基于RBF神经网络的FADS系统及其算法研究

以典型的十字形布局的大气数据传感系统及其跨声速应用为研究对象,基于RBF神经网络,设计了新的FADS算法和故障检测处理方法。将测压点按不同功能进行精细的划分和组合,形成更加精简、目的性更强且相互独立的RBF网络处理子模块,利用各子网络模块提供的冗余特性,使用基于故障特征向量表的方法,实施简单而有效的故障检测与处理。仿真验证表明,迎角与侧滑角的测量误差不大于0.5°,且故障检测是有效的。

Research of Algorithms of Flush Airdata Sensing System Based on RBF Neural Network

Based on RBF neural network,a new estimating algorithm,fault detection and diagnosis algorithm have been designed for a typical cruciform pattern flush air data sensing system,which is used for transonic speed air data parameters measurement.All pressure orifices are divided and combined into small modules by different function,forming more concise,more purposeful and independent RBF NN processor sub-module.Because of the redundancy of each sub-module,a simple and effective fault detection and diagnosis algorithm has been taking,by using the method of standard failure indicator vector scheme.Simulation results show that measurement error of angle of attack and sideslip are less than 0.5 degree,and the fault detection is effective.