Machine fault prediction and diagnosis method based on three-level neural network modeling

Abstract

The invention relates to a machine fault prediction and diagnosis method based on three-level neural network modeling, and belongs to the technical field of machine fault detection methods and artificial intelligence. According to the invention, the deep neural network modeling technology is adopted to improve the intelligence of diagnosis; whether a fault exists or not is diagnosed firstly; re-determining type, and finally, determining three-level neural network modeling of a severity level. Fault judgment can be achieved under the condition that data accumulation is not sufficient at the initial stage of system deployment, fault type judgment is conducted slowly and deeply along with data accumulation, finally, fault prediction is conducted on the progressive fault type, the period frominput to output of the diagnosis system is shortened through three-level neural network modeling, and practicability is improved. The fault sample set generated by the method comprises a large numberof low-level fault samples which cannot be identified by the existing method, and the model obtained by training has higher diagnosis accuracy and prediction capability compared with the existing method.

Description

technical field [0001] The invention relates to a machine fault prediction and diagnosis method based on three-level neural network modeling, which belongs to the technical fields of machine fault detection methods and artificial intelligence. Background technique [0002] A large number of key machinery and equipment in industrial production, especially the key machinery and equipment of assembly line operations, cannot be shut down for maintenance during a production cycle. Unexpected shutdown will cause major production accidents. For the operation and maintenance of these key machinery and equipment, the traditional way is to adopt a planned maintenance plan. The planned maintenance scheme does not consider the actual running state of the machine, so there is a problem that the machine in good condition is shut down for maintenance (over-maintenance), while the machine on the verge of failure is ignored (under-maintenance). The dangers of under-maintenance are obvious. ...

AI Technical Summary

Problems solved by technology

However, since the artificial intelligence model requires targeted training for different machines and equipment, each monitoring and diagnosis solution requires a long period of data accumulation before it is officially launched.

Moreover, since the training samples are completely marked by human experts, although the trained diagnostic model saves a lot of manpower and can replicate the experience of experts for tireless and continuous monitoring, it is ineffective in predicting potential failures that are difficult to detect manually. limited ability to diagnose

Images