Electromagnetic actuator equivalent magnetic field strength modeling method based on Gaussian process regression

Abstract

The invention provides an electromagnetic actuator equivalent magnetic field strength modeling method based on Gaussian process regression. The Gaussian process regression method in the statistical machine learning theory is applied to electromagnetic actuator equivalent magnetic field strength modeling. The electromagnetic actuator equivalent magnetic field strength modeling method based on Gaussian process regression comprises the steps that based on a selected covariance function and an established Gaussian process regression model, hyper-parameters of the covariance function are optimizedby maximizing edge logarithmic likelihood, and an electromagnetic actuator equivalent magnetic field strength model finally established is obtained. The method has the advantages that the electromagnetic actuator equivalent magnetic field strength modeling method based on Gaussian process regression does not depend on a mechanism model, the established model is a conditional probability model, andthe method has high universality and is particularly suitable for realizing electromagnetic actuator equivalent magnetic field strength modeling with complicated magnetic field distribution.

Description

technical field [0001] The invention belongs to the technical field of magnetic levitation control and statistical machine learning, and in particular relates to a modeling method for equivalent magnetic field strength of an electromagnetic actuator based on Gaussian process regression. Background technique [0002] Establishing a high-precision model of the equivalent magnetic field strength of the electromagnetic actuator is the key to realizing the high performance of the electromagnetic actuator. At present, the modeling methods of the equivalent magnetic field strength of electromagnetic actuators include mechanism analysis modeling, experimental modeling and finite element analysis modeling. Mechanism modeling is usually carried out under ideal assumptions, assuming that the magnetic field is uniform and ignoring factors such as leakage flux and edge effects. Due to the restriction of the processing technology level, the structural size and the number of coil turns of...

AI Technical Summary

Problems solved by technology

[0005] However, the current modeling methods for the equivalent magnetic field strength of various electromagnetic actuators are usually parametric models, which are not suitable for solving the problem of modeling the equivalent magnetic field strength of electromagnetic actuators with complex magnetic field distribution.

Images