RBF neural network optimization method based on improved manta ray foraging optimization algorithm

Abstract

The invention relates to the technical field of neural network optimization, in particular to an RBF neural network optimization method based on an improved manta ray foraging optimization algorithm, which adopts the improved manta ray foraging optimization algorithm to optimize RBF and improve sea clutter prediction and suppression precision. According to the invention, the population is initialized by adopting a dynamic general reverse learning strategy, the population diversity is enriched, and a possible solution of a search space is further mined; secondly, the golden sine algorithm is adopted to improve the prying foraging strategy, the optimization mode is optimized, and the global search and local development capabilities of the algorithm are enhanced; in addition, the optimal solution is disturbed through adaptive probability and mixed variation, the convergence speed of the algorithm and the capability of jumping out of local optimum are improved, and an ideal result is found. The precision and the convergence speed of the improved manta ray foraging optimization algorithm are improved to a certain extent, and the RBF can be helped to find the optimal initial parameters.

Description

technical field [0001] The invention relates to the technical field of neural network optimization, in particular to an RBF neural network optimization method based on an improved manta ray foraging optimization algorithm. Background technique [0002] As a technical means of ocean safety monitoring, high-frequency ground wave radar is widely used in the fields of sea surface target detection and sea state information monitoring because of its advantages of all-weather, over-the-horizon, and low cost. With the increasing complexity of the marine environment and international security situation, it has become an inevitable trend to face threats and challenges from the ocean. At the same time, higher requirements are placed on radar detection performance. However, when the high-frequency ground wave radar detects the target, the first-order sea clutter will be mixed in the radar echo, resulting in missed or false alarms. Therefore, the ability of target detection is closely r...

AI Technical Summary

Problems solved by technology

MRFO uses randomly generated data to initialize the population, so the initial diversity of the population is difficult to guarantee, and the global search ability of the algorithm is limited

Secondly, the position of the somersault strategy in the manta ray foraging optimization algorithm is updated, and the value of the tumbling factor S affects the balance between exploration and development capabilities. Improper values ​​can easily cause the population to gather quickly in the later stage and reduce the possibility of jumping out of the local optimum

Finally, in the late stage of the standard MRFO iteration, manta ray individuals gradually converge, and the phenomenon of "premature" appears, especially when it comes to high-dimensional problems, the population often converges to an optimal solution and lingers around it and is not easy to jump out, which leads to the optimization of the algorithm The effect becomes worse

Images