【作者】 寇攀高；

【导师】 周建中；

【作者基本信息】 华中科技大学 ， 水利水电工程， 2012， 博士

【摘要】 精确的水轮发电机组数学模型不仅是水力发电系统仿真的基础,而且对于互联电力系统的计算分析、规划运行和控制保护设计具有重要意义。随着水电能源的大力开发,我国已建及在建的水电站单机容量大、引水管道长、水流惯性巨大、引水系统布置复杂,同时承担电力系统调频、调峰和事故备用等任务,对电力系统的安全稳定运行及模拟分析提出了严峻挑战,然而目前国内电力系统计算分析所用的水轮机调速器数学模型往往采用简化的模型,与实际投运的水轮机调节系统模型结构存在较大差异。因此,有必要研究水轮发电机组的先进辨识方法及控制策略,建立精确的水轮发电机组数学模型,提高水电机组仿真精度及机组的控制性能,确保水电机组及电力系统的安全高效运行。基于辨识的控制系统建模是获取研究对象数学模型的一种重要手段,传统的水力发电系统辨识研究中,辨识模型多采用线性模型,辨识方法多基于线性系统辨识理论,缺乏对水力发电系统非线性因素及参数时变的考虑,难以满足电力系统的动态稳定分析、中长期电压稳定分析、自动电压控制及系统频率稳定分析的需求。为此,在深入研究水轮发电机及其调速系统数学模型的基础上,凝练出水轮发电机组辨识及控制所面临的科学问题,结合多新息理论、先进智能优化方法、动态系统设计方法及变结构控制策略,对水电机组的参数辨识方法、先进控制策略进行了系统深入的研究,提出了水轮发电机组辨识-控制一体化体系。论文的主要工作及创新性成果如下：(1)针对水轮发电机组频率调节及控制的需求,研究并建立了水轮机调节系统频率调节、开度调节及功率调节数学模型,对其中的非线性环节进行了重点解析,建立了基于Simulink的水力发电系统多工况仿真平台,为系统辨识及控制策略的研究打下了基础。(2)以建立精确的水轮发电机及其调速系统线性模型为研究目标,引入多新息辨识理论,在传递函数模型的基础上推导出同步发电机任意工况下一次性抛载试验及水轮机调节系统的线性回归模型,并通过实例验证了多新息辨识理论在水电机组控制系统参数辨识的可行性。(3)以水轮机调节系统非线性模型参数辨识为目标,在引入量子粒子群优化算法的理论框架下,将水轮机调节系统参数辨识问题转化为单目标函数有约束优化问题；针对水轮机调节系统对象参数时变、非最小相位、复杂非线性的特点,提出了一种基于混沌变异的QPSO (Quantum Particle Swarm Optimization)算法,并通过不同工况仿真试验验证了IQPSO (Improved Quantum Particle Swarm Optimization)在水轮机调节系统参数辨识中的有效性。(4)针对水轮发电机非线性状态空间模型的参数辨识问题,在粒子群优化算法的基础上引入量子计算,提出了粒子群-量子操作(Particle Swarm Optimization with Quantum Operation, PSO-QO)优化算法,并将该方法应用到同步发电机非线性模型参数辨识中,通过实例验证了该方法在同步发电机参数的离线辨识中的有效性；并在粒子群-量子操作算法中引入敏感粒子,实现了同步发电机参数的在线跟踪与辨识。(5)在总结以往参数辨识研究思路的基础上,尝试从控制论的角度认识参数辨识问题,提出了基于动态系统设计的参数辨识方法及动态系统设计原则,并以Hopfield神经网络为例,研究了基于Hopfield (?)神经网络动态系统的同步发电机非线性状态空间模型参数辨识,并通过实例验证了基于动态系统设计的参数辨识研究思路的可行性及有效性。(6)针对传统PID参数难以满足水轮机调节系统参数时变、多工况运行的特点,将菌群优化(Bacterial Foraging Optimization, BFO)算法引入到水轮发电机组调速器PID参数优化整定中,综合BFO算法良好的全局搜索能力及粒子群优化(Paticle Swarm Optimization, PSO)算法收敛速度快的特点,提出了菌群-粒子群优化算法(BFO-PSO);在传统ITAE(?)旨标的基础上,考虑菌体间相互吸引、相互排斥、相互学习的行为特点,提出了一种新型适应度函数,构造出基于智能优化算法的水轮机调速器PID参数整定框架,实现了水轮机调节系统任意工况下参数整定。(7)针对传统PID控制策略随水轮机工况切换、参数时变适应性差的问题,将滑模变结构控制引入到水轮机调节系统中；考虑到传统水轮机调节系统状态空间模型忽略了机组转速给定项无法仿真机组空载扰动且线性最优控制下机组转速存在稳态误差的问题,推导出一种三控制输入的水轮机调节系统状态方程,揭示了状态方程稳态误差产生机理,提出了消除状态方程稳态误差的方法,在此基础上设计了水轮机调节系统的滑模变结构控制器。更多还原

