On Non-Stationary Signal Transformation and Feature Extraction

Abstract

Wei Wang and Yongjian Sun

With the continuous development of the current society, the national economic level has been improving, and the people's material demand is higher and higher. Currently, it is necessary to improve the output and efficiency of materials. Therefore, mechanical automation equipment is also developing at a high speed, and the production efficiency is constantly improving. However, there are many important large-scale mechanical equipment, such as steam turbine or large ball bearing. If they work in a harsh environment for a long time, it is easy to cause damage. Whether it is the fault caused by manual disoperation or the fault caused by bearing damage, the shutdown may cause economic losses, or casualties and environmental damage. From the beginning of human beings, signal is the carrier of language and the medium of all kinds of things. According to human experience, the signal is divided into two parts, deterministic signal, and uncertain signal. Uncertain signals are divided into two kinds, stationary signals, and non-stationary signals. Nonstationary signals will follow time to transform, and the statistical feature is a function of time. For example, fault diagnosis can prevent and avoid bearing faults by transforming the non-stationary signals in various random signals generated by noise during the operation of equipment and engineering systems and extracting the required fault diagnosis features for diagnosis. If China can vigorously promote the use of non-stationary signal fault diagnosis, and many large equipment can be detected through fault diagnosis, it is likely to avoid the occurrence of faults. Because the loss of shutdown is sometimes very large, especially in the processing industry. For example, for a large synthetic fertilizer machine, the loss caused by shutdown for one day can be as high as several million yuan. If the system fault is diagnosed through nonstationary signals, it can completely prevent economic losses and even casualties. Therefore, the study of non-stationary signals is very meaningful, and it is also a subject that the country should vigorously develop.

PDF