基于自适应步长FOA-SVM算法的卡泵故障诊断

Research on Fault Diagnosis of Stuck Pump Based on Adaptive Step Size FOA-SVM Algorithm

为了提高机采井卡泵故障诊断精度,提出一种基于自适应步长FOA-SVM混合算法模型的机采井卡泵诊断方法。在支持向量机对示功图诊断分类的基础上,引入改进的自适应步长果蝇优化算法(AS_FOA)对SVM的惩罚因子和核函数参数进行寻优,避免人为选择参数的盲目性。为了实现果蝇优化算法的全局与局部寻优能力的平衡,应用自适应步长方法对其进行改进,使果蝇算法能够根据上一代的适应度值和当前迭代次数来自适应改变果蝇个体搜索步长。通过采油厂真实示功图数据进行仿真实验,比较AS_FOA、FOA、GA三种算法在支持向量机参数寻优中的性能。实验结果表明,AS_FOA收敛速度更快,寻优能力更佳。与其他算法相比,AS_FOA-SVM混合算法模型在卡泵故障诊断中准确率更高,泛化能力更强。

In order to improve the fault diagnosis accuracy of stuck pump,a fault diagnosis method based on adaptive step size FOA-SVM hybrid algorithm model is proposed.Based on the classification of indicator diagram diagnosis by support vector machine,an improved adaptive step size drosophila optimization algorithm(AS_FOA)is introduced to optimize the penalty factor and kernel function parameters of SVM,so as to avoid the blindness of artificial selection of parameters.In order to achieve the balance of global and local optimization ability of drosophila optimization algorithm,the adaptive step method is used to improve it,so that the drosophila algorithm can adapt to change the individual search step according to the fitness value of the previous generation and the number of current iterations.Through the simulation experiment of the real indicator diagram data of oil production plant,the performance of AS_FOA,FOA and GA in the parameter optimization of support vector machine is compared.The experiment shows that AS_FOA has faster convergence speed and better optimization ability.Compared with other algorithms,AS_FOA_SVM hybrid algorithm model has higher accuracy and stronger generalization ability in the fault diagnosis of stuck pump.