【作者】 李云伍；

【导师】 乌建中；

【作者基本信息】 同济大学 ， 机械电子工程， 2006， 博士

【摘要】 小型燃气发动机热电联产将能源就地转换，有效地提供电能和热能，总能源利用率达85％以上，将会是21世纪能源技术的主流之一。在我国能源日益紧张的情况下，研究燃气发动机热电联产机组及其控制系统对于提高能源利用率具有重要意义。 燃气发动机热电联产机组控制系统包括了气体流动、燃烧与膨胀，往复式机械运动，电磁转换等众多复杂的瞬变动态过程，具有时变、非线性和多不确定性干扰等特性。在分析机组运行特性的基础上，运用非线性控制理论和智能控制理论，详细研究了机组供电质量相关的频率控制、励磁控制、励磁与频率调节相互影响的机理及其解耦控制以及机组余热回收及其控制等问题，并通过试验和仿真对提出的控制策略和方法进行了验证。 1、建立燃气发动机热电联产试验机组，研究了机组结构、燃气发动机改装技术以及控制系统的硬件组成。根据燃气发动机和同步发电机的运行特性建立了机组的机理数学模型，然后利用神经网络对有限试验结果进行拟合得到机组输入输出参数间的关系曲线，用拟合关系曲线对机理数学模型参数进行修正。由此得到了符合机组实际运行特性的机组数学模型。 2、研究直接以频率值为反馈，负荷电流为前馈，采取前馈与反馈协调控制策略，通过步进电机调节节气门开度进行频率控制的方法。前馈补偿采用模糊神经网络算法(FNN)，以负载电流i及电流变化量ic作为输入，以步进电机转动的角度θ为输出，以反馈控制器的输出误差平方和最小为控制训练目标。反馈控制采用PID神经网络算法(PID-NN)，设计了PID-NN各层神经元个数、连接方式、连接权值，同时研究了以系统频率给定值r和机组实际频率输出值y为输入，输出误差平方和为最小的学习算法。FNN前馈快速反应机组负荷的变化并做出相应的调节，PID-NN通过自学习，自适应前馈对系统的调节作用并对系统误差进一步调节，减少系统偏差和增加系统的稳定性和鲁棒性。试验结果表明FNN前馈补偿与PID-NN反馈协调进行频率控制其效果明显优于PID和模糊控制，能适应机组非线性特性和参数的变化，减少负荷突变时的频率稳定时间和稳态误差。更多还原

【Abstract】 Gas engine cogeneration unit transforms and provides electric and heat energy on the spot. Its total efficiency of energy utilization is reach to 85%. On the situation of energy short, it is important to research gas engine cogeneration unit and its control system.The control system of gas engine cogeneration unit is a complex one with time-variable and nonlinear properties, in which contains a lots of transient and dynamic processes such as gas flow, combustion, reciprocating mechanical movement and electromagnetism transform. Based on analyzing the problems of the exiting control method and the running characters of the generattion unit, by combing nonlinear control and intelligent control theory, this dissertation aims at research and investigation thoroughly on power frequency control strategy, exciting control strategy, the interrelationship of power frequency and exciting control and its decoupling control strategy, the structure and control method of waste hot utilization system of the unit. Further more, these strategies and methods are tested through establishing experiment unit or simulation model.1. A model is studied for complicated multi-parameters engine generating unit. Firstly, the mechanism model is deduced base on the operating characteristics of natural gas engine and synchronous generator. Then, according to the limited test results, the relationship curves of input and output parameters are gained by artificial neural network calculating. Last, the parameters of mechanism model are adjusted according the relationship curves.2. The control method of power frequency is studied, of which the measured value of power frequency is as feedback, the current of load is as feedforward and throttle angle is adjusted by step motor. PID controller and fuzzy controller are designed and tested through experimental unit. The results indicate that PID controller and fuzzy controller can’t get appropriate control precision for power frequency control. Then frequency control strategy combining feedforward and更多还原