凡 简¹， 窦明莹¹， 赵 飞¹， 马存志²， 张如昕¹， 金云峰²， 李舒月¹， 张永民¹

1. 中国石油大学（北京）重质油全国重点实验室，北京 102249；

2. 南京天洑软件有限公司，江苏 南京 211100

引用格式：

凡简， 窦明莹， 赵飞， 等. 不同机器学习模型的流化床状态识别和故障诊断效果［J］. 中国粉体技术， 2025， 31（6）： 1-12.

FAN Jian， DOU Mingying， ZHAO Fei， et al. Effectiveness of different machine learning models for state recognition and fault  diagnosis in fluidized beds［J］. China Powder Science and Technology， 2025， 31（6）： 1−12.

DOI：10.13732/j.issn.1008-5548.2025.06.009

收稿日期： 2025-02-13， 修回日期： 2025-03-27， 上线日期： 2025-05-29。

基金项目： 国家自然科学基金项目，编号： 22308377； 中核集团青年英才项目，编号： KY202212。

第一作者简介：凡简（2000—），女，硕士研究生，研究方向为多相反应过程强化。E-mail：jian_fan0113@163. com。

通信作者简介：张永民（1978—），男，教授，博士生导师，青年学术拔尖人才、 教育部新世纪优秀人才，研究方向为多相反应过程强化及装备。E-mail：zhym@cup. edu. cn。

摘要： 【目的】验证机器学习方法应用于气固流化床状态识别和故障诊断的可行性，并进一步对比不同机器学习模型的优劣。【方法】　针对流化床常见的内部颗粒结块、分布板堵塞2种类型的故障，考察不同机器学习模型用于流化床状态识别和故障诊断的可行性； 考察半径近邻回归模型（radius nearest neighbor， RNN）、 支持向量回归（support vector regression， SVR）、 K近邻算法（K-nearest neighbor， KNN）和随机森林算法（random forest， RF）4种传统机器学习模型与智能算法AI-agent在故障诊断方面的应用，利用不同运行状态下测量得到的实验室小型流化床动态压力信号进行不同模型预测质量的评判。【结果】　机器学习模型可以通过测得的动态压力信号进行流化床的状态识别与故障诊断，在所选择的传统模型中，KNN具有最高的准确率和区分率，分别达到93. 75%和81. 25%，自研算法AI-agent的也展现出优良的诊断效果。【结论】　通过对比5类较为主流的机器学习模型，确定针对流化床状态监测与故障诊断的最优模型KNN。

关键词： 流化床；智能算法；压力脉动；故障诊断

Abstract

Objective　The complex gas-solid flow in fluidized beds often leads to operational failures such as particle agglomeration， distribution plate clogging， gas short-circuiting， and hydrodynamic instability， jeopardizing production safety and economic benefits.Compared to the machine learning （ML） methods， manual monitoring and signal processing in traditional approaches have limitations in signal extraction and multi-type fault differentiation. This study evaluates the feasibility of ML methods for state identification and fault diagnosis in gas-solid fluidized beds， benchmarking and optimizing typical ML models using the built-in AI model of DTEmpower. The research results provide insights into the development of artificial intelligence-based stabilization techniques for industrial fluidized bed systems.

Methods　A small cylindrical bubbling bed cold mold was constructed with transparent glass walls to visualize flow dynamics.The flow characteristics of fluid catalytic cracking （FCC） particles under normal working conditions， particle agglomeration  （10%-50% by volume）， and distribution plate blockage（ 12. 5%-62. 5% by area ） were simulated. Transient pressure pulsation signals were collected， and bed eigenvalues were calculated based on pressure sensor data. For fault diagnosis， four ML models， k-nearest neighbors （KNN）， random forest （RF）， support vector regression （SVR）， and radius nearest neighbor  （RNN）， as well as the AI-agent of DTEmpower， were selected based on their characteristics and model features. Five types of models were constructed and their model performance was validated and compared.

