Blind Separation Of Sources And Application In Fault Diagnosis Of Type 2D12 Reciprocating Compressor

The research on Blind Signal Separation (BSS) is more and more deep in recent decade. It's widely used in many fields. Because the blind processing techniques could separate signals without a prior information of the original signals and the channel. It can be used in fault diagnosis to separate mixed signal. Second order blind independence (SOBI) could separate non-stabile sources. In this paper, the method is used in the fault diagnosis of compressor of type 2D12 reciprocating compressor to exploit character parameter.Based on subject background, in this text, mainly do work on several parts below:1. The valve is the common fault in reciprocating compressor, model on dynamic and thermodynamic. Analyse the movment and the press of valve. Research the vibration on gas crock cover, the main inspirit sources are valve impact on seat and the press of gas crock.2. Analyse the independent criterion of blind source separate, introduce a natural gradient separate method based on uniform aim function, advance its power function and do simulink to compute their separate performance. Validate it has better astringency and stability. Simulink several second order blind independent method. Compare their separate performance under different signal-noise ratio.3. Combine blind source separate and other signal process method, separate mixed vibration and press signals, realize pretreatment of machine signal to exploit character parameter, achieve fault diagnosis, validate its feasibility.