Random drift particle swarm optimization method having von Neumann structure

Abstract

The invention discloses a random drift particle swarm optimization method having a von Neumann structure. For biological network modeling, such as synthetic gene loop modeling, a differential equation model of a biological system needs to be established, and model parameters are inferred according to observed values; and in the parameter calculating process, an appropriate objective function and restraint conditions need to be defined first, and meanwhile, support of an optimization algorithm is needed. Parameter estimation of a differential equation is equivalent to a problem of nonlinear programming having differential-algebraic constraints, and an appropriate parameter set is found to fit measurement data; for the boundedness that optimization calculation in the parameter estimation process is easy to run into partial optimization, the von Neumann structure is introduced to the random drift particle swarm optimization algorithm to enhance global search capability; and the improved algorithm is improved for the boundedness of a full-mesh global topological structure, and adopts a local topological structure to serve as an inter-particle information sharing mode, so that global search capability of the algorithm is enhanced, and high-reliability model parameters can be given.

Description

technical field [0001] The invention relates to the field of biological network identification and parameter estimation, in particular to a random drift particle swarm optimization method with von Neumann structure. Background technique [0002] The rapid development of biotechnology has brought continuous impetus to fields such as food and medicine, environmental governance, and new energy. Governments of various countries have gradually increased their attention and investment in biotechnology represented by synthetic biology. The purpose of synthetic biology is to establish artificial biological systems and make them function according to specific laws, from gene fragments, DNA molecules, gene regulatory networks to artificial design and synthesis of cells. Collins et al. of Boston University have developed a "Toggle Switch" device, and the selected cell functions can be switched on and off at will. The synthetic gene oscillation circuit developed by Elowitz et al. of th...

AI Technical Summary

Problems solved by technology

[0005] In order to improve the reliability of biological network parameter estimation and solve the local optimal problem faced by the existing particle swarm optimization algorithm, the present invention provides a random drift particle swarm optimization method with von Neumann structure, which can be used as parameter reasoning The algorithm handles biological system modeling problems such as gene networks better, and improves the global search ability in the parameter optimization process

Images