求解航空发动机数学模型的混合智能方法

An intelligent algorithm for solution of nonlinear mathematical model for aeroengine

针对传统求解方法收敛性不强而遗传算法求解效率较低的问题,利用BP神经网络逼近发动机平衡方程的反函数,将求解结果作为Newton-Raphson法的初值,提出了求解模型的混合智能方法。仿真结果表明,该方法可以保证非线性数学模型在整个飞行包线范围内收敛,与遗传算法相比又提高了求解效率。

Current solutions are not always convergent while genetic algorithm is inefficient. Because of this, BP neural networks was used to approach the inverse function of balance equations, and the approximate solution was used as the initial value of Newton-Raphson algorithm, thus an intelligent algorithm is proposed. Simulation results show that this algorithm can make nonlinear mathematical model for aeroengine convergent in the entire flight envelope, and also has higher efficiency compared with genetic algorithm.