关键词: 人工免疫, 故障检测, 无量纲指标, 免疫检测器, 缓变故障

Abstract:

As slowly-varying faults are difficult to detect, a detection method based on the nondimensional immune detector is proposed. In this method, first, a generation algorithm of immune detector based on biological immune principle and evolutionary mechanism is combined with the nondimensional parameters to construct five nondimen- sional immune detectors. Next, in order to detect small variations, the negative selection algorithm is adopted and a proper code length is selected to deal with the extracted data. Then, with the cross-detection using multiple non- dimensional immune detectors and with the integration and fusion of test data, comprehensive fault information is obtained. After an offline training, the multiple nondimensional immune detectors are finally used for the online de- tection of slowly-varying faults. Simulated results and practical application both indicate that the proposed method effectively detects slowly-varying faults on line

Key words: artificial immunity, fault detection, nondimensional parameter, immune detector, slowly-varying fault