Research On Fault Detection And Diagnosis Of Local Ventilator In Coal Mine Based On WNN

Local ventilator is the key equipment of air supply to underground excavation face in the coal mine, and the fault of local ventilator is a main factor to induce gas explosion. Therefore, it has great significance that fault detection and diagnosis of local ventilator. To ensure normal and stable work of local ventilator, by using virtual instruments technology, wavelet packet analysis, back propagation (BP) neural network technology and multidisciplinary technology. A fault detection and diagnosis system based on vibration detection is studied and developed in this dissertation, which is expected to realize effective fault diagnosis of local ventilator. The performed researches are as follows:(1) The common mechanical failure and mechanism, vibration frequency and diagnostic methods of the local ventilator are detailedly analyzed, and it is established that general research programme of vibration signals acquisition and management by using virtual instruments, wavelet packet energy extraction eigenvector, BP Artificial Neural Network (ANN) recognition. This programme could improve the automation and accuracy, reduce the cost of hardware, and shorten the software development cycle.(2) Fault diagnosis model is established through integrating of wavelet analysis and neural network. Choose db9-wavelet and definition five wavelet packet tree (WPT) for wavelet packet decomposition. The value of wavelet decomposition serves as input vector of BP ANN, while common fault as output vector. ANN model has been set up through sample test training, which is showed that network scale is moderate, computing time is short, recognition credibility is high.(3) The hardware system is designed. In accordance with the vibration signals acquisition, LC0151T sensor and USB6221 data acquisition (DAQ) card and other hardware equipments are chosen. Four measuring point distribution and sensors layout based on vibration detection are determined.(4) The software function, development and structure model are analyzed and designed. Adopting spiral model to develop software, choosing LabVIEW and MATLAB as software development platform, designing three-tiered progressive software structure have achieved. Local ventilator offline, online diagnosis, data acquisition and management, fault database maintenance, web report generation and data backup and etc. are implemented. Optimization algorithm is applied to software development to improve subroutine's reusability and efficiency. Human computer interactive interface has adopted the event trigger mechanism so that the system can respond user command in time and save computer resources.(5) Verification and performance evaluation of system are finished. Analysis of surge and foundation loosening of local ventilator has showed that experience analysis corresponds with diagnosis result. The assessment of the response time and reliability through parameters and test statistics has showed that the system has a reasonable response time and good reliability.Virtual instruments technology, wavelet packet analysis and ANN have been applied to fault detection and diagnosis of local ventilator. The prototype system that integrates signal storage management, fault diagnosis, fault database maintenance, and report generation is researched and developed. It has improved the level and efficiency of integration, automation, intelligence of local ventilator fault detection and diagnosis, and has positive significance to reduce the accident of gas explosion in underground excavation face.