Adaptive Dynamic Information Extraction And State Evaluation For Machinery Under Varying Speeds

The dynamic signals of machinery in varying speed operations are more complex than those in sationary operations.Those signals are non-stationary in nature and characterized as frequency modulation,amplitude modulation and phase modulation,which bring difficulty for the fault symptom extraction.For those reasons,there is a high demand for new diagnostic theory for the condition monitoring and fault diagnosis of machinery in varying speed operations.The mechanism of fault signal generation and propagation is investigated firstly in this dissertation.On this basis,the adaptive extraction methods for the instantaneous amplitude,frequency,phase of non-stationary signal are developed.In addition,the evolutions of fault signature with the rotating speed of mechinery are also investigated.Those methods developed in this dissertation could not only expand the scope of current diagnostic approaches,but also form a solid foundation for the fault diagnosis of machinery with varying speed conditions.The effectiveness of those methods are demonstrated by a variety of industrial applications,such as rotor crack detection,gearbox run-up signal analysis,fault diagnosis of rolling element bearing under varying speed and transmission error identification of machine tools.The significance and difficulty of this dissertation is introduced firstly.Then,the history and development of fault diagnosis methods for machinery under varying-speed conditions are reviewed and categorized from the aspects of signal acquisition and preprocessing,non-stationary signal analysis and intelligent fault classification,etc.The shortcomings of current methods are also discussed.The instantaneous rotating speed carries a lot of useful information about the running condition of machinery.It is also a key parameter for the condition monitoring and fault diagnosis.Traditional rotating speed estimation methods suffer from large estimation error due to low frequency resolution of STFT.To overcome this limitation,an instantaneous speed estimation method based on short-time chirp-Fourier transform is proposed.In addition,an adaptive chirp rate selection scheme is further established.The analyzed signal could be decomposed in an adaptive way according to its time-frequency distribution in theproposed method.Hence the smearing problem could be avoided,and the computational efficiency is enhanced as well.Order tracking is an effective tool which is capable of transforming non-stationary signals into cyclo-stationary ones.However,the implementation of current order tracking methods require additional hardware(optical encoder,tachometer,keyphaser,etc)to provide a reference signal,which not only increases the cost,but also brings difficulties in the installation and adjustments.To extend the applications of order tracking,a new tacholess order tracking method for large speed variation is established by combining the adaptive short-time chirp-Fourier transform and Vold-Kalman filtering.The proposed method can be performed without tachometer,thus providing a powerful tool for the non-stationary diagnosis.In the traditional gearbox diagnostic methods,the fault symptoms are extracted under a constant rotating speed.However,it is pointed out that the detectability of fault is highly related to the rotating speed of gearbox.For this reason,the diagnostic information extracted under constant speed is partial,and cannot reflect the real condition of gearbox.It explains why the rate of misdiagnosis is high in traditional methods.To overcome this deficiency,a tacholess short-time phase demodulation(STPD)technique is proposed to detect the gear fault in the run up/down process.In this technique,by integrating the merits of flexible time-domain averaging and short-time phase demodulation,the fault introduced phase modulation is presented as a two-dimensional function of rotating speed and rotating angle of gear.It is shown that STPD could not only give a clear indication of fault,but also provides a guideline for the optimal rotating speed selection for the gear fault diagnosis,which is a great helpful for the condition monitoring and fault diagnosis of gearbox.Rolling element bearings are critical mechanical components in rotating machinery and their failure may lead to fatal breakdown and significant economic losses.Conventional bearing diagnostic methods are based on the assumption of constant running speed.However,in practice,almost all of the bearings experience different kinds of speed variations.Under varying speed conditions,the repetition frequencies of impulses also vary with time and hence the corresponding envelope signals are non-stationary in nature.The direct application of frequency-based methods(such as envelope spectrum analysis,spectral correlation)to those vibration signals of bearings will lead to spectral smearing and false diagnosis.To overcome the shortcomings of conventional methods,a tacholess envelope order analysis technique is established in this paper.In this technique,a tacholess order tracking method is first proposed based on generalized demodulation transform.By using this method,the tacho information of the bearing could be recovered from the vibration signal itself.On this basis,envelope order spectrum is utilized to transform the non-stationary envelope signal in the time-domain into a cyclo-stationary signal in the angular domain.In this way,the smearing problem caused by speed variation is solved effectively.The effectiveness of the proposedmethod is demonstrated by real vibration signals collected from locomotive roller bearings with faults on inner race,outer race and rollers,respectively.Analyzed results show that the proposed method could identify different bearing faults effectively and accurately under speed varying conditions.Although the vibration analysis is one of the most effective approaches to the condition monitoring of mechinery,however,the vibration signal is not always available due to environmental restrictions.How to exploit new diagnostic information source is a hot and interesting topic in recent years.Aiming at this issue,we investigated the fault diagnosis scheme by using the built-in encoders of mechinery.In order to capture the fault introduced speed variation accurately,an instantaneous speed estimation method based on digital differentiator is established.The influence of window type and window length selection on the performance of digital differentiator is also researched.The effectiveness of proposed method is verified by an experiment on a CNC machine tool.The vibration source of the rotary table is identified successfully.By utilizing the built-in encoder information and the proposed non-stationary signal analysis methods,we developed a dynamic transmission error measurement and identification system.This system has been used in several CNC machine tool companies,and has solved many industrial problems,such as transmission error,multi-axis synchronization error,assembly error measurement and identification effectively.