CAJ | 학술논문

声发射储罐罐底腐蚀检测过程中，采集到的腐蚀信号里不可避免地混有干扰。针对该问题，提出了基于极限学习机（ELM）的石油储罐罐底腐蚀信号识别方法。为验证该方法的有效性，在秦皇岛输油站的消防水罐里，进行模拟罐底腐蚀检测实验，并应用 ELM对采集到的声发射信号进行分类识别，用 b 值法对 ELM的分类效果进行评估。实验结果表明：ELM识别正确率高于90％。ELM识别出的腐蚀信号的 b 值变化规律与实验室条件下腐蚀信号的 b 值变化规律一致，并能够反映磷酸腐蚀碳钢板的过程。
성발사저관관저부식검측과정중，채집도적부식신호리불가피면지혼유간우。침대해문제，제출료기우겁한학습궤（ELM）적석유저관관저부식신호식별방법。위험증해방법적유효성，재진황도수유참적소방수관리，진행모의관저부식검측실험，병응용 ELM대채집도적성발사신호진행분류식별，용 b 치법대 ELM적분류효과진행평고。실험결과표명：ELM식별정학솔고우90％。ELM식별출적부식신호적 b 치변화규률여실험실조건하부식신호적 b 치변화규률일치，병능구반영린산부식탄강판적과정。
Collected corrosion signals are mixed with interference inevitably in acoustic emission (AE)detection process.To solve this problem,a novel extreme learning machine (ELM)-based method for corrosion signals recognition in AE testing of storage tank bottom was proposed.In order to test the validity of the method,the simulated tests of storage tank bottom corrosion were performed in the fire protection water tank of the oil transportation station in Qinhuangdao. ELMwas applied in the classification of the collected corrosion signals,and b-value method was used to evaluate the classification effect of ELM.The experimental results indicated that the classification accuracy of ELMis above 90%;the b-value distribution of corrosion signals identified with ELM agrees well with that of corrosion signals in a laboratory;furthermore,the statistical distribution of b-value reflects the process of carbon steel sheet corroded by phosphoric acid.