Small unmanned helicopter modeling method based on a hybrid kernel function PSO-LSSVM

The mathematical modeling of a small unmanned helicopter (SUH) with multivariable, highly nonlinear and complex dynamic characteristics is considered. This paper presents a modeling method for SUHs based on a particle swarm optimization least squares support vector machine (PSO-LSSVM) with a hybrid kernel function. The proposed method is based on a least square support vector machine and uses linear weighting of the polynomial kernel function (POLY) and Gaussian kernel function (RBF) to form a hybrid kernel function, and uses a particle swarm optimization algorithm to search for the optimal parameters. Finally, a mathematical model of the longitudinal and lateral passages of a SUH is established. According to the flight test data, the longitudinal and lateral channel models are trained and verified in the hover and low-speed forward flight states of a SUH. The experimental and comparison results demonstrate that the model established via this method has higher prediction accuracy and more accurate prediction results than a model established using a least squares support vector machine with a single kernel function. The identification accuracy of the SUH model is improved effectively.