基于信号时频特征的电站锅炉性能诊断技术

The Performance Diagnosis Technology for Coal-fired Boiler Based on the Time-frequency Characteristics of Signals

根据信号的时频特征,建立了多分量信号与单分量信号之间的对应关系,并以此为基础提出了一种新的基于信号时频特征的非线性系统性能诊断技术。利用经验模态分解方法将多分量信号分解为一系列内蕴模态分量和残差函数,然后进行主成分分析,避免了模态分量间的相互关联与混叠,从而提高了Hilbert-Huang变换处理非平稳信号的准确性。以某电厂锅炉排烟温度波动诊断进行算例分析,结果表明此改进的HHT方法可有效分析各独立分量对系统性能参数的关联度。

Based on the time-frequency characteristics of signals, a corresponding relationship between a multi-component signal and a mono-component signal is established. And lays a foundation by this achievement, a new performance diagnosis technology for nonlinear systems is proposed in this paper. The connection and allasing among the intrinsic mode functions can be avoided by principle component analysis, which is used to deal the analysis results of empirical modal decomposition. Thus, the accuracy of Hilbert-Huang transform in processing non-stationary signals is improved perfectly. Through two performance diagnosis examples on the exhaust gas temperature of some coal-fired boiler, it can be demonstrated that the improved Hilbert-Huang transform can effectively analysis the relational degree between every independent component and the performance parameter.