为了提高工业生产过程故障检测的精度,保证产品的质量和生产过程的安全,提出了一种基于算术优化算法(arithmetic optimization algorithm,AOA)优化支持向量机(support vector machine,SVM)的故障检测方法。首先,对工业过程中产生的数据进行标准化处理;然后,将处理后的数据作为训练样本建立SVM模型,同时采用算术优化算法对支持向量机中的惩罚参数C和核函数参数g进行优化,通过多次迭代对模型进行训练,建立最佳故障检测模型;最后,将测试数据导入建立的故障检测模型中进行故障检测。将提出的AOA-SVM方法应用于田纳西-伊斯曼过程进行实验验证,并与传统SVM、灰狼优化算法优化的支持向量机(GWO-SVM)方法进行比较,该研究提出的模型具有更高的准确率。实验仿真结果表明,提出的AOA-SVM故障检测模型具有更好的表现。

In order to improve the accuracy of fault detection in industrial production process,and ensure product quality as well as production process safety,a fault detection method is proposed based on arithmetic optimization algorithm optimized support vector machine.Firstly,the data generated in industrial process is standardized.Next,the pre-processed data is utilized as training samples to create an SVM model.Meanwhile,the penalty parameter C and the kernel function parameter g in the Support Vector Machine are optimized by using Arithmetic Optimization Algorithm.The model is trained through multiple iterations to establish the optimal fault detection model.Finally,the test data are imported into the established fault detection model for fault detection.The AOA-SVM method proposed in this study is employed for experimental validation on the Tennessee Eastman process.It is then compared with the conventional SVM method and Grey Wolf Optimization algorithm-optimized Support Vector Machine method.It is indicated that the proposed model in the study has higher accuracy.Experimental simulation results show that the proposed AOA-SVM fault detection model has better performance.

李鑫妮(1998-),女,硕士生,研究方向为多元统计建模和故障诊断在工业过程的应用,lixinni1564@163.com;王亚君(1978-),女,博士,教授,研究方向为工业过程监测方法研究、过程控制,通讯联系人,wyj_lg@163.com。