Results and Discussion　As particle agglomeration worsened， the bed pressure drop and density decreased steadily， while the distribution plate pressure drop fluctuated， showing an overall downward trend. Distribution plate clogging increased plate pressure drop but decreased bed pressure drop and density. The trend variations could be used to determine the operational status of fluidized beds. An experienced operator could identify simple faults from trend data but not complicated faults. To achieve accurate identification of fault category and severity under conditions of particle agglomeration and distribution plate blockage， ML methods should be used. For known fault severities（ limited domain）， the AI-agent model and KNN achieved the highest accuracy（ 88. 89%）， followed by RF（ 88. 67%） and SVR（ 19. 44%）. For unknown fault severities（ unqualified domain）， RF had the highest accuracy（ 100%）， followed by AI-agent（ 95. 84%）， SVR（ 87. 5%）， and RNN（ 87. 5%）. Overall， the KNN model demonstrated superior accuracy（ 93. 75%） and differentiation rate（ 81. 25%）， while the AI-agent also excelled in fault diagnosis under comparison.

Conclusion　This study offers insights into the development of AI-based technologies in operational stability solutions for indus⁃

trial fluidized bed systems. Through a comprehensive comparison of five mainstream ML models， the optimal model for condition monitoring and fault diagnosis in fluidized beds was identified. The proposed method can distinguish the differences between particle agglomeration and distribution plate blockage while quantitatively assessing failure severity， significantly enhancing operational safety. It provides a foundation for industrial-scale state identification and fault diagnosis systems based on actual industrial data. However， it should be noted that this study was conducted at a laboratory scale with simplified operating conditions and limited parameter variations. Future work should focus on applying these methods to industrial-scale applications. Historical DCS data should be employed， and specific reaction conditions and mechanisms should be considered， which will require specialized model adjustments.

Keywords： fluidized bed； intelligent algorithm； pressure pulsation； fault diagnosis

参考文献（References）

［1］PARISIEN V， PJONTEK D， MCKNIGHT C A， et al. Impact of catalyst density distribution on the fluid dynamics of an ebullated bed operating at high gas holdup conditions［J］. Chemical Engineering Science， 2017， 170： 491-500.

［2］周泉， 姜从斌， 崔坤达， 等 . 斜分布板对鼓泡流化床流动特性的影响与数值模拟［J］. 中国粉体技术， 2020， 26（2）：13-19.

ZHOU Q， JIANG C B， CUI K D， et al. The influence of inclined distribution plate on the flow characteristics of bubble fluidized bed and numerical simulation［J］. China Powder Science and Technology， 2020， 26（2）： 13-19.

［3］刘阿珍， 张卫义， 李汉勇， 等. 液-固流化床颗粒流态化特性实验研究［J］. 中国粉体技术， 2021， 27（3）： 43-49.

LIU A Z， ZHANG W Y， LI H Y， et al. Experimental study on the fluidization characteristics of liquid-solid fluidized bed particles［J］. China Powder Science and Technology， 2021， 27（3）： 43-49.

［4］赵贵兵， 阳永荣， 候琳熙. 流化床发射机理及其在故障诊断中的应用［J］. 化工学报， 2001， 52（11）： 941-943.

ZHAO G B， YANG Y R， HOU L X. Fluidized bed emission mechanism and its application in fault diagnosis［J］. Chinese Journal of Chemical Engineering， 2001， 52（11）： 941-943.

［5］郑燕萍， 黄轶伦， 陈伯川， 等 . 利用压力脉动信号特征预测流化床结块故障［J］. 高校化学工程学报， 2000， 14（2）：128-133.

ZHENG Y P， HUANG Y L， CHEN B C， et al. The pressure pulsation signal characteristics are used to predict the agglomeration fault of the fluidized bed［J］. Journal of Chemical Engineering of Chinese Universities， 2000， 14（2）： 128-133.

［6］SONG J C， FAN L T， YUTANI N. Fault detection of the fludized bed distributor by pressure fluctuation signal［J］. Chemical Engineering Communications， 2007， 25： 105-116.

［7］ZHANG J C， CHEN X， HOU J X， et al. Early risk warning method for fluidized beds using generalized extremum distribution of pressure fluctuation［J］. Process Safety and Environmental Protection， 2021， 156： 29-42.

