A fault diagnosis method for gas turbines based on improved data preprocessing and an optimization deep belief network. (25th October 2019)

• Record Type: Journal Article
• Authors: Yan, Li-Ping; Dong, Xue-Zhi; Wang, Tao; Gao, Qing; Tan, Chun-Qing; Zeng, De-Tang; Zhang, Hua-liang; Chen, Hai-sheng
• Is Part Of: Measurement science & technology. Volume 31:Number 1(2020)
• Publication Date: 2019-10-25
• Subjects: fault diagnosis -- data preprocessing -- deep belief network -- gas turbine; Physical measurements -- Periodicals; Scientific apparatus and instruments -- Periodicals; Equipment and Supplies -- Periodicals; Science -- instrumentation -- Periodicals; Technology -- instrumentation -- Periodicals; Mesures physiques -- Périodiques; Physical measurements; Scientific apparatus and instruments; Periodicals; 502.87
• DOI: 10.1088/1361-6501/ab3862
• Languages: English
• ISSNs: 0957-0233

Abstract

Abstract: A classifier trained by a normalized simulation parameter could not identify an actual fault. In order to solve this problem, improved data preprocessing is proposed which normalizes the deviation of the simulation parameter, thus making preprocessed simulation data more accurate at revealing the performance of an actual gas turbine. Furthermore, an optimization deep belief network (DBN) based on a genetic algorithm is developed, which shows a good classification ability. The superiority of these two methods is validated respectively by a three-shaft gas turbine platform. It has also been found that based on the DBN optimization method, adding outlet temperature parameter T 3 to a high-pressure compressor can significantly improve diagnostic accuracy, increasing it by 10.1%. Finally, the fault experimental result validates the effectiveness of improved data preprocessing combined with an optimization DBN to diagnose faults in actual gas turbines.