Research On Fault Diagnosis Technology Of Sintering Fan Based On OSELM

The sintering fan is the core of iron and steel plants sintering equipment which can provide enough oxygen for the full combustion of the sintering mixture in the sintering process.In operation,the abnormal vibration will be caused by fan's fault.So,vibration signals are regarded as the effective basis for the fan fault diagnosis.In the thesis,the study focused on sintering fan and vibration signal analysis method is used to extract fault features and artificial intelligence method of fault feature recognition is proposed based on wavelet packet and extreme learning machine fault diagnosis methods.In-depth study include some methods of fun's fault diagnosis involved in the technology of fault diagnosis of the exhauster of theories,methods and key technologies.The main works was as follows:(1)Research difficulties and key techniques of fault diagnosis.Current situation of research on fault diagnosis technology at home and abroad,cause and damage of the sintering fan fault were summarized.The significance of fault diagnosis was analysed.The common signal processing and fault diagnosis methods and applied fields were introduced and the common failure types and fan sign and implementation process of fault diagnosis were elaborated.(2)The fault feature extraction based on wavelet packet analysis.According to the features of the fan signal feature,Wavelet packet analysis method be used to extract the fault signal that the signal at each time and space changes on different scales were analysed.Through the simulation experiment proves: the fan vibration characteristics extracted effectively and excellent distinguish between different states of fan vibration signals was displayed when the wavelet packet transform be used in fan's fault diagnosis.(3)Fan fault diagnosis based on online sequential learning algorithm.Fan which usually installed on severe environment in actual industrial processes is hard to accurately classification of the fan of state.Fast and reliable diagnosis is given based on the OSELM method of fault diagnosis of the exhauster,which has fast training of the algorithm,generalization,also online sequential learning study the information that feature vector extracted through wavelet packet analysis,adaptive work environments to identify failures.(4)Case analysis.OSELM used in fault detection of sinter plant exhauster withactual experimental data.In the experiment,fault diagnosis model for sintering fan be established,OSELM algorithms are analyzed with different parameters and hidden-layer activation function on the performance of its impact.Through case analysis,the feasibility of OSELM in fault diagnosis of fan were proved.Compared with the ordinary ELM network,OSELM on the premise of allowable reduce training time slightly but the accuracy of fault diagnosis has improved.It has good application prospects and popularized value in steel industry.(5)The fault diagnosis system of fan base on Matlab GUI.After realization of fault diagnosis,MatIab GUI tools be used to program a fault diagnosis system which can operate on Matlab environment in user-friendly method and engineering application integration design of fault diagnosis.Through interactive way to improve ease of use,flexibility,methods that can be used in this vivid display in front of the user,make it easier for users to understand and promote.