Method of predicting residual life of gearbox by utilizing modified gray model

Abstract

The invention discloses a method of predicting the residual life of a gearbox by utilizing a modified gray model. The method comprises the following steps of (1) computing a correction factor epsilon of a prediction model based on damage evolution dynamic information of a physical model of the gearbox; (2) extracting a mixed-domain feature vector Fm and computing a weight vector wm thereof; (3) obtaining history data for vibration monitoring of the gearbox and computing a corresponding mixed-domain feature vector Fmj(k) and a weight vector wmj thereof; (4) collecting the vibration monitoring data of the gearbox in a real-time manner, and computing a mixed-domain feature vector to be tested Ft; (5) taking the feature vector to be tested Ft and the weight vector Wmj as inputs of the modified gray model, and taking the output of the model as FP; (6) substituting the output Fp of the current model into a life calculation model so as to obtain a prediction result Nf of the residual life of the gearbox. The method has the advantages of simple principle, simplicity in operation, stability and reliability, high precision and applicability to changeable or multi-interference work conditions.

Description

technical field [0001] The invention mainly relates to the technical field of gearbox failure prediction and health management, in particular to a method for predicting the remaining life of a gearbox by using an improved gray model. Background technique [0002] In the prior art, a preliminary life trend analysis of the gearbox is often performed through a single time-domain statistical index, and it is difficult to effectively, stably and reliably predict the remaining life of the gearbox. In this way, the vibration signal of the gearbox acquired by the sensor is first pre-processed to reduce noise, and then the corresponding characteristic index is calculated by the time-domain statistical method, and then the current state of the gearbox is judged and predicted by setting multiple thresholds , the changing trend of the future state. [0003] Generally speaking, there are a lot of noise and other power frequency interference in the working environment of the gearbox. It ...

AI Technical Summary

Problems solved by technology

[0003] Generally speaking, there are a lot of noise and other power frequency interference in the working environment of the gearbox. It is often difficult to overcome the above interference factors by using a single characteristic parameter to predict the state change trend, and it is also unable to adapt to changes in working conditions such as speed and torque. Cause high false alarm or false alarm, affect the normal operation of the equipment

Images