Application Of Information Fusion Theory In Fault Diagnosis Of Servo Motors

With the continuous advancement of science and technology,the design and manufacture of electric motors are developing with the goal of high speed,high precision and high reliability.The motor plays a pivotal role as the power or tuning component of most automation equipment.Due to the harsh working environment such as high temperature,oil pollution,overload,and manufacturing and assembly errors,the operation of the motor will be accompanied by potential risks.In the event of failure of some important electrical equipment,huge losses will occur,endangering the lives of personnel.Therefore,signal monitoring and fault diagnosis during motor operation are particularly important.The traditional method of motor fault diagnosis is based on some observable measurements,which are processed and analyzed by some mathematical or signal processing methods.These observations mainly include vibration,temperature,noise,current,voltage,and the like.Analysis of these signals requires engineers with a wealth of practical experience and a strong background in electrical equipment.Due to the weak signal in the early stage of motor failure,the motor operating environment is complex,the requirements for sensor accuracy are high,and the economy is poor.Moreover,the conclusions obtained from the single-parameter sensor have inherent uncertainties,and the diagnosis based on this may lead to problems such as false positives and false negatives of the fault.In view of the inherent uncertainty of the single-parameter method,considering the information obtained by synthesizing multiple sensors,the different observations of the same object are redundantly complemented to obtain a more accurate description of the object.This method of synthesizing information obtained by multiple sensors and comprehensively utilizing it is an information fusion method.Such methods were mainly researched on data when they were proposed,so they were also called data fusion at that time.After decades of development and research,the object of information fusion research has not only been limited to data,but the concept of information fusion has become more extensive.In this paper,the concept,model,theoretical system,development status and main algorithms of information fusion are introduced.The advantages and disadvantages of various algorithms are summarized.The fault characteristics of servo motor are discussed in combination with the characteristics of the algorithm.In this paper,the types of motor faults and their characterizations are analyzed.The characteristics of motor fault signals are discussed.For the diagnosis of traditional motors,most of them have low diagnostic accuracy based on single sensors.The data fusion method is proposed to integrate data.And use the data after integration for diagnosis.According to the different operating states of the motor,two methods of motor bearing fault extraction based on information fusion and FastICA and motor bearing composite fault feature extraction method based on Kalman filtering and EWT are proposed.Two different experimental data are used for verification.Simulate the operating environment of the servo motor fault test bench with poor operating conditions,determine the hardware and acquisition equipment,and collect the experimental data to verify the proposed algorithm,which proves that the algorithm has good diagnostic accuracy and strong robustness.