A method and system for real-time intelligent diagnosis of bearing faults based on attention CNN model

Abstract

The present invention provides a real-time intelligent diagnosis method and system for bearing faults based on the Attention CNN model, including using a vibration sensor to collect faulty bearing vibration signals, and then segmenting the faulty bearing vibration signals using a fixed-length random segmentation method to obtain data samples; After the data samples are affixed with labels corresponding to each type according to the state type of the rolling bearing, they are divided into a training set, a verification set and a test set according to a certain ratio; according to the data in the training set and the verification set, a variety of data in the The bearing fault data set in the unbalanced state and all the bearing fault data sets produced constitute the unbalanced data set; construct the above model, train the above model with different bearing fault data sets respectively, and obtain the above training model; use the above training model Real-time fault detection is performed on the rolling bearing. The invention can identify the operating state of the bearing in real time, accurately and automatically, thereby effectively maintaining the normal operation of mechanical equipment.

Description

technical field [0001] The invention relates to the technical field of equipment health management, in particular to a method and system for real-time intelligent diagnosis of bearing faults based on the Attention CNN model. Background technique [0002] Bearings are a key component of modern industrial equipment. The working scene of bearings is complex. Once a failure occurs, it may cause serious safety accidents, causing a large number of casualties and huge economic losses. Bearings are one of the key supporting components in equipment such as helicopters, aero engines, and wind turbines. Therefore, it is very important to detect faults in a timely and accurate manner and eliminate potential safety hazards of machinery. Therefore, how to diagnose the faults of rotating machinery in real time, accurately and automatically is of great significance to ensure its normal operation and safe production. [0003] The traditional signal-based method refers to the use of various ...

AI Technical Summary

Problems solved by technology

[0006] Although there are many rotating machinery fault detection models and good experimental results have been obtained, there are still many challenges in the field of rotating machinery fault diagnosis.

① In order to obtain higher accuracy, the traditional deep learning fault detection model often adopts the method of multi-layer neural network superposition, so the model complexity is too high

The existing models are all experimented with an ideal number of samples. The actual industrial situation is that the fault samples are less than the normal samples, so the data imbalance phenomenon generally exists in the industrial field.

Images