【摘要】 对于高阶复杂的动态系统,实时故障检测与隔离对保证和提高系统的精度和可靠性尤为重要.本文将状态x~2检验和Fuzzy ART神经网络相结合,提出一种新的动态系统故障检测方法.该方法通过实时监测动态系统的状态分量,不仅能实时地确定动态系统量测值的有效性,而且能将故障的特征值提取出来,并利用Fuzzy ART神经网络实时确定故障的类别,从而实现动态系统的故障定位与隔离.将该方法应用于INS/GPS组合系统的故障检测中,仿真结果证明这种方法效果很好.更多还原

【Abstract】 To maintain and improve the level of precision and reliability in complicated high-dimensional dynamic systems, it is necessary and important that failures should be detected, located and isolated promptly. A novel fault detection method using State Chi-square Test and Fuzzy ART neural network is put forward in this paper. Via monitoring components of the state vector, this method can not only confirm the validity of measurement, but also extract the fault characteristics, identify the fault pattern using Fuzzy ART neural network in real time, and then implement fault locating and isolating in dynamic systems. This method is adopted to detect faults in INS/ GPS integrated system. Simulation result proves that its effect is satisfactory. It has many advantages such as on-line self-organizing, self-adaptive clustering, fast abiogenetic failure catching, rapid convergence and realtime diagnasis.更多还原