摘要
传感器输出数据的可靠性是保障其发挥作用的前提。鉴于主元分析(PCA)处理非线性问题能力的不足,提出了一种基于核主元分析(KPCA)的非线性微惯性测量单元(MIMU)传感器故障诊断方法。通过构建KPCA模型预测误差和传感器变量贡献量变化值实现故障监测与定位,为了减少参数选择的盲目性和建模工作量,利用模糊推理改进的自适应遗传算法(AGA)对KPCA核函数参数进行自动优选。仿真结果表明,所提出的方法对MIMU传感器具有良好的故障监测与识别能力,相比于常规KPCA,故障监测的平均准确率提高了18.44%,证明了方法的有效性和优势。
The reliability of sensor output data is the premise to ensure its function. In view of the weakness of principle component analysis(PCA) in dealing with nonlinear problems, a fault diagnosis method for nonlinear micro inertial measurement unit(MIMU) sensors based on kernel principal component analysis(KPCA) is proposed. By constructing the KPCA model, fault monitoring and location can be realized by predicting errors and variable values of sensor contribution. In order to reduce the blindness of parameter selection and modeling workload, the adaptive genetic algorithm(AGA) improved by fuzzy inference is applied to optimize the KPCA kernel function parameters automatically. The simulation results show that the proposed method has good fault detection and identification ability for MIMU sensors. Compared to conventional KPCA, the average accuracy of fault detection is increased by 18.44%, which proves the effectiveness and advantages of the method.
作者
高运广
蔡艳平
盛安
GAO Yunguang;CAI Yanping;SHENG An(Engineering Machinery College,Hunan Sany Polytechnic College,Changsha 410129,China;Combat Support College,Rocket Force University of Engineering,Xi’an 710025,China)
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2022年第6期835-840,共6页
Journal of Chinese Inertial Technology
基金
陕西省自然科学基金(2019JQ-712)。
关键词
微惯性测量单元
传感器
故障诊断
核主元分析
自适应遗传算法
micro inertial measurement unit
sensor
fault diagnosis
kernel principal component analysis
adaptive genetic algorithm
