Fault Detection Methods Of Machineries Based On Spectrum Zoom Analysizing Of Audio Signals

With higher and higher technical index requirement of precision, load intensity, reliable operation time, mechanical equipment fault detection and diagnosis needs more urgent.Therefore,the theory and methods of fault detection and diagnosis have become a research hotspot.In numerous fault detection methods, based on the analysis of the audio signal non-contact fault detection became main technical mean in special applications. The audio signal includes the mechanical failure audio informations. In other words, the mechanical failure can be reflected on some traits of the audio signal informations. The method mainly depends on the audio signal feature extraction for fault detection, audio signal feature extraction accuracy and effectiveness becoming the main technical factor which directly affects the primary technology of fault detection and diagnosis. Therefore, the audio signal noise suppression and spectrum zoom technique have become the key technology of this kind of detection method.Regarding the mechanical equipment fault detection as the research background, treating non-contact detection method based on audio signal feature extraction technology as the technical means, putting the audio signal spectrum zoom technique as the research object,combined with the characteristics of existing audio signal spectrum room technology, the paper focused on audio signal spectrum room optimization method, error spectrum room correction method which applied to precision mechanical equipment fault detection, and verified the methods through theoretical modeling and simulation software in the proposed.The main research contents of this paper are as follows:Firstly, analyze two classic methods of the audio signal frequency spectrum zoom,namely the Chirp-Z Transform (CZT) algorithm, Multiple Modulation Zoom spectrum Fourier transform algorithm (Zoom-FFT, ZFFT) technology and performance of the algorithm,and introduces a kind of modified ZFFT algorithm based on band-pass filter.Secondly, build software simulation model of audio signal spectrum zoom method model based on CZT and improved ZFFT, analyzing the technical performance of the two methods.The reason which causes data errors of the above two methods for audio signal spectrum thinning is focused on study, combined with the audio signal characteristics of precision bearing fault generation, the error correction method based on Hanning window of three point convolution method proposed, and put forward the scheme based on the error correction and ZFFT audio signal spectrum subdivision, simulating and analyzing the performance of the scheme.Finally,to verify feasibility and technical performance of the scheme,respectively,based on the MATLAB platform and (FPGA platform)Verilog language, construct the software simulation model of the proposed audio signal frequency spectrum room scheme with the corresponding simulation analysis.