【Abstract】 Accurate mathematic model of hydro turbine unit not only provides basis for hydro turbine system simulation, but also plays an important role in interconnected power system calculation and analysis, operation and plan, protection device and controller design. With great developing of the hydro-electric power, the built and building hydro power stations are with large capacity generators, long penstock, big water inertia, and complicated water diversion system, meanwhile, they also undertake the task of primary frequency modulation, peak load regulation and emergency reserve for accident. This causes great challenge for power system modeling and control. However, models of the hydraulic turbine governing system used in power system stability calculation nowadays in China usually cited simplified models, which is different from the field hydraulic turbine governing system (HTGS) in operation. Hence, it is very necessary to study of the identification method and control law for accurate modeling of hydro turbine unit not only for improving the simulation accuracy and the controller performance, but also for the hydro turbine operation with high efficient and interconnected power system operation safely.System modeling based on identification is an important way to provide accurate models of control systems. Conventionally, linear system model and linear identification theory was studied for parameters identification of hydro power unit. While, they neglected the nonlinear factor such as saturation, amplitude limitation in servomotor system and time-varying factor in turbine, which are hard to satisfied the need of power system dynamic analysis, mid-term and long-term voltage stability calculation, auto generation control and system frequency analysis. Focus on these problems and research on models of synchronous generator (SG) and its speed mover governor system (PMGS) of hydropower unit, the paper proposed the scientific problems of parameters identification and control law design of hydropower unit. Combing the multi-innovation identification theory, intelligent optimization method, dynamic system design method and sliding mode control law, the paper studied detailed parameters identification technology and control laws of hydropower unit, and built the frame of parameters identification and system control for hydropower unit. Main contribution and innovations of the paper is organized as follows:(1) Focus on the need of frequency regulation and control of power system, the paper studied and built models of HTGS under frequency regulations, opening regulations and power regulation. Meanwhile, nonlinear sets of the system are detailed described and a simulation platform for hydropower system is built in Simulink, which provide a basis for parameters estimation and control law design of hydropower system.(2) To get accurate models of SG and its PMGS, the paper introduced the multi-innovation least square (MILS) identification method. A linear regressive model of SG and its PMGS is derived from their transfer function model. Case study showed the effect of the multi-innovation least square method in parameters identification of hydropower unit.(3) To estimate parameters of HTGS, the paper introduced the Quantum Particle Swarm Optimization (QPSO) algorithm and converted the problem of parameters identification of HTGS into a constrained optimization problem with single object function. Focus on the characteristic of time-varing, non-minimum phase and complex nonlinearity of HTGS, an improved QPSO (IQPSO) with chaos mutation operator is proposed and is used to solve the problem of parameters estimate of nonlinear HTGS. Simulations under different operations show the effectiveness of IQPSO.(4) To identification parameters of nonlinear state space of SG, this paper presents an improved algorithm named Particle Swarm Optimization with Quantum Operation (PSO-QO) to solve both offline and online parameters estimation problem for SG. An illustrative example for parameters identification of SG is provided to confirm the validity. Meanwhile, PSO-QO is also improved with a sentry particle introduced into the swarm to detect and determine parameters variation. Simulation results confirm that the proposed algorithm is a viable alternative for online parameters detection of SG.(5) With reviewing the identification methods in literatures, this paper tried to provide a novel idea to deal with parameters estimation as a problem of dynamic system design. Principles for dynamic system design are proposed in this paper, and a specific dynamic system based on Hopfield Neural Networks (HNN) is listed in the paper. With the dynamic system based on HNN, the paper estimated parameters of the nonlinear SG model formulated by third order differential equations and feasibility of the method is verified by simulation test.(6) Focus on conventional PID controller can not satisfied the need of HTGS with multi-condition and time-varying parameters, this paper introduced Bacterial Foraging Optimization (BFO) into PID parameter optimization tuning. Considering the slow convergence of BFO algorithm and the good convergence of particle swarm optimization (PSO) algorithm, a novel method named BFO-PSO algorithm was proposed. A new performance index integrated with ITAE and the Jcc index which can reflect the effect of bacterial swarm’s mutual attraction, mutual repellence and mutual learning was proposed. Thus, the paper constructed the frame of PID parameter optimization tuning based on intelligent optimization method and provides an alternative for PID parameter optimization tuning of HTGS in any conditions.(7) Considering that the PID control law is not suitable for HTGS with multiple conditions and parameter time-varying, the sliding mode control (SMC) law was illustrated in this paper. With control input of unit reference speed neglected, the no-load disturbance can not be simulated in conventional state space model of HTGS and steady speed error exists in this state space model under linear optimal control theory. To overcome these problems, the paper proposed an improved state space model of HTGS with three control inputs. With the steady error of the improved state space model analyzed, the paper display reason of the steady error and supplied a technique to eliminate the steady error of the improved model of HTGS. Meanwhile, a sliding mode controller for HTGS is designed based on the improved state space model of HTGS with three control inputs.更多还原