A new approach to robust fault detection of model-unknown nonlinear systems

作者：Su Shengchao^*; Zhang Wei

来源：International Journal of Control and Automation, 2014, 7(8): 181-192.

DOI：10.14257/ijca.2014.7.8.16

• Confidence levels
• Engineering systems
• Least squares support vector regression
• Model of systems
• Model-unknown
• On-line fault detection
• Robust
• Robust fault detection

Most of actual engineering systems have the characteristics such as nonlinearity, difficult modeling, and high cost to experiment with fault. So fault detection in these systems is difficult. In this paper, a new approach to robust fault detection is proposed for model-unknown nonlinear systems. Firstly, a general framework for fault detection is discussed. This framework is capable of associating each detection result with a confidence level. An online learning least squares support vector regression algorithm is then used to design the framework. Meanwhile, robust strategy and forgetting strategy are added to the algorithm. Accordingly, a robust online fault detection method is developed with less amount of calculation. Finally, the proposed method is applied to fault detection of model-unknown fighter F-16. The simulation results show that the proposed method can detect fault quickly without the model of system or any prior knowledge of fault.

• 出版日期2014
• 单位上海大学; 东华大学