基于支持向量机的非线性工业过程故障检测与预测研究

发布时间：2018-09-08 21:10

【摘要】：随着全球工业智造的大行其道,人们对工业生产系统的稳定性、工业生产运行过程的经济性及产品质量等各个方面的要求愈加严格。工业自动化市场规模的急剧扩张使得现代工业系统和设备愈加复杂,要保证大型复杂工业系统正常运行,需要面临诸多挑战。因此,为实现对工业过程实时有效地监控与检测,确保生产过程的安全可靠,利用支持向量机方法对对非线性工业过程的大数据进行故障检测与预测具有重要的理论价值和实际意义。本文分析了支持向量机的基础理论,推导了该算法的建模原理和过程。针对非线性工业过程中大数据的故障检测和预测,首先采用交叉验证优化方法对支持向量机进行核参数优化。然后分别利用支持向量机、主成分分析法和增强偏最小二乘法对连续搅拌釜式加热器过程进行故障检测,并对各个算法的故障检测结果进行分析比对,实验结果表明,SVM分类器在实际复杂工业过程中具有优异的预测能力和理想的运行时间。针对非线性工业过程的故障预测问题,通过学习半监督学习方法,利用孪生支持向量机和改进算法(S~4VM)对工业过程的故障状态进行有效地预测分析。S~4VM对初始参数设定值不敏感,能同时考虑多个候选大边界低密度分界线,并在最坏情况下优化标签分配,在解决非线性工业过程大数据的故障预测的问题上表现优异。
[Abstract]:With the popularity of global industrial intelligence, the requirements for the stability of industrial production system, the economy of industrial production process and the quality of products are becoming more and more stringent. The rapid expansion of industrial automation market makes modern industrial systems and equipment more complex. To ensure the normal operation of large-scale complex industrial systems, many challenges need to be faced. Therefore, in order to realize the real-time and effective monitoring and detection of the industrial process and ensure the safety and reliability of the production process, The support vector machine (SVM) method is of great theoretical value and practical significance for the fault detection and prediction of big data in nonlinear industrial processes. In this paper, the basic theory of support vector machine is analyzed, and the modeling principle and process of the algorithm are deduced. Aiming at the fault detection and prediction of big data in nonlinear industrial process, the kernel parameters of support vector machine are optimized by cross-validation optimization method. Then, support vector machine, principal component analysis and enhanced partial least square method are used to detect the faults of the continuous stirred tank heater, and the results of each algorithm are analyzed and compared. The experimental results show that the SVM classifier has excellent prediction ability and ideal running time in complex industrial processes. In order to solve the problem of nonlinear industrial process fault prediction, by learning semi-supervised learning method, twinning support vector machine and improved algorithm (S~4VM) are used to effectively predict the fault state of industrial process and analyze that Sch _ 4VM is insensitive to the initial parameter setting value. It can simultaneously consider multiple candidate large boundary low density boundaries and optimize label assignment in the worst case. It is excellent in solving the problem of big data's fault prediction in nonlinear industrial processes.
【学位授予单位】：渤海大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP277

【参考文献】