ETR-LDA-based method for identifying wearing state of hard disc magnetic head

Abstract

The invention relates to an ETR-LDA-based method for identifying wearing state of a hard disc magnetic head. An acoustic emission sensor is used for collection to obtain a vibration signal of a hard disc magnetic head during a wearing period; the obtained vibration signal is processed by zone division within a time-domain amplitude range; on the basis of statistic information, a distribution rule of the signal in all zones is obtained by statistics and a high-dimensional statistic feature of equipment state information having the same dimension as the number of the zones is obtained; an objective function of an original TR-LDA algorithm is modified and thus a minimum inter-class distance is considered during a projection plane iterative solution process and the minimum inter-class distance is maximized during the iteration process, a weak link in state identification is corrected, sample data are projected into a two-dimensional plane, data sets having similar wearing degrees are collected together, and thus samples with different wearing degrees are distinguished. Therefore, while the algorithm convergence and global optimality are guaranteed, classification accuracy is improved obviously. An effective analysis method is provided for evaluation of a wearing state of a hard disc magnetic head.

Description

technical field [0001] The invention belongs to the field of fault diagnosis of mechanical equipment, and in particular relates to an ETR-LDA-based identification method for the wear state of a magnetic head of a hard disk. Background technique [0002] Fault diagnosis is an important link in industrial production. Accurate and timely identification of key equipment component failures is the key basis for ensuring safe operation of equipment, avoiding major accidents, and reducing casualties. Due to the various forms of equipment failure, and the early failure response during the operation of the equipment is weak and difficult to identify, there is a huge potential safety hazard. Once the fault type and degree are diagnosed incorrectly, the corresponding maintenance decision cannot correct the system to a normal state in time. As the defect further deteriorates, it will cause a major accident of the equipment and even threaten the life safety of the on-site personnel. [0...

AI Technical Summary

Problems solved by technology

[0005] At present, the fault diagnosis with the help of TR-LDA dimensionality reduction algorithm still relies on its original objective function, that is, to minimize the intra-class distance while maximizing the total inter-class distance, but TR-LDA cannot effectively solve the problem of weak faults with similar responses. fault identification problem

Images