Optimal estimation of parameters of dynamical systems by neural network collocation method

摘要: Abstract In this paper we propose a new method to estimate parameters of dynamical system from observation data on the basis neural network collocation method. We construct an object function consisting squared residuals model equations at points and deviations observations their corresponding computed values. The is then trained by optimizing function. proposed demonstrated performing several numerical experiments for optimal estimates two different nonlinear systems. Firstly, consider weakly highly cases Lorenz apply optimum values under various conditions. Then it one-dimensional oscillator with damping restoring terms representing ship roll motion Satisfactory results have been obtained both problems.