超临界火电机组多变量前馈预见预测控制

Multi-variable Feedforward Foreseeable Predictive Control for Supercritical Thermal Power Units

采用BP神经网络算法建立超临界火电机组协调控制系统的多输入多输出数学模型,并在此基础上利用前馈控制、预见控制及预测控制对机组进行优化控制,形成一个多变量前馈预见预测(FFP)控制系统。结果表明:神经网络前馈预见预测控制使机组更快、更准确地跟踪机组AGC曲线、滑压目标值和中间点温度设定值;前馈控制加剧了主汽压和中间点温度的波动,预见控制能使这2项指标趋于平稳;采用前馈预见预测控制算法来控制超临界火电机组,能够提高机组的AGC响应速度、经济性、稳定性和安全性。

A multi-input multi-output mathematical model was established using neural network algorithm for the coordinated control system of a supercritical thermal power unit,based on which,the feedforward,foreseeable and predictive method were used to optimize the control of the unit,thus to form a multivariable feedforward foreseeable predictive(FFP)control system.Results show that the neural network predictive control enables the unit to track the AGC curve,the sliding pressure target value and the midpoint temperature set value more quickly and accurately,whereas the feedforward control aggravates the fluctuation of the main steam pressure and the intermediate point temperature,while the foreseeable control has the effect of making these two indicators tend to be stable.Therefore,using the feedforward foreseeable predictive(FFP)algorithm to control the supercritical thermal power unit could improve the AGC response speed,the economy,the stability and the safety of the unit.