively compensated by combining geometric and non-geometric error compensation d