针对储罐罐底在线声发射腐蚀检测评价过程中事件数集中区域难以确定的问题,提出一种基于自适应仿射传播的罐底声发射腐蚀评价方法。该方法采用自适应扫描偏向参数空间与调整阻尼因子获得最优聚类数目,确定声发射事件数集中区域,并按相关检测标准对罐底的腐蚀状态级别进行评价。数据分析结果表明:该方法克服了传统K均值算法无法准确获知聚类数目的缺点,精确地确定了罐底事件数集中区域,有效提高了罐底腐蚀评价的准确率。
A method of corrosion evaluation on tank bottom based on adaptive affinity propagation(adAP)algorithm is proposed for finding acoustic emission(AE)events clustering regions.The number of clusters is searched for finding the optional clustering solution by adaptive scanning of parameters space and adaptively adjusting damping factors.In addition,the AE events focusing areas are determined by the clustering results and the level of tank bottom's corrosion state is evaluated according to the acoustic emission testing standards.The test results of storage tanks show that the limitation of the traditional K-means algorithm which does not know the clusters number is overcome by the method and the clustering regions of AE events are accurately found.Thus the accuracy of corrosion evaluation on storage tank bottom is improved.