［8］韩美玉. 基于机器学习的高密度循环流化床流动特性预测［D］. 北京： 中国石油大学（北京）， 2023.

HAN M Y. Machine learning-based prediction of flow characteristics of high-density circulating fluidized beds［D］. Beijing： China University of Petroleum（Beijing）， 2023.

［9］CHEW J W， COCCO R A. Application of machine learning methods to understand and predict circulating fluidized bed riser flow characteristics［J］. Chemical Engineering Science. 2020， 217： 115503

［10］KIM M， JUNG S， KIM B， et al. Fault detection method via k-nearest neighbor normalization and weight local outlier factor for circulating fluidized bed boiler with multimode process［J］. Energies， 2022， 17（15）： en15176146

［11］张蕾. 基于声发射的流化床故障诊断方法研究［D］. 包头： 内蒙古科技大学， 2015.

ZHANG L. Research on fault diagnosis method of fluidized bed based on acoustic emission［D］. Baotou： Inner Mongolia University of Science and Technology， 2025.

［12］俞红萍， 罗保林 . 基于神经网络和宏观动力学模型的流化床反应器离线故障诊断［J］. 化工冶金， 1999， 20（3）：290-294.

YU H P， LUO B L. Offline fault diagnosis of fluidized bed reactor based on neural network and macroscopic dynamics mode［l J］. The Chinese Journal of Process Engineering， 1999， 20（3）： 290-294.

［13］林伟国， 张鹏， 陈磊， 等. 流化床反应器结块故障的声纹特征提取及监测技术［J］. 化工学报， 2012， 63（9）： 2851-2858.

LIN W G， ZHANG P， CHEN P， et al. Voiceprint feature extraction and monitoring technology for agglomeration faults in fluidized bed reactors［J］. Chinese Journal of Chemical Engineering， 2012， 63（9）： 2851-2858.

［14］杨新， 麻哲瑞， 申赫男， 等. 基于多尺度特征能量-核极限学习机的双循环流化床气流堵塞故障智能诊断［J］. 化工

学报， 2019， 70（7）： 2616-2625.

YANG X， MA R Z， SHEN H N， et al. Intelligent diagnosis of airflow blockage fault in dual-cycle fluidized bed based on multi-scale characteristic energy-nuclear extreme learning machine［J］. Chinese Journal of Chemical Engineering， 2019，70（7）： 2616-2625.

［15］郭中山， 袁世威， 孙国刚， 等 . 旋风分离器排料堵塞与料腿漏气的故障诊断［J］. 中国粉体技术， 2022， 28（4）：109-117.

GUO Z S， YUAN S W， SUN G G， et al. Fault diagnosis of cyclone separator discharge blockage and air leakage of material leg［J］. China Powder Science and Technology， 2022， 28（4）： 109-117.

［16］BREIMAN. Random forests［J］. Machine Learning， 2001， 45（1）： 5-32.

［17］WALTER J， WILLIAMS J， Li Y. Comparative study of distance functions for nearest neighbors［C］//Proceedings of the Advanced tTechniques in Computing Sciences and Software Engineering. Springer， 2010： 79-84.

［18］ROSA T， PRIMARTHA R， WIJAYA A. Comparison of distance measurement methods on k-nearest neighbor algorithm for classification［J］. Advances in Intelligent Systems Research， 2019， 172： 258-261.

［19］张清华， 肖嘉瑜， 艾志华， 等. 自适应半径选择的近邻邻域分类器［J］. 模式识别与人工智能， 2022， 35（11）： 989-998.

ZHANG Q H， XIAO J Y， AI Z H， et al. Nearest neighbor classifier for adaptive radius selection［J］. Pattern Recognition and Artificial Intelligence， 2022， 35（11）： 989-998.

［20］ZHAO G J， PAN X Q， YAN H， et al. Predicting engineering properties of controlled low-strength material made from waste soil using optimized SVR models［J］. Case Studies in Construction Materials. 2024， 20： e03325.