.png){kind=icon}
.png){kind=icon}
Return
Harmonic Reducer Performance Prediction Algorithm Based on Multivariate State Estimation and LargeVis Dimensionality Reduction
.png){kind=icon}
OA
AI
DOI：10.1109/ACCESS.2022.3166921.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
As an important transmission component of industrial robots, the harmonic reducer determines the positioning accuracy, bearing capacity and service life of the robot end-effector. Predicting the performance can grasp the working status in advance and avoid major losses caused by uncertain factors such as component damage. The current paper focuses on a harmonic reducer performance prediction algorithm based on Multivariate State Estimation Technique (MSET) and LargeVis dimensionality reduction. Firstly, an accelerated life test platform is designed to collect multi-dimensional parameters that can characterize the operating state of the harmonic reducer throughout the life cycle. Afterwards, as far as the MSET method is concerned, the fault warning threshold is set according to the residual between the constructed memory matrix of the health state data and the actual observed value. Finally, utilizing LargeVis to reduce the dimensionality of multi-dimensional features, combining with Mahalanobis distance to construct a health index degradation model, and then selecting Long Short-Term Memory (LSTM) network to predict the downward trend of the harmonic reducer. The analysis of the accelerated life test data of the harmonic reducer demonstrates that the proposed method can send out the fault warning signal 18 minutes in advance in the sample with a life of 5.7 hours, and has a strong ability to predict the degradation trend of the harmonic reducer.
Keywords:
Harmonic reducer
performance prediction
multivariate state estimation technique
LargeVis
features dimensionality reduction
long short-term memory network
Journal
IF:
3.6
Papers:
9.8W
Citations:
29.4W