Bearing health assessment method based on composite model

Abstract

The invention relates to a bearing health assessment method based on a composite model, which comprises a signal acquisition stage, an index extraction stage, a model training stage, a bearing matching stage and a model application stage, and specifically comprises the following steps of: in the signal acquisition stage, selecting a target bearing to carry out a full life cycle experiment; the invention provides a bearing health assessment method based on a composite model, the method is simple and universal, the data processing precision is high, the robustness of a core algorithm is good, the neural network model, the index health model and the logistic regression model are adopted for assessment for different types of bearings, and the assessment accuracy is high. By judging the optimal health degree curve, the evaluation result is more accurate and effective, proper maintenance measures can be conveniently taken in advance, and bearing faults are avoided.

Description

technical field [0001] The invention relates to the field of bearing monitoring, in particular to a bearing health assessment method based on a composite model. Background technique [0002] Bearing is an important component in a rotating mechanism, and the operation of the bearing is often critical. At present, the fault monitoring methods of bearings mainly include the mechanism model method based on the degradation process and the machine learning method based on data. The principle based on the mechanism model method is relatively Simple and easy to understand, but the algorithm has low precision and poor adaptability; the data-based method is prone to over-fitting, the algorithm model is more complex, and there are many initial parameters to set, which is not easy to migrate the algorithm, resulting in inaccurate bearing health. The evaluation will eventually lead to an unexpected situation, or the bearings in good condition are abandoned, wasting resources. SUMMARY O...

AI Technical Summary

Problems solved by technology

[0002] As an important part of the rotating mechanism, the bearing's operation is often very important. At present, the fault monitoring methods of bearings mainly include the mechanism model method based on the degradation process and the machine learning method based on data. The principle of the method based on the mechanism model is relatively Simple and easy to understand, but the accuracy of the algorithm is low, and the adaptability is poor; the data-based method is prone to overfitting, the algorithm model is more complex, and there are many initial parameters to set, it is not easy to migrate the algorithm, resulting in inaccurate bearing health evaluation, which eventually leads to unexpected situations, or bearings in good condition are scrapped, wasting resources

Images