无人机传感器的故障诊断和容错控制是一项关键技术,为了能够实时监测传感器的运行状态、快速定位故障和控制重构,文章采用BP神经网络设计了一种由主、从神经网络构成的无人机传感器故障诊断算法,其中主网络用于传感器的故障检测,从网络完成对故障的识别;该算法减少了故障诊断运算量,提高了故障诊断的实时性;通过仿真研究表明了该算法可以有效地检测、识别出故障,并能给出故障传感器估计值用于容错控制。
The paper presents the UAV failure diagnosis method for the sensor with the neural network algorithm, which is composed of main network and local network. Based on the BP algorithm, a new kind of tolerant fault detection method is presented and the fault of sensors can be detected on line with this method. A simulation example is given for illustration.