Estimation of a five-parameter JONSWAP spectra with an improved particle swarm optimization

Abstract

An improved particle swarm optimization (PSO) algorithm is presented to estimate a five-parameter JONSWAP spectrum for modeling of certain wave states, including normal events and extreme events at deep-water and shallow-water stations, respectively. The choices of spectral parameters (energy scale coefficient α, peak enhancement coefficient γ, and shape parameter σ) through vast amounts of in situ observations and spectral nonlinear factors are of significant importance for safety of marine structures and activities. Several experiments are carried out for five JONSWAP spectral parameters (α, γa, γb, σa, and σb) using relatively unimpaired and long series significant wave height, dominant wave period, average wave period, maximal spectral value, and ratio of dominant period and average period. And the choices of γ from γa and γb, as well as σ from σa and σb depend on the relationships between current frequency and the peak frequency. Two improvements of PSO algorithm are made, one is the fitness function, another is that the population average replacing particle’s individual optimum in the updating step. Well fitted spectra and statistic results show that the improved PSO model is more suitable to be applied to estimate the wave spectrum at a given record time. This heuristic model can be used without restrictions of water depth and sea state with its intuitive background, easy-programming and portable feature.