The invention provides a bearing
health index construction method based on prognostic standard embedding and constraint fusion process, which is characterized by comprising the following steps: step S1, collecting
full life cycle vibration signals of a bearing, carrying out data preprocessing, and dividing a model
training set and a
test set; s2,
training set degradation labels are constructed, an early fault point is determined by using a
root mean square 3 criterion, a healthy section is 0, and a fault section is 1; s3, extracting features from dimensions such as a
time domain, a
frequency domain, a sparsity feature, an entropy feature, an envelope spectrum feature and the like, and carrying out
nondimensionalization on dimensional features such as a
peak value and the like through
standardization; s4, constructing a health mixing criterion (HM) as a quantitative index of a prognostic standard for evaluating the degradation description capability of each feature; and S5, constructing an optimization objective function with double regularization by taking the prognosis criterion constructed in the step S4 as a guide item. And S6, monotonicity and tendency constraints of the health indexes are added by using the objective function constructed in the S5, and a total optimization model is obtained. And step S7, using an alternating direction
multiplier method (ADMM) in combination with a projection operator on the
training set to solve a constrained objective function to obtain a feature weight. And a step. And S8, performing
health index construction and performance
verification on the
test set based on the weight obtained in the S7.