Reduction gearbox fault prediction method and device, computer equipment and storage medium

Abstract

The invention discloses a reduction gearbox fault prediction method and device, computer equipment and a storage medium. The method comprises: obtaining a vibration signal, an operation feature vector and an environment feature vector of a final polycondensation stirrer reduction gearbox; performing variational mode decomposition on the vibration signal to obtain a plurality of mode components; performing data fusion on the plurality of modal components, the operation feature vector and the environment feature vector to obtain a data source in a data matrix form; performing first feature extraction on data in the data source by using a first self-attention network to obtain a first feature; performing second feature extraction on the first feature by using a second self-attention network to obtain a second feature; and inputting the second feature into a full connection layer, and performing weight calculation on an output result of the full connection layer by using a classification algorithm to obtain a classification result as a fault prediction result of the final polycondensation stirrer reduction gearbox. According to the invention, the precision of fault prediction of the reduction gearbox can be effectively improved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method, device, computer equipment and storage medium for fault prediction of a gearbox. Background technique [0002] In the SSP (polyester solid phase polycondensation) production line, the final polycondensation stirrer is the main component of the main reactor, which has the largest volume and the most complex structure, and is connected in series with other equipment. If a failure occurs, on the one hand, it is necessary to stop the entire production line for fault diagnosis and maintenance, resulting in unplanned downtime equipment idle losses; on the other hand, sudden unplanned downtime will also cause chemical materials in the equipment to react The material is wasted due to insufficient reaction. [0003] As an important part of the mixer, the standard gear reducer of the final polycondensation mixer is a failure-prone and frequent part of the instrument. For exam...

AI Technical Summary

Problems solved by technology

[0005] However, traditional machine learning algorithms such as SVM (Support Vector Machine), RF (Random Forest), and GBDT (Gradient Descent Tree) are not very effective in processing high-dimensional nonlinear models; CNN networks (Convolutional Neural Networks) do not take time into account. The context of sequence data, and because industrial serialized data cannot be converted into images, it is not conducive to CNN to take advantage of feature extraction, and the prediction of results is inaccurate; RNN network (cyclic neural network) has a high probability of gradient for long-term sequence data The risk of disappearing; the LSTM network (long short-term memory network) has a poor classification effect on high-latitude small-sample data and is prone to over-fitting, and does not take into account the production environment of chemical materials, setting up a targeted algorithm mechanism model

Images