前向神经网络快速学习算法在发动机模型辨识中的应用
A Rapid Learning Algorithm of Feedforward Neural Network for Identification of Aeroengine Model
-
摘要: 介绍一种采用前向神经网络辨识发动机非线性模型的快速学习方法,与现有同类方法相比,该方法采用矩阵的QR分解求解线性方程组,解决了维数过高的矩阵直接求逆带来的解失真问题,并把这种方法应用于航空发动机的非线性模型辨识。算例仿真验证了算法的可行性。Abstract: To build a precise model is a key technique to aeroengine adaptive control.A rapid learning algorithm of feedforward neural network for identification of aeroengine model is presented in the paper.By contrast with similar algorithms,the present approach employs the decomposition of Matrix QR to solve a set of linear equations.As a result,inversing of huge matrix inverse and fuzzy solutions can be avoided.The algorithm was applied to the aeroengine nonlinear model identification,and validated by the simulation results of the calculation samples.
-
Key words:
- aerospace propulsion system /
- neural networks /
- learning algorithm /
- engine model /
- model identification
点击查看大图
计量
- 文章访问数: 1425
- HTML浏览量: 3
- PDF量: 459
- 被引次数: 0