| Ocean wave simulation technology has been widely used in many fields,such as engineering,science,virtual reality,commerce and so on.With the gradual deepening of ocean wave simulation research and the rapid development of computer software and hardware,people put forward higher requirements for real-time and realistic ocean wave simulation.Since the development of radar technology,the field of sea surface remote sensing and monitoring has been paid more and more attention.Therefore,the modeling of sea surface geometry and electromagnetic scattering,and the prediction of sea clutter are of great significance.In this paper,geometric modeling and real-time scene simulation of sea surface,electromagnetic scattering modeling and statistical analysis of sea surface,simulation and prediction of sea clutter and so on are studied.The main contents include the following points: Firstly,in order to ensure the authenticity and real-time performance of the ocean rendering simulation system,and to improve the rendering efficiency,a fast Fourier transform(FFT)modeling method based on spectral analysis is studied in this paper.Firstly,the influence of wind region on wave height is considered,and the Phillips spectrum is the sea wave spectrum of infinite wind region.Based on the sea surface model of Phillips spectrum,a calculation model based on JONSWAP spectrum is established,so that it can fully reflect the characteristics of the wind and the impact of the wind on the waves.Then,based on JONSWAP spectral model,the corresponding double-sided angular distribution function is proposed to reflect the real three-dimensional sea surface.Finally,the projection mesh algorithm is used to speed up the rendering efficiency,and in order to present the geometric model of time-varying sea surface,OpenGL is used to draw wave height field.The results show that the simulation method presented in this paper is suitable for real-time simulation of large scene sea surface.Secondly,after studying the geometric modeling of the sea surface,it is necessary to carry on the sea surface electromagnetic modeling on the basis of the geometric model.First,according to the rough surface scattering mechanism,the Semi-deterministic facet scattering model and the Small slope approximation method are adopted to calculate the electromagnetic scattering of the sea surface respectively,the scattering coefficient distribution of the total field and the cell field is analyzed,and the electromagnetic scattering result of the cell field is counted,and the amplitude distribution characteristic is analyzed.Then,the probability density function and parameter estimation method for describing the four theoretical distributions of sea clutter are studied.The estimated theoretical curve is fitted with the actual statistical scattering coefficient amplitude distribution model,and the goodness-of-fit test is carried out.Thus,the best fitted PDF model is logarithmic normal distribution.Finally,the power spectrum estimation method of clutter is studied,and the power spectrum estimation of sea clutter data is carried out.Finally,in order to ensure that the sea clutter simulation results are most close to the actual sea clutter data,on the basis of obtaining the PDF model with the best fitting degree to the electromagnetic scattering distribution and the appropriate parameters,the sea clutter simulation related to the specified distribution is further studied.Firstly,the two-dimensional sea clutter is simulated by Monte Carlo method,and the simulated clutter sequence is transformed by zero-memory nonlinear transformation,and the correlated lognormal distribution clutter sequence is simulated.Then,the sea clutter data are predicted based on AR model and neural network theory.Finally,a simulation experiment is designed,and compare the prediction effects of the two simulation prediction methods,and the simulation results are compared with the actual results and the error calculation is carried out.The results show that the prediction result of sea clutter based on neural network model is more accurate than that of linear prediction based on AR model,and the error can be reduced by setting the target value. |