Rotating machine fault diagnosis method based on fusion of genetic programming and weighted evidence theory

Abstract

The invention discloses a rotating machine fault diagnosis method based on fusion of genetic programming and a weighted evidence theory, comprising a diagnosis model, genetic programming, a weighted evidence theory and a fusion algorithm. Compared with the prior art, the method comprises the steps as follows: firstly, extracting multiple fault feature parameters through genetic programming, and preliminarily diagnosing a rotating machine; secondly, by taking the preliminary diagnosis results of the feature parameters to faults as evidence bodies and taking the diagnosis accuracy of the feature parameters to faults as the weights of the evidences, correcting the current diagnosis result based on historical data of fault diagnosis; and finally, fusing the multiple evidences based on the weighted evidence theory for decision making, thus reducing the uncertainty of fault diagnosis and realizing accurate diagnosis of faults. Experimental results show that the method improves the accuracy of fault diagnosis.

Description

technical field [0001] The invention relates to a rotating machinery monitoring technology, in particular to a rotating machinery fault diagnosis method integrated with genetic programming and weighted evidence theory. Background technique [0002] The current state monitoring and fault diagnosis of rotating machinery is often to analyze and observe the vibration information of rotating machinery, extract effective fault characteristic parameters from it, and use the extracted fault characteristic parameters to carry out fault diagnosis. High-quality fault characteristic parameters can often improve the accuracy of diagnosis, but sometimes the optimal fault characteristic parameters extracted cannot accurately distinguish faults, which brings limitations to the effective diagnosis of rotating machinery faults. The faults of rotating machinery can be diagnosed more accurately by considering the diagnostic information of multiple extracted optimal fault characteristic paramete...

AI Technical Summary

Problems solved by technology

However, in the process of extracting characteristic parameters, genetic programming will converge to a solution according to different operating parameters, but it cannot guarantee that the converged solution is the global optimal solution, and may be an approximate optimal solution.

For the problem that different fault characteristics of rotating machinery present complex symptoms such as multi-coupling and ambiguity, if only a single approximate optimal fault characteristic parameter is used for diagnosis, there will inevitably be randomness, and it is difficult to achieve a high fault classification effect

Images