基于GA-SVM的刚性罐道故障诊断

针对刚性罐道故障类型识别精度低这一难题，提出了一种基于遗传算法(genetic algorithm，GA)和支持向量机（support vector machine，SVM）的刚性罐道故障诊断方法。搭建了立井提升系统实验台，模拟2种典型的罐道故障，并采集提升容器振动加速度信号。运用经验模态分解（empirical mode decomposition，EMD）方法对振动加速度信号进行分解，选取前4个固有模态函数（intrinsic mode function，IMF），然后运用奇异值分解（singular value decomposition，SVD）方法计算出每个IMF的奇异值作为故障特征参数，将得到的故障特征参数作为SVM的训练集，通过GA参数寻优方法得到SVM关键参数c和g的最优值，并选取新的测试样本检测SVM的诊断效果。实验结果表明:基于GA-SVM的刚性罐道故障诊断方法的平均分类准确率达到93%。研究结果表明该方法能精确地识别刚性罐道的典型故障类型，为立井提升系统等非线性非平稳复杂系统的故障诊断提供一种通用可行的解决方法。

Fault diagnosis for rigid guide based on GA-SVM

In order to improve the identification accuracy of rigid guide fault type, a method based on genetic algorithm (GA) and support vector machine (SVM) was proposed to solve the fault diagnosis problem. Vertical shaft lifting system experimental platform was set up to simulate two typical types of rigid guide fault, and the vibration acceleration signal of the lifting vessel was collected. The vibration acceleration signal was decomposed by empirical mode decomposition (EMD), and the singular values of first four intrinsic mode functions (IMF) were calculated by singular value decomposition (SVD) method as the fault characteristic parameters. The fault characteristic parameter was used as the training set of the SVM, and the optimal values of c and g of the SVM were obtained by the GA parameter optimization method. The new test samples were selected to detect the diagnostic effect of SVM. The experimental results showed that the average classification accuracy of rigid guide fault diagnosis method based on GA-SVM reached 93%. This method can accurately identify the typical fault types of rigid guide, and provide a general and feasible solution for the fault diagnosis of nonlinear and non-stationary complex systems such as vertical shaft lifting system.