基于CEEMDAN与CBBO-SVM的汽轮机转子故障诊断研究  

Study of the Diagnosis of the Faults Occurred to the Rotor of a Steam Turbine Based on the CEEMDAN(complete ensemble empirical mode decomposition with adaptive noise) and CBBO-SVM( biogeography-based optimization with chaos-supporting vector machine)

作　　者：石志标[1] 葛春雪 曹丽华 赵军 

机构地区：[1]东北电力大学机械工程学院,吉林吉林132012 [2]雅砻江流域水电开发有限公司,四川成都610051

出　　处：《热能动力工程》2018年第1期69-74,共6页Journal of Engineering for Thermal Energy and Power

基　　金：国家自然科学基金(51576036);吉林省科技发展计划项目(20100506)~~

摘　　要：为了提高汽轮机转子故障诊断的准确率和识别效率，提出了一种基于CEEMDAN（自适应噪声完备集合经验模态分解）和CBBO（混沌生物地理学优化算法）优化SVM（支持向量机）相结合的故障诊断方法。首先利用CEEMDAN对转子振动信号进行分解，提取PE（排列熵）作为故障特征值，并构造特征向量；其次将混沌理论引入到BBO（生物地理学优化算法）中，得到CBBO，通过CBBO优化SVM得到诊断模型的最优参数。最后通过ZT-3转子试验台模拟汽轮机转子故障，利用得到的4种状态下的试验数据验证优化模型的有效性与先进性。结果表明：CBBO优化SVM模型可以准确、高效地对汽轮机转子进行故障诊断；与CPSO（混沌粒子群算法）优化SVM模型相比，该方法的故障诊断准确率和识别效率更高。To enhance the accuracy and identification efficiency to diagnose any faults in the rotor of a steam turbine, proposed was a method for diagnosing any faults based on a combination of the CEEMDAN and CBBO-SVM. Firstly, the CEEMDAN was employed to perform a decomposition of the vibration signals from rotors, extract the permutation entropy （PE） as the fault characteristic values and create the characteristic vectors. Subsequently, the chaos theory was introduced into the BBO to obtain a CBBO-based algorithm and the optimum parameters of the diagnostic model by optimizing the SVM through the adoption of the CBBO. Finally, the fault of the rotor of the steam turbine was simulated on the ZT-3 rotor test rig and the test data obtained in the four states were utilized to verify the effective ness and advanced nature of the optimization model. It has been found that to optimize the SVM model by using the CBBO can accurately and efficiently diagnose any faults of steam turbine rotors. Compared with CPSO （ chaos particle swarm op- timization）-based SVM model, the method in question has an even higher accuracy and identification efficiency to diagnose any faults.

关 键 词：汽轮机转子 故障诊断 支持向量机 混沌生物地理学优化算法 自适应噪声完备集合经验模态分解 

分 类 号：TK267[动力工程及工程热物理—动力机械及工程]