Research On Constrained Independent Component Analysis For Gearbox Fault Diagnosis

With the rapid development of modern industrial technology and the improvement of automation, mechanical equipment become more and more large, integrated, high-speed and high-load, while its working environment is very complex. Once the key components of the mechanical equipment fail unexpectedly, the unexpected failure can affect the whole equipment and even the whole manufacturing system, result in huge economic losses and catastrophic accidents. Mechanical fault diagnosis technology is one of the main measures to guarantee the equipment operation effectively, they can predict and judge the cause of the faults, put forward countermeasures to reduce the accidents. Therefore, it is of great significance and practical engineering value to carry on the fault diagnosis technology research.The extraction of failure feature information is the core content of the mechanical fault diagnosis technology, how to effectively extract the real signal feature under stony noise and apply to practical engineering is a hot research field of fault diagnosis.New theory,technology and methods come out one after another to enrich and improve the mechanical fault diagnosis technology. The gear of the gearbox are the research objects of the paper, the constrained independent component analysis(Constrained independent component analysis,cICA) is the signal analysis tool. The feasibility and validity of the cICA to gearbox fault diagnosis is demonstrated by both experiment and case studies.The paper mainly includes the following contents:(1) The mathematical foundation, the basic principle and algorithm associated with independent component analysis(Independent component analysis,ICA) are introduced. First the mathematical foundation is introduced, then the model, the uncertainly and identifiability of model and the related principle of ICA are studied. The advantage, the algorithm and the establishment of the reference signal of the cICA are finally studied.(2) From the structure of the gearbox respectively expounds the characteristics of the vibration source, and to analyze its vibration mechanism. The main parts and components of the gearbox fault types and the characteristics of the corresponding fault signal are introduced. Finally the mixing mechanism of the gearbox are studied, establish the vibration signal model of gearbox.(3) Experiment and analysis. Experiment is conduced to demonstrate the feasibility and validity of constrained independent component analysis in the gearbox fault diagnosis.Using Drivetrain Diagnostics Simulator to simulate the partial fault of the gear, then use the data acquisition instrument and the corresponding software to acquire the vibration of the missing teeth and local missing teeth of gearbox vibration signal. And then through the method of cICA to extract the fault feature of the missing teeth and local missing teeth, draw the corresponding conclusion from the analysis of the experiment data, to verify the validity of the algorithm and its characteristics.(4) The engineering case analysis. In view of the weak fault signals of the mine hoist gearbox,combined with the prior knowledge of the fault information.The cICA technology is applied to fault diagnosis in the mine hoist gearbox, extract the gear fault feature from the complex vibration signals,eliminate the redundant information of the characteristic, and puts forward effective features, implements the precision fault diagnosis of the hoist gearbox.It proves that using the cICA technology in the hoist gearbox fault feature extraction is effective and practical. It has important practical significance to ensure the safety of the coal mine hoist production.