Comprehensive Alarm Method for Equipment Condition Based on Parzen Window Estimation
LIU Han1,3, ZHANG Qing1,2, MENG Lihua1, YANG Kai1, XU Guanghua1,2
1.School of Mechanical Engineering,Xi′an Jiaotong University,Xi′an 710049,China; 2. State Key Laboratory for Manufacturing Systems Engineering, Xi′an Jiaotong University,Xi′an 710049,China3. Shanghai Baosteel Industry Inspection Corporation, Shanghai 201900, China
Abstract:To improve the adaptability of traditional monitoring and alarming method and fuse multiple monitoring parameters, a comprehensive alarm method for condition monitoring based on Parzen window estimation was proposed. The joint probability density function of multiple monitoring parameters was estimated to describe the distribution of monitoring data. The alarm boundary was set by Parzen window function of monitoring data located in the marginal distribution area. Along with the accumulation of monitoring data, the alarm boundary was continuously adjusted to accommodate new operation performance. The proposed method was evaluated by monitoring data measured on rotor test rig and heating furnace fan. Results show that faults under different running conditions are well identified.