作者：梁涛¹, 程立钦¹, 姜文², 王剑峰²

作者单位：1. 河北工业大学人工智能与数据科学学院，天津 300131;
2. 河北建投能源投资股份有限公司，河北 石家庄 050011

关键词：火力发电;故障诊断;Murphy规则;结果融合

摘要：

锅炉故障是火力发电厂的一个重要问题，它具有高温高压高耦合性的特点。针对这一问题，提出一种基于改进Murphy规则的智能融合故障诊断方法。首先利用Relief算法对锅炉的各个变量进行特征提取与选择，获得11个模型输入变量；然后利用SVM、LVQ、 PNN、BP 4种不同的分类器进行故障模型训练，并对每个模型进行性能评估；最后利用改进Murphy规则对4个分类器的结果进行融合，得到最终的故障诊断结果。运行结果证实该智能融合故障诊断方法可以有效诊断出锅炉故障，提高故障诊断的准确率，有效降低故障诊断的误报率与漏报率。

Intelligent fusion fault diagnosis method for boiler based on improved Murphy rule
LIANG Tao¹, CHENG Liqin¹, JIANG Wen², WANG Jianfeng²
1. College of Artificial Intelligence and Data Science, Hebei University of Technology, Tianjin 300131, China;
2. Hebei Jointo Energy Investment Co., Ltd., Shijiazhuang 050011, China
Abstract: Boiler failure is an important problem in thermal power plant, which has the characteristics of high temperature, high pressure and high coupling. To solve this problem, an intelligent fusion fault diagnosis method based on improved Murphy rules is proposed. Firstly, the Releif algorithm is used to select the features of each variable of the boiler, and 11 model input variables are obtained. Then four different classifiers, SVM, LVQ, PNN and BP, are used to train fault models, and the performance of each model is evaluated. Finally, the improved Murphy rule is used to fuse the results of the four classifiers to get the final fault diagnosis results. The operation results show that the intelligent fusion fault diagnosis method can effectively diagnose the boiler fault, improve the accuracy of fault diagnosis, and effectively reduce the false alarm rate and missing alarm rate of fault diagnosis.
Keywords: thermal power;fault diagnosis;Murphy rules;result fusion
2020, 46(7):133-140  收稿日期: 2019-12-15;收到修改稿日期: 2020-03-05
基金项目: 河北省科技支撑计划资助项目（17214304D，19210108D）
作者简介: 梁涛(1975-),男,河北石家庄市人,教授,博士,研究领域为大数据、人工智能与新能源
参考文献
[1] VERBERT K, BABUŠKA R, SCHUTTER B. Bayesian and Dempster-Shafer reasoning for knowledge-based fault diagnosis-A comparative study[J]. Engineering Applications of Artificial Intelligence, 2017, 60: 136-150
[2] VILEINISKIS M, REMENYTE-PRESCOTT R, RAMA D, et al. Fault detection and diagnostics of a three-phase separator[J]. Loss Prevent. Process Indust, 2016, 41: 215-230
[3] 李洋, 赵鸣, 徐梦瑶, 等. 多源信息融合技术研究综述[J]. 智能计算机与应用, 2019, 9(5): 186-189
[4] 袁海满, 吴广宁. 基于多信息融合的变压器故障诊断[J]. 高压电器, 2018, 54(9): 103-110
[5] LIU S X, ZHAO E M, ZHANG Y J, et al. Application research of multi-source information fusion technology in power network fault diagnosis[J]. Journal of Physics: Conference Series, 2019, 1187(2): 24-32
[6] 李金拓. 基于大数据的火力发电远程监督诊断的研究与应用[J]. 仪器仪表用户, 2018, 25(10): 66-69
[7] PRIYA M S, KANTHAVEL R, SARAVANAN M. Fault diagnostics on steam boilers and forecasting system based on hybrid fuzzy clustering and artificial neural networks in early detection of chamber slagging/fouling[J]. Circuits and Systems, 2016, 7(12): 4046-4070
[8] ROSTEK K, MORYTKO L, JANKOWSKA A. Early detection and prediction of leaks in fluidized-bed boilers using artificial neural networks[J]. Energy, 2015, 89: 914-923
[9] 岳军, 黄诚, 任瑞云. PNN神经网络模型在变压器故障诊断建模中的应用[J]. 自动化技术与应用, 2016, 35(10): 80-82
[10] 侯晓东, 蔡斌斌, 金炜东, 等. 基于证据距离和模糊熵的加权证据融合新方法[J]. 广西师范大学学报(自然科学版), 2015, 33(1): 45-51
[11] 张雅媛, 孙力帆, 郑国强. 基于改进Jousselme证据距离的多传感器决策融合方法[J]. 仪表技术与传感器, 2019(7): 82-88