Sea radar image sea surface wind speed inversion method based on RBF neural network

Abstract

The invention discloses a sea radar image sea surface wind speed inversion method based on an RBF neural network, and belongs to the field of sea surface wind speed inversion by using a remote sensingmeans. The invention discloses a sea radar image sea surface wind speed inversion method based on an RBF neural network. The method comprises four parts of sea radar image data preprocessing, RBF neural network input layer construction, RBF neural network model determination and sea surface wind speed information extraction. The sea surface wind speed inversion process is completed based on a model obtained by training a single hidden layer RBF neural network, and an RBF neural network input layer sample is constructed by adopting a normalization result of a sea surface wind field energy spectrum, sensor information and sea condition information; meanwhile, an application subtraction clustering algorithm is proposed, a neural network is determined according to the density index of an input sample and clustering judgment conditions to determine the number of hidden layer units and the center and expansion constant of a primary function, and a network output layer connection weight is obtained by using recursive least squares.

Description

technical field [0001] The invention relates to the technical field of sea surface wind speed remote sensing, in particular to a method for retrieving sea surface wind speed from a marine radar image based on an RBF neural network. Background technique [0002] The sea surface wind field is of great significance to navigation operations and ocean dynamics research, and is an important means of understanding ocean changes and predicting risks. Sea surface wind field information mainly includes two aspects of sea surface wind direction and sea surface wind speed, and the present invention is used to invert the sea surface wind speed information. [0003] There are two main types of traditional methods for extracting sea surface wind speed information: site-based observation and remote sensing inversion. Station observations usually use anemometers on board or on the shore, but due to the turbulence effect caused by the hull structure, buildings on board or shore foundation, t...

AI Technical Summary

Problems solved by technology

However, the BP neural network has inherent deficiencies, mainly manifested in the problems that it is easy to be limited to local minimum values, the convergence speed of the learning process is slow, and the hidden layer and the number of nodes in the hidden layer are difficult to determine, which makes the network unsuitable for some radar data. Defects of low performance accuracy and calculation time that cannot meet engineering requirements

Images