Aircraft Engine Control Based On Multiple-Model Adaptive Dynamic Programming

Aeroengine has a characteristics of high complexity and high precision,and the research on aeroengine has became the frontier of the aeronautics and astronautics field.With the development of aeroengine and corresponding control theory,the requirement of the control system is also improved.Based on the data-driven approaches,we mainly focus on the optimal control of aeroengine,combining with the multiple model method.Firstly,a brief introduction of aeroengine and control theory is presented.Then the modeling of aeroengine control system is proposed with a fixed equilibrium point for further research.And the data from the simulation model are used for the following research.Considering the curse of dimensionality,the adaptive dynamic programming(ADP)approach is applied as the key iterative algorithm for the optimal control problem of nonlinear system.This method can obtain the optimal control law of aeroengine by approximating and solving the HJB equation of the system,so that ADP method is suitable for the system with strong nonlinearity and high complexityConsidering multiple working modes of aeroengine,we combine ADP algorithm and multiple model method to improve dynamic performance.Based on the above,we put forward a method of second level adaptation dynamic programming.Using convex hull property,we present the system linearly,accelerate convergence speed and solve the problem of sensitivity to initial condition.A turbo-fan engine is selected as the object of research,and the servo signal and two steps of normalization are used to avoid the introduction of noise and reduce damage to the system.Through the program,the HJB equation of the system is approximated and iterated.System’s working modes are represented by data and we use switching and indirect weighting way to accelerate convergence speed.The experimental results show that the proposed algorithm can be applied to the engine control system with strong nonlinearity.