数据分析与BP神经网络相结合的乙烯装置智能故障诊断系统

摘要/Abstract

摘要： 针对化工生产流程中各单元具有强关联性、连续性、复杂性的特点，以某乙烯生产装置为对象进行Aspen仿真，提取数据并进行灵敏度分析，发现设备故障可以通过分馏塔再沸器和冷凝器的热负荷变化来反映；进而，以脱甲烷塔、脱乙烷塔和乙烯精馏塔为例建立动态模型，把引起热负荷变化的温度、压力、甲烷含量和进料流量4种可观测参数数据集进行故障分类编码，再利用3层BP神经网络建立热公用工程（主要指蒸汽）智能诊断系统模型。对不同可观测参数及其组合数据各选取10 000组数据样本进行训练和验证，在获得阶段性诊断结果后，综合设备热负荷变化得到完整的装置故障情况。测试集样本验证结果显示，所建BP神经网络诊断模型的准确率较高，可达到99.75%。实际应用结果表明，该故障诊断系统在实际操作中能够快速有效地判断出设备故障诱因。

关键词: 公用工程, 神经网络, 乙烯装置, 故障诊断

Abstract: In view of the characteristics of the strong correlation, continuity and complexity of each unit in the chemical production process, the Aspen simulation was carried out on an ethylene production unit. The data were extracted and the sensitivity analysis was carried out. It was found that the equipment failure could be reflected by the change of heat load of reboiler and condenser in fractionating column. Then, a dynamic model was established by taking the demethane column, deethane column and ethylene rectifying column as examples. The data sets of four observable parameter, temperature, pressure, methane content and feed flow, which caused the change of thermal load, were classified and coded, then the intelligent diagnosis system model of thermal utility (mainly steam) was established by using three-layer BP neural network. 10 000 samples of different observable parameters and their combined data were trained and verified. After the periodic diagnosis results were obtained, the complete fault condition of the integrated equipment was determined. The results of test set sample verification showed that the accuracy of BP neural network diagnosis model was high, reaching 99.75%. The practical application results showed that the fault diagnosis system could quickly and effectively judge the cause of equipment failure in the actual operation.

Key words: utility, neural network, ethylene plant, fault diagnosis

常亚娜武锦涛代玉强. 数据分析与BP神经网络相结合的乙烯装置智能故障诊断系统[J]. 石油炼制与化工, 2023, 54(6): 97-104.

参考文献

[1] A Perspective on Smart Process Manufacturing Research Challenges for Process Systems Engineers [J]. Engineering, 2017, 3(2): 161-5,76-81.
[2] QIAN F. Smart Process Manufacturing Systems: Deep Integration of Artificial Intelligence and Process Manufacturing [J]. Engineering 2019, 5(6): 981.
[3] CHEN P, TOYOTA T, LIN Y, et al. Failure diagnosis of machinery by self-reorganization of symptom parameters in time domain using genetic algorithms [J]. The International Journal of Intelligent Control and Systems, 1999, 3(4): 571-85.
[4] LO C H, WONG Y K, RAD A B. Intelligent System for Process Supervision and Fault Diagnosis in Dynamic Physical Systems [J]. IEEE Transactions on Industrial Electronics, 2006, 53(2): 581-92.
[5] PAIXAO K. Feature engineering for machine learning and data analytics [J]. Computing reviews, 2019, 60(6): 242.
[6] SHANG C, YOU F. Data Analytics and Machine Learning for Smart Process Manufacturing: Recent Advances and Perspectives in the Big Data Era [J]. Engineering, 2019, 5(6): 1010-6.
[7] 王翔. 基于深度学习的故障诊断算法研究 [D]; 浙江理工大学, 2020.
WANG X. Research on Deep Learning Base Fault Diagnosis Algorithms [D]; Zhejiang University of Science and Technology, 2020.
[8] 齐长兴, 毕义明, 李勇, 等. 改进的萤火虫算法优化BP神经网络及应用 [J]. 现代防御技术, 2018, 46(05): 32-8.
QI C X，BI Y M，LI Y， et al. BP Neural Network Optimized by Approved Firefly Algorithm and Its Application. [J]. Modern Defence Technology, 2018, 46(05): 32-8.
[9] 田雨波, 潘朋朋. 粒子群算法优化神经网络结构的研究 [J]. 现代电子技术, 2011, 34(04): 110-2.
TIAN Y B, PAN P P. Particle Swarm Algorithm Used for Optimizing Neural Network Structure [J]. Modern Electronics Technique, 2011, 34(04): 110-2.
[10] 衷路生, 夏相明. 基于深度残差网络的化工过程故障诊断 [J]. 过程工程学报, 2020, 20(12): 1483-90.
ZHONG L S, XIA X M. Fault diagnosis for chemical processes based on deep residual network (in Chinese). [J]. The Chinese Journal of Process Engineering, 2020, 20(12): 1483-90.
[11] 曹美, 崔国民, 肖媛, 等. 换热网络优化中换热单元均布处理的周期性影响分析 [J]. 热能动力工程, 2020, 35(06): 15-21+142.
CAO M，CUI G-M，XIAO Y，et al． Periodic analysis on even distribution of heat exchanger units in heat exchanger network optimization［J］. Journal of Engineering for Thermal Energy and Power，2020，35( 6): 15-21+142．
[12] 王世豪, 田一彤, 李绍军. 基于双层优化策略的柔性换热网络同步优化方法[J]. 高校化学工程学报, 2021, 35(05): 905-14.
WANG S H, TIAN Y T, Li S J. A simultaneous synthesis based on a bi-level optimization strategy for flexible heat exchanger network [J]. Journal of Chemical Engineering of Chinese Universities, 2021, 35(05): 905-14.
[13] 王子健, 车景华, 宋昭峥, 等. AEA在常减压装置换热网络优化中的应用分析 [J]. 石油石化绿色低碳, 2021, 6(04): 17-21+51.
WANG Z J，CHE J H，SONG Z Z, et al. Application Analysis of AEA Optimization in Heat Exchange Network of AVU [J]. Green Petroleum & Petrochemicals, 2021, 6(04): 17-21+51.
[14] 齐少宁. 煤制甲醇基于温差贡献值〖△T〗_con的能效优化 [D]; 天津大学, 2014.
QI S N. Energy efficiency optimization of coal to methanol process based on temperature difference contribution〖△T〗_con [D]; Tianjin University, 2014.
[15] Mechanical Properties Prediction of the Mechanical Clinching Joints Based on Genetic Algorithm and BP Neural Network [J]. Chinese Journal of Mechanical Engineering, 2009, 22(1): 36-41.
[16] XIAO M, ZHANG W, Wen K, et al. Fault Diagnosis Based on BP Neural Network Optimized by Beetle Algorithm [J]. Chinese Journal of Mechanical Engineering, 2021, 34(6): 252-61.
[17] 李仑.乙烯装置脱甲烷系统中甲烷分离浅析 [J]. 乙烯工业, 2021, 33(03): 33-7+4.
LI L. Discussion on Methane Separation in Demethanation System of Ethylene Plant [J]. Ethylene Industry, 2021, 33(03): 33-7+4.
[18] 张黎骅, 赵超, 李庆东, 等. 基于Matlab6.5的归一化二阶系统单位阶跃响应的模拟演示 [J]. 中国农机化, 2005, (03): 35-8.
ZHANG L H, ZHAO C, Li Q D, et al. Computer Simulation of Fundamental Second-order System’s Response to Unit Step Function Base on Matlab6.5 [J]. Chinese Agricultural Mechanization, 2005, (03): 35-8.