Design And Implementation Of The Wind Turbine Status Monitoring And Fault Diagnosis System

With the deepening of the energy crisis, countries are vigorously developing theclearer energy technology. As a relatively mature renewable energy technology, thewind power develops rapidly in recent years with great growth in both the powergeneration of the wind turbine and new installed capacity. However, at the same timeof the rapid progress of wind power, it also brings a series of challenges which ishighlighted in the frequent quality problems of wind turbine, which seriously affectthe normal production and electricity generation, causing huge economic losses.Therefore it becomes particularly necessary and urgent to develop the onlinemonitoring and fault warning system for wind turbine.As a large rotary machine, wind turbine is for long term in the wild field withpoor working environment, so it’s fault prone. As the core component of the drivingsystem in the wind turbine, the gearbox continuously bears the alternating impactforce and heavy loads which makes it the high site of the fault. Therefore this paperfocuses on the study of the fault diagnosis problems of the wind turbine gearbox.Common failures of gearbox are gear wheel brokenness, damage to the inner andouter bearing rings, the shaft imbalance, and so on. These failures will cause thevibration signal modulation of the gearbox. The fault in different parts will causedifferent vibration signal characteristics frequency, and also the varying degrees offault in the same part will cause varying degrees of vibration amplitude. Through theanalysis of vibration signal of the fault, the health status of the gearbox can beappropriately assessment. The structures and force condition of the wind turbinegearbox is very complex, which leads to the great complexity in gaining the vibrationsignal of the gearbox, so the completion of the equipment fault diagnosis often needsprofessional technicians, subject to the experience and skills restriction of differenttechnicians. There exit big differences of the judgment from different ones, also with low efficiency. To solve the above problems, this paper intends to adopt the intelligentfault diagnosis technology based on BP neural network. The main research content ofthis paper is divided into the following aspects:Firstly, this paper is to study the gearbox structure of the wind turbine and toclarify the mechanism of vibration generation in the gearbox. It mainly focuses on thestudy of vibration signal forms and vibration characteristics of the gear and bearing inthe gearbox when they go fault.Secondly, due to the complexity of the force changes in the wind turbine gearbox,the vibration signal of the gearbox typically belongs to non-stationary signals. Thispaper studies the processing algorithms for the vibration signal, including thealgorithm of time domain analysis, frequency domain analysis and time-frequencyanalysis. Through the comparison of their characteristics, it ultimately define that thefault signal characteristic value of the gearbox be abstracted using the method ofwavelet packet decomposition. A failure of the gearbox will inevitably lead to changesin the vibration energy and different failure will cause changes in different frequencybands energy. Based on this characteristic, the energy changes in different frequencybands are calculated using the method of wavelet packet decomposition.Thirdly, to improve the efficiency and accuracy of the gearbox fault diagnosis,this paper studies the application of BP neural network in this regard. BP neuralnetwork is a kind of mature feed forward network with strong nonlinear mappingability and relatively simple structure, but the network training and learning effect isvery good. The system combines the fault characteristics abstracted through thewavelet packet method, and send the characteristic values to the BP neural networkfor training and diagnosis, through which the intelligent fault diagnosis can berealized.Fourthly, a set of complete online monitoring and fault diagnosis systemsoftware is developed based on the existing gearbox experiment rig. Based on theconcept development of the virtual instrument, the software forms good humanmachine interface using LabVIEW programming language, and completes a series offunctions such as the data collection, transmission, processing and preservation of the vibration signal.This paper studies the principle of the gearbox fault, and determines the faultprocessing algorithms based on the wavelet packet decomposition and the BP neuralnetwork as the core. A set of gearbox online monitoring and fault diagnosis systemsoftware is developed using the mixed programming technology of LabVIEW andMATLAB. The system can make accurate judgment about the inner fault type of thegearbox which is with certain practicality.