The Research On Motor Fault Diagnosis Based On Multi-Sensor Information Fusion Technology

Motor is the main electromechanical energy conversion device, motor has irreplaceable position, not olny in the field of various energy and manufacturing of national economy in, but also in people’s daily life. Research of motor fault diagnosis technology has great economic and social significance, after several decades of development, motor fault diagnosis technology has made considerable progress, both in signal processing and methods of diagnostic, however, the most commonly-used motor fault diagnosis system which based on single parameter and single feature still has a lot of uncertainty in the process of diagnosis, it is sometimes difficult to ensure the accuracy of diagnosis, on this basis, this dissertation presents a motor fault diagnosis method based on multi-sensors information fusion technology.The object of study in this dissertation is motor fault diagnosis, in the first place, this dissertation introduced the background, significance and the development of motor fault diagnosis technology, meanwhile, gave a brief introduction of multi-sensors information fusion technology, and common faults such as motor stator fault, motor rotor fault, bearing fault and air-gap eccentric fault were analyzed. In the aspact of signal processing and feature extraction, this dissertation solved the problem of modal mixture and false components in EMD which is the core content of HHT, through the simulation experiment, verifyed the feasibility of EEMD in restraining modal mixture problem; through the simulation of the comparing between the correlation coefficient method, verifyed the effectiveness of grey relational degree method in detecting false components. And on this basis, using the IMF energy structure fault feature vector. In the aspact of local fault diagnosis methods, this dissertation adopted BP neural network which is the most widely used and most mature, introduces the basics, principles, structure and learning process of neural network. Use the excellent properties of neural network to provide more accuracy and higher reliability input informations for the multi-sensors information fusion method based on D-S inference. In the aspact of the multi-sensors information fusion algorithm, this dissertation introduced and analyzed the basic concept of D-S inference and D-S combination rules, in the structure of basic probability assignment, using the local diagnosis of neural network as the foundation, considering the error factors at the same time, not only solved the difficulties of how to build the basic probability assignment in D-S inference, but also avoiding the defects of D-S combination rules in Dealing with conflict evidence, effectively combined the neural network and D-S inference. Finally, this dissertation constructed a motor fault diagnosis system model based on multi-sensor information fusion technology, and select bearing faults which is the most common motor fault as the experimental object, having experiments and data analysis. Through the experiment, verifyed the feasibility, correctness and effectiveness of the motor fault diagnosis system based on multi-sensor information fusion technology which this dissertation had constructed.