Research On Fault Feature Extraction Of Rotary Kiln Based On Empirical Wavelet Transform

Rotary kiln is a large production equipment of building materials,metallurgy,chemical industry and other industries,it mainly consists of the shell,tyres,rollers and drive system.The long-term operation of the rotary kiln in the harsh conditions of high temperature and heavy load often lead to hidden faults such as kiln center line deviation and shell bending deformation.The fault characterization is not obvious.When the fault develops seriously,the roller bearing bush heating causes production failure,which brings safety risks and economic losses to the enterprise.At present,there is a lack of research on the theory and method of early fault detection in rotary kiln,and the processing method of rotary kiln fault signal needs to be improved.Therefore,this paper analyzes the main faults of rotary kiln,compares the advantages and disadvantages of classical time-frequency analysis methods in fault feature extraction,and proposes a fault detection method based on empirical wavelet transform(EWT).The main research contents are as follows:(1)This paper qualitatively analyzes the force of the supporting roller,and discusses the effects and relationship of the two kinds of failures of the shell bending deformation and the kiln center line offset on the deflection of the roller.It is found that the roller deflection signal contains this two kinds of fault characteristics information.The deflection expression of the roller contains characteristic parameters such as the shell operating frequency,the roller operating frequency,the shell bending deformation and the equivalent mass of the shell at the kiln pier,which indicates that the shell bending deformation affects the amplitude of the shell operating frequency harmonics,and the equivalent mass of the shell affects the amplitude of both roller and shell operating frequency harmonics.(2)According to the above measurement analysis results,a measuring system for the deflection signal of the roller is designed,which introduces the selection of each sensor in the measurement system,the design and manufacture of the sensor fixture and the design of the LabVIEW program of the host computer software.The measuring system is applied to the industrial site,and the roller deflection signals are successfully obtained.(3)Since the measured roller deflection signal shows non-stationarity,several classical time-frequency joint analysis methods are used to deal with it and can all decompose the fault characteristic harmonics.But due to its own method problems,the decomposition effect is not good.Therefore,a feature extraction method of roller deflection signal based on empirical wavelet transform is proposed.It absorbs the advantages of empirical mode decomposition and wavelet transform and effectively prevent the occurrence of modal aliasing.Both the simulated signal and the actual signal show that the empirical wavelet transform can effectively extract the characteristic modes in the signal,and the adaptive decomposition effect is better than the empirical mode decomposition.Finally,the EWT results of the roller vibration signal are compared with the results of other rotary kiln detection methods,which verifies the correctness and feasibility of EWT in fault detection of rotary kiln.