At present, oil analysis, noise monitoring, vibration monitoring, nondestructive testing, torsional vibration analysis, temperature and energy consumption analysis are commonly used methods for gearbox fault diagnosis. Because the characteristics of the vibration signals in the normal operation and fault state of the gearbox will be different in the time and frequency domain, the vibration signals can be used as the fault feature carrier, and the effective technical means can be used for analysis to determine the location, cause and development degree of the fault. The detection process can be realized without shutdown, which can save time. So far, the analysis methods developed for different fault signal characteristics are more and more abundant, and the analysis effect is also significantly improved, which has become one of the main technologies of gearbox fault diagnosis, and has been widely used. People began to study the vibration and noise of gearbox at the beginning of the 20th century, but it was not until 1968 that it was regarded as one of the standards to determine whether the performance of a gear device meets the requirements, and it has attracted worldwide attention. According to the mechanism of gear vibration and noise, H. optiz, Buckingham and Niemann, British scholars, put forward their own opinions on the vibration and noise of gearbox.
From 1970 At the beginning of the year, some simple parameters, such as vibration peak value, root mean square value, etc., were used to diagnose the gearbox fault as an index to evaluate the performance state of the gear. Later, some dimensionless parameters (such as peak coefficient, etc.) were used to eliminate the adverse effect of machine load change on Fault Analysis and measurement. However, at that time, these technologies were sensitive and accurate to fault diagnosis The rate is not high. From the 1970s to the end of the 1980s, B. Randall and James I. Taylor et al. Have done a lot of research on the frequency domain analysis technology in the gear fault diagnosis, and have achieved a certain effect. The analysis technology has done a lot of research in the gear fault diagnosis, and has achieved a certain effect [6]. [6]. The research of related fields in our country started much later than that in foreign countries. Until entering the 21st century, the fault diagnosis of equipment by using vibration theory has been gradually popularized in practical engineering application. Some research institutions and scholars have done a lot of work, and have some theoretical and practical application results. For example, Qu Liangsheng and he Zhengjia analyzed the time-frequency characteristics of the gear fault vibration signal, and elaborated it in detail in the mechanical fault diagnosis; Han Jie and others proposed that the gear fault vibration signal has the characteristics of long period and short period.
In the practical engineering application, in order to better analyze the characteristics of a specific signal, improve the signal-to-noise ratio of fault signal analysis, and improve the accuracy of fault diagnosis, many algorithms are put into the signal analysis and processing, and constantly improve the shortcomings of the original analysis methods, derive a large number of methods that can be used for the analysis and processing of gearbox fault vibration signals, and also promote the status of fault diagnosis The improvement and development of state monitoring and analysis system. For example, in order to improve the problem of low accuracy in the process of using spectrum analysis, discrete spectrum correction technology is proposed to improve the analysis accuracy; in order to improve the frequency spectrum analysis of the modulated signal of gearbox fault signal, the mixing phenomenon that may be generated by the high pass absolute value analysis, detection filtering and square filtering of traditional generalized detection filtering, Hilbert change In order to improve the performance of the traditional complex system, the complex analytical band-pass filter based complex system is proposed. In order to improve the performance of the traditional complex system, the complex analytical band-pass filter based complex system is proposed. In order to improve the traditional fast Fourier transform, only the whole spectrum of the signal can be obtained, but not the local information of the signal In order to solve the problem that the time-frequency window of Gabor transform cannot change its time width and frequency width, some short-term features of signals will be hidden during analysis, and the wavelet transform with multi-scale analysis function for coordinate axis translation and expansion can be used.
In practical application, different methods have different processing effects for different types of fault signals. They often adopt specific methods or adopt several different methods to analyze and process specific signals at the same time. For example, some domestic experts put forward the time-frequency scoring theory based on the establishment of archives. It uses spectrum analysis, refinement analysis and narrowband refinement envelope analysis to analyze the time domain signal and envelope time domain signal, and calculates the total score based on the score corresponding to the eigenvalue. When the total score exceeds the set value, it is considered that the fault occurs. This method has been widely used in the fault diagnosis of gear, shaft and rolling bearing; Taiyuan University of technology has put forward A neural network diagnosis method based on time, frequency and wavelet analysis, which combines time, frequency and wavelet analysis to identify gearbox fault signals, and uses neural network to train and diagnose the obtained information, so as to improve the accuracy and reliability of gearbox fault diagnosis.