Machine and pump health diagnosis method based on support vector machines

Abstract

The invention relates to a machine and pump maintenance technology, and aims to provide a machine and pump health diagnosis method based on support vector machines. The method includes the following steps that real-time vibration signal data during operation of machines and pumps are acquired, and time-frequency domain analysis is performed on vibration signals; time-frequency domain feature signals are extracted from a frequency domain oscillograph obtained through time-frequency domain analysis; the time-frequency domain feature signals are substituted into a trained support vector machine health diagnosis model for diagnosis, and machine and pump operation state fault conditions are obtained. According to the method, multiple second-category support vector machines are used, so that thestructure is simple, calculation is convenient, and the fault diagnosis speed and accuracy are improved; the number of the support vector machines is greatly reduced, and the phenomena of classification overlapping and incapability of classification are avoided; the support vector machine diagnosis model can be updated in real time with change of practical samples; the requirements for data sample structures and numbers are reduced, the problem difficulty is greatly reduced, and the method has the very high practical value.

Description

technical field [0001] The invention relates to machine pump maintenance technology, in particular to a support vector machine-based machine pump health diagnosis method, which is the application of statistical analysis technology and machine learning technology in machine pump health diagnosis. Background technique [0002] After the support vector machine was officially published in 1995, due to its excellent performance in text classification tasks, it quickly became the mainstream technology of machine learning, and directly set off a learning climax of "statistical learning". Support vector machine is a new nonlinear pattern recognition method, which has better classification accuracy and generalization ability than traditional neural network and Bayesian network algorithm. In addition, the support vector machine is a typical binary classifier, which has low requirements on the number of sample data and is easy to implement. At the same time, with the development of var...

AI Technical Summary

Problems solved by technology

[0003] Mechanical and pump equipment such as fans, water pumps, and air compressors are key equipment in modern enterprises such as petrochemical, metallurgy, steel, and electric power. Its fault diagnosis and health monitoring have always been difficult, and there is no specific implementation using support vector machines. Method of pump health diagnosis

Images