RBF neural network optimization method based on improved whale algorithm

Abstract

The invention belongs to the technical field of neural network optimization, and particularly relates to an RBF neural network optimization method based on an improved whale algorithm, and the methodcomprises the steps: using improved whale algorithm for searching an optimal initial parameter of an RBF neural network, building a sea clutter prediction model through a training network, and carrying out prediction and inhibition of sea clutters of an adjacent unit; dynamically calculating the fitness mean value of each generation of population in the whale algorithm iteration process, setting the fitness threshold of the next generation of population, dividing the whole population into a high-quality whale sub-population and a non-high-quality whale sub-population, and enabling the high-quality whale sub-population and the non-high-quality whale sub-population to be close to the global optimum at different step lengths; besides, when contraction updating is executed, the idea of substance exchange is introduced so that newly generated particles can be globally recognized and can be stably searched for in the globally optimal direction, the improved whale algorithm has global and local search capacity in the iteration process, and the convergence speed and precision are improved.

Description

technical field [0001] The invention belongs to the technical field of neural network optimization, in particular to an RBF neural network optimization method based on an improved whale algorithm. Background technique [0002] High-frequency radar emits high-frequency electromagnetic waves, and short waves can diffract and propagate along the ocean surface, realizing all-weather and beyond-horizon monitoring of the ocean. At present, high-frequency radar has been widely used in many fields such as maritime early warning, maritime resource detection, and maritime rescue. The innovative research on the high-frequency radar system of RANGER, an EU maritime traffic monitoring project, has enabled high-frequency radar to be applied better and better in more and more fields. However, when high-frequency radar detects sea targets, the echoes are often mixed with a large number of interference echoes, and the main interference component is sea clutter, which appears relatively larg...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the technical problems that the existing whale algorithm lacks flexibility, ignores global information when updating, converges slowly, and is easy to fall into local optimum. On this basis, a RBF neural network optimization based on the improved whale algorithm is proposed method

Aiming at the problem that particles lack the ability to combine global information, this invention introduces the idea of ​​multiverse algorithm, which replaces the original method of randomly looking for particles, so that the update benchmark has global information and ensures that particles are updated to the global optimal direction

Images