Research On Intelligent Fault Diagnosis Of Electro Mechanical Equipment Based On Integrated Neural Networks

Electromechanical equipment has become increased automation,information,and intelligence with the perfection of its functions.The research in condition monitoring and fault diagnosis of devices has become the effective means to ensure the safety of production and growth of economic.The vibration signal is the most widely used sensing information in the condition monitoring and fault diagnosis because of its easy acquisition,rich information and high sensitivity.However,the operating environment of mechanical equipment is noisy and the fault signal is non-linear and non-smooth,which make the existing method of condition monitoring and fault diagnosis have many shortcomings.The paper does some optimization and improvements in noise suppression and fault feature information extraction and improves the reliability,accuracy,and real-time in the condition monitoring and fault diagnosis of equipment through introducing the artificial intelligence.The paper does much research in existing technology of signal processing and methods of condition monitoring and fault diagnosis,analyzes the advantages and disadvantages of methods based on the traditional signal processing and explains the importance of effective extraction of fault feature information in the fault diagnosis technology.According to the feature that the vibration signals of equipment are from various sources,non-linear and non-smooth,the paper provides the method that combines EMD and ICA and uses the EMD-ICA technology to finish the source signals dissociation,de-noising and feature separation,which solves the problem of mode mixing in EMD and the limitation to the signal-channel in ICA.As a result,the method obviously improves the result of noise suppression and contributes to the clear extraction of fault feature information.The paper provides a method that makes use of integrated neural network to make fault diagnosis because traditional pattern identification,expert system and the single neural network are not helpful enough in the judgment of the operation situation for equipment and fault diagnosis.The method separates the diagnosis into several grades,finishes multi-features fusion of the same fault of vibration signal with the help of sub-network and makes the final decision about the equipment fault after a preliminary identification.The method obviously improves the real-time ability of the system,holding the accuracy rate of fault diagnosis at the same time.The paper proposes the idea and realization strategy of signal fault feature information extraction and fault diagnosis.The paper also builds the general frame,designs the intelligent fault diagnosis system based on MATLAB platform and finishes the simulation of roll bearing,which is one of the critical components in equipment.The result proves that the system in the paper improves the accuracy and real-time ability greatly when compared with the fault diagnosis system which is based on tradition signal processing methods.