Study On Fusion Cluster Method Of Acoustic Emission Signal Of Tank Bottom Corrosion

Because of the corrosion of metal tanks in storage, the corrosion of tank bottom is detected by acoustic emission technology, which is widely used in nondestructive testing. The technique uses sensors to collect acoustic emission signal processing and analysis, which comprises a noise removal to exclude interference and clustering of different acoustic emission source type to distinguish, and the overall evaluation of the tank bottom corrosion accordingly.The in laboratory and field on-line detection research based on the three aspects of the research: first, using wavelet packet to signal of wavelet transform and denoising; secondly, to identify different clustering algorithm is optimized; finally collected in the experiment of different types of data clustering analysis, data collection and field compared.In order to simulate the corrosion condition of the actual tank bottom, the steel corrosion experiment of the PCI-2 acoustic emission acquisition system was used to extract three kinds of acoustic emission signal samples. The training sample signals into the algorithm, the establishment of data model, the calibration calculation and tag clustering center. An improved classification algorithm based on kernel validity index is proposed, and the initial cluster centers are selected by the fuzzy subtraction clustering. Corrosion Acoustic emission signal is a non-stationary random signal, difficult to the matrix parameters directly applied to clustering algorithm calculation, parameters are mapped to high dimensional feature space by using improved Gaussian kernel fuzzy clustering algorithm(KFCM), to overcome the shortcomings of the signal feature is not obvious. In order to get better partitioning result.In the sample data processing stage, in order to ensure that the sample data to be accurately labeled, using the differential algorithm to calculate the membership of the cluster center to ensure the accuracy of the data. In the experiment, it is proved that the method is feasible.By comparing the field online detection data with the laboratory data, the method can identify three types of tank bottom corrosion damage. It provides an effective method and basis for corrosion testing of oil tank bottom.