基于矩阵分解的管道流动系统递推滤波

吉林大学学报(信息科学版) ›› 2026, Vol. 44 ›› Issue (2): 284-290.

基于矩阵分解的管道流动系统递推滤波

高宏宇^a,b , 胡银鸽^a,b , 于皓然^a,b , 蔡金瑞^a,b

东北石油大学 a. 电气信息工程学院; b. 人工智能能源研究院, 黑龙江大庆 163318

收稿日期:2025-03-16 出版日期:2026-04-14 发布日期:2026-04-14
作者简介:高宏宇(1979— ),女，黑龙江大庆人，东北石油大学副教授，硕士生导师，博士，主要从事复杂系统滤波、先进控制理论和智能控制研究，（Tel）86-15090529379（E-mail）gaohongyunepu@126.com。
基金资助:
国家自然科学基金资助项目(62103096); 黑龙江省自然科学基金资助项目(LH2023F006)

Matrix Factorization Based Recursive Filtering for Pipeline Flow Systems

GAO Hongyu ^a,b , HU Yinge ^a,b , YU Haoran ^a,b, CAI Jinrui^a,b

a. School of Electrical Information and Engineering; b. Artificial Intelligence Energy Institute, Northeast Petroleum University, Daqing 163318, China

Received:2025-03-16 Online:2026-04-14 Published:2026-04-14

摘要/Abstract

摘要：

针对智慧管道监测的能效优化与安全防护双重需求, 构建了融合占空比调度 ( DCS: Duty Cycle Scheduling)与拒绝服务(DoS: Denial of Servic) 攻击特征的协同分析模型, 创新性整合矩阵分解(MF: Matrix Factorization)技术与新型递推滤波算法。通过建立含多重噪声及随机非线性的离散化管道系统模型和滤波器模型, 提出了根据黎卡提差分方程求解的递推滤波算法, 并对滤波误差协方差的有界性进行了严格的分析, 求出了系统最优的滤波器增益。仿真结果表明, 所设计的方法在测量稀疏下可以实现管道系统传感器网络能耗降低的同时, 保持输出数据的完整性, 且能有效补偿噪声和随机非线性因素造成的状态滤波偏差, 能实现对管道系统流量与压力的精确滤波。

关键词: 天然气管道系统, 占空比调度, 拒绝服务攻击, 矩阵分解, 递推滤波

Abstract:

Long-distance pipelines spanning, complex geographical environments and diverse climate zones require monitoring systems to address multiple technical difficulties, while existing theories predominantly rely on mathematical models under idealized conditions. To address the requirements of energy efficiency optimization and security protection for intelligent pipeline monitoring, a collaborative analysis model integrating DCS(Duty

Cycle Scheduling) and DoS(Denial of Service) attack characteristics is constructed. This framework innovatively combines MF ( Matrix Factorization) technology with a novel recursive filtering algorithm. By establishing a discretized pipeline system model and a filter model incorporating multi-source noise and stochastic nonlinearities, a recursive filtering algorithm derived from solving the Riccati difference equation is proposed. A rigorous analysis of the boundedness of the filtering error covariance is conducted, and the optimal filter gain for the system is derived. Simulation results demonstrate that the proposed method achieves reduced energy consumption in pipeline sensor networks under sparse measurements while maintaining the integrity of output data. It effectively compensates for state estimation deviations caused by noise and stochastic nonlinear factors, enabling precise filtering of pipeline system flow rates and pressures.

Key words: natural gas pipeline systems, duty cycle scheduling, denial-of-service attacks, matrix factorization, recursive filtering

中图分类号:

TP14

高宏宇, 胡银鸽, 于皓然, 蔡金瑞.

基于矩阵分解的管道流动系统递推滤波

[J]. 吉林大学学报(信息科学版), 2026, 44(2): 284-290.

GAO Hongyu , HU Yinge , YU Haoran , CAI Jinrui .

Matrix Factorization Based Recursive Filtering for Pipeline Flow Systems

[J]. Journal of Jilin University (Information Science Edition), 2026, 44(2): 284-290.

[1]	张雨烟, 陈万忠, 张涛, 李明阳. 基于非负矩阵分解的癫痫脑电自动检测[J]. 吉林大学学报(信息科学版), 2017, 35(5): 551-559.
[2]	周鹏宇, 杨欣, 周大可, 刘加. 基于非负矩阵分解的阴影检测方法[J]. 吉林大学学报(信息科学版), 2013, 31(6): 575-581.