Hybrid grid quality optimization method based on probability model

The invention belongs to the field of three-dimensional tetrahedral mesh quality optimization, and relates to a hybrid mesh quality optimization method based on a probability model. The method comprises the following steps: modeling and dispersing a target, generating a three-dimensional tetrahedral mesh, and obtaining mesh vertex information and corresponding topological information; secondly, determining a grid unit quality measurement criterion, establishing an optimization model, and setting an error function form and a target function form; establishing an overall iteration model, and setting the maximum overall iteration times; thirdly, retrieving low-quality grid units, and respectively constructing local optimization domains for vertexes of the low-quality grid units; and iteratively solving the optimal value of the target function through a hybrid grid quality optimization method based on a probability model to complete grid quality optimization. According to the method, the modified gradient descent