【作者】 朱红路；

【导师】 刘吉臻；

【作者基本信息】 华北电力大学（北京） ， 控制理论与控制工程， 2010， 博士

【摘要】 随着我国电站信息化进程的迅速发展,逐渐形成了以DCS(分散控制系统)、SIS(监控信息系统)、MIS(管理信息系统)为核心的过程控制、信息监控及管理系统。与此同时,许多先进控制、计算、诊断、优化功能在此平台上得以实施,极大促进了机组控制、运行、管理水平的提高。这些功能的实施依赖于各种现场测量信号。现场信号的可靠性、准确性、一致性将直接影响功能算法最终结果。因此对检测信号进行必要的预处理以减少各种不确定性的影响是十分必要的。另一方面,理论研究和工程实践都需要对过程被控对象更多的了解。因此,研究一种能够对热工过程信号、对象进行有效分析和处理的信号处理手段具有十分重要的意义。本文以自适应滤波作为理论依据,深入的研究了自适应滤波技术的相关理论及其应用,将其引入到热工过程信号的噪声消除,热工过程对象建模和热工过程对象的纯迟延估计,模型算法控制等应用中。论文的主要工作内容和研究成果包括：研究了自适应滤波理论,包括LMS算法和RLS算法的基本算法、结构和性质。针对热工过程对象特点,对算法进行部分改进以更好的适应于热工过程信号的分析和处理。针对工业过程检测和控制中存在的噪声扰动问题,应用ANC(Adaptive Noise Canceling,自适应噪声消除)技术对噪声进行抑制和消除。提出了基于自适应滤波技术的热工过程对象建模方法,建模结果为FIR(Finite Impulse Response,有限脉冲响应)模型。通过得到的FIR模型的抽头权值分布,提供了一种可视的直观的建模结果,仿真结果说明了这种建模方法具有较强的鲁棒性,并且算法简单,易于实现。研究了TDE(Time-Delay Estimation,时间延迟估计)的相关理论,验证了自适应延时估计对输入信号中存在的加性噪声有着良好的抑制能力。对热工过程对象中的纯迟延问题,提出了基于建模方法的热工过程对象纯迟延估计方法,结合热工对象特点对这种纯迟延估计方法进行了仿真验证。研究了模型算法控制理论,在仿真实验中发现控制器参数和预测模型之间的相互关系,并据此对模型算法控制进行改进,以获得更为精确的预测模型和较低的计算复杂度。将基于自适应滤波技术的自适应建模方法和模型算法控制相结合,提出了一种基于自适应滤波器的自适应模型算法控制,并通过仿真实验说明了这种方法的有效性。更多还原

【Abstract】 For the booming of information process in our country’s power unit, DCS (Distributed Control System), SIS (Supervisory Information System in plant level), MIS (Management Information System) have been widely used as the key part of the process control system, information monitoring system and management system. Advanced control algorithm, computing function, diagnostics and optimization have been applied in those platforms, which accelerate the control level, operation level and management level. The implementation of these algorithms needs a lot of signals from the field. So the reliability, the accuracy and the consistency of those signals plays a very important role. Therefore, it is necessary to do the data preprocessing job for those field signals. Otherwise, it is essential to have knowledge about the controlled object for theoretical study and engineering. So it is of great significance to learn a signal processing technology for the thermal process signal. In this paper, take the adaptive filter theory as a tool, and use it to eliminate the noise in thermal process signal, build the model for the unknown thermal object, estimate the delay of the thermal process object, and incorporate with the model algorithmic control.The main content and research result:Study the adaptive filtering theory which includes Least-mean-square algorithms and Recursive-least-square algorithms. According to the characteristics of the thermal process, part of the algorithm is improved for better using to the analysis of thermal processes and signal processing. For the existing noise disturbance problem in industrial process monitoring and control, implement ANC (Adaptive Noise Canceling) technology for the noise suppression and elimination to the industry process signal and control.A thermal process modeling method based on adaptive filtering technology is proposed, the modeling results is FIR (Finite Impulse Response) model. The tap weight distribution of the FIR model provides a visible modeling result. The simulation results show that this modeling method has strong robustness, and the algorithm is simple, easy to implement.Study the relevant theories of the TDE (Time-Delay Estimation), verify the adaptive time delay estimation has a very good performance for suppression the noise to the input. For the delay of the thermal process object, proposed a method of estimating delay for the process object, which is based on modeling method, and simulation result shows its perfect effect.Study the algorithm theory of the model algorithmic control, and find the relationship between the controller parameters and the prediction model during the simulation. The model algorithmic control method is improved to have a more accurate prediction models and low computational complexity. Combine the adaptive modeling method based on adaptive filtering and model algorithmic control, proposed an adaptive model control algorithm based on adaptive filter, and the simulation results show the effectiveness of this approach.更多还原