数据驱动的溶解氧浓度在线自组织控制方法  

Data-driven Online Self-organizing Control for Dissolved Oxygen Concentration

作　　者：权利敏 杨翠丽 乔俊飞 QUAN Li-Min;YANG Cui-Li;QIAO Jun-Fei(Faculty of Information Technology,Beijing University of Technology,Beijing Laboratory of Smart Environmental Protection,Beijing Key Laboratory of Computational Intelligence and Intelligent System,Beijing 100124;School of Information and Control Engineering,Qingdao University of Technology,Qingdao 266520)

机构地区：[1]北京工业大学信息学部,智慧环保北京实验室,计算智能与智能系统北京市重点实验室,北京100124 [2]青岛理工大学信息与控制工程学院,青岛266520

出　　处：《自动化学报》2023年第12期2582-2593,共12页Acta Automatica Sinica

基　　金：国家自然科学基金(62021003,61890930-5,61973010);科技创新2030——“新一代人工智能”重大项目(2021ZD0112302);北京市自然科学基金(4202006)资助。

摘　　要：针对城市污水处理过程的非线性、不确定性以及非高斯等特点,提出一种数据驱动的溶解氧(Dissolved oxygen,DO)浓度在线自组织控制方法.首先,设计一种基于相关熵的自组织模糊神经网络控制器(Correntropy-based self-organizing fuzzy neural network,CSOFNN),采用相关熵与规则贡献度指标实现控制器结构与参数的自动构建或修剪.其次,设计基于相关熵诱导准则的补偿控制器及参数自适应律,充分利用相关熵抑制非高斯噪声的能力,能够有效地降低系统中的不确定性.然后,分析所提出的控制方法的稳定性,从而保证其在实际应用中的可靠性.最后,基于基准仿真1号模型(Benchmark simulation model No.1,BSM1)的实验验证了所提方法的有效性.To deal with the nonlinearity,uncertainty and non-Gaussianity of urban wastewater treatment processes,this paper proposes a data-driven online self-organizing control method for dissolved oxygen(DO).First,a correntropy-based self-organizing fuzzy neural network(CSOFNN)controller is designed.For CSOFNN,its structure and parameters can be automatically generated or pruned based on the correntropy and rules-contribution indexes.Second,the compensation controller and parameter adaptive laws are developed using the correntropy-induced criterion,thus can tackle non-Gaussian noise and reduce the system uncertainty.Third,the stability of the proposed control method is analyzed theoretically,thus ensuring its feasibility in practice.Finally,the proposed control method is tested in the benchmark simulation model No.1(BSM1).The experimental results show its effectiveness.

关 键 词：数据驱动控制 相关熵 在线自组织 模糊神经网络 溶解氧浓度 稳定性分析 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] X703[自动化与计算机技术—控制科学与工程]