Identification of Helicopter Dynamics based on Flight Data using Nature Inspired Techniques

摘要

The complexity of helicopter flight dynamics makes modeling and helicoptersystem identification a very difficult task. Most of the traditional techniquesrequire a model structure to be defined apriori and in case of helicopterdynamics, this is difficult due to its complexity and the interplay betweenvarious subsystems.To overcome this difficulty, non-parametric approaches arecommonly adopted for helicopter system identification. Artificial NeuralNetwork are a widely used class of algorithms for non-parametric systemidentification, among them, the Nonlinear Auto Regressive eXogeneous inputnetwork (NARX) model is very popular, but it also necessitates some in depthknowledge regarding the system being modeled. There have been many approachesproposed to circumvent this and yet still retain the advantageouscharacteristics. In this paper we carry out an extensive study of one suchnewly proposed approach using a modified NARX model with a two tiered,externally driven recurrent neural network architecture. This is coupled withan outer optimization routine for evolving the order of the system. Thisgeneric architecture is comprehensively explored to ascertain its usability andcritically asses its potential. Different instantiations of this architecture,based on nature inspired computational techniques (Artificial Bee Colony,Artificial Immune System and Particle Swarm Optimization) are evaluated andcritically compared in this paper. Simulations have been carried out foridentifying the longitudinally uncoupled dynamics. Results of identificationindicate a quite close correlation between the actual and the predictedresponse of the helicopter for all the models.