Research On Magnetic Memory Detection Signals Of Belt Steel-cords Using Wavelet Analysis

Currently, there are many methods for detecting belt steel-cords at home and abroad, but it is not high accuracy to get and analyze the failure signal in the early stage. Especially,there are many difficulties in signal processing and characteristic quantity extraction,which made these early methods fail in achieving the desired effect. Generally, failures were usually identified and eliminated only depending on periodic downtime inspections. Signals often contain a large number of time-varying,short-term shocks and sudden nature elements, so the traditional approaches such as Fourier transform is not conducive to analyze and determine fault signal. Therefore, it is necessary to find a new and more effective method for early diagnosis of belt steel-cord.In this paper, begining with the structural characteristics of steel cord conveyor belt, it was pointed out that fault of steel core in belt was caused by twitching and fracture, and then the mechanism and characteristics of occurrence fault were analyzed in detail. Through researching and analyzing fault detection and signal processing defects technology about steel cord belt, a new method of diagnosis to the early fault signals of magnetic memory testing techniques was proposed with wavelet analysis.First, the basic principles of magnetic memory testing, features and its advantages in fault detection were deeply researched, and influencing factors and detection principle of magnetic memory were discussed in detail. By comparing magnetic memory with traditional methods, it was demonstrated the feasibility of magnetic memory testing of tape steel cord. Moreover, from both diagnostic theory and features of fault signals, it was illustrated that fault signals by wavelet analysis is inherent requirement in the diagnosis process.Second, based on comparison of wavelet transform with Fourier transform and fast Fourier transform in this paper, the features of wavelet analysis was summarized, and wavelet transform and multi-resolution was discussed. Through using different methods for de-noising contrast of simulation signal, it was finally proved that the soft thresholding method has the best noise reduction and the example was used to prove the singularity detection principle of wavelet transform.Finally, it is focused on that the collected magnetic memory signal was processed. In the process of signal analysis using MATLAB as a platform, different wave functions were used to eliminate noise in the detection signal, and the best de-noising function was obtained after comparison of the results. The same tested signal was decomposed by different scales, and four soft threshold methods were selected to quantize the high frequency coefficients by a threshold, then low-frequency coefficients and a threshold quantized frequency coefficient were extracted to complete wavelet reconstruction. By comparison, the threshold selection method for magnetic memory signals was determined finally. Ultimately, by analyzing filtered signal curve of magnetic memory testing, combining the signal peak to peak, difference scale of adjacent signals, signal characteristic parameters were extracted and that the signal point of failure was pointed out. And test results was comparied with its method of X-ray machine, detection effect to the early fault is remarkable.The innovation of this topic is that both magnetic memory techniques and processing signal technology by wavelet were used in the steel cord for belt failure detection. It can extract early fault features not obtained with other methods. It provides a feasible and effective new method for early fault detection technology of belt steel cord.