一种模糊模型预测控制在锅炉汽轮机中的应用  

Application of Fuzzy Model Predictive Control in Boiler Turbine System

作　　者：祁星晨 卓旭升[1] QI Xing-chen;ZHUO Xu-sheng(School of Information and Electrical Engineering,Wuhan Institute of Technology,Wuhan 430205,China)

机构地区：[1]武汉工程大学电气信息学院,武汉430205

出　　处：《自动化与仪表》2021年第6期17-21,26,共6页Automation & Instrumentation

摘　　要：随着新能源发电占比不断上升,火力发电的灵活性对电网的安全稳定运行显得尤为重要,对此火电机组的控制性能提升是一个重要技术手段。相较于传统控制算法,采用模型预测控制技术的控制器不仅具有良好的性能,而且能够严格遵循非线性系统的物理约束条件,此外也能适应各种不同类型的模型。该文利用k-means++聚类和具有动量的随机梯度下降算法建立了T-S模糊模型,并给出了动量随机梯度下降算法和标准随机梯度下降算法的辨识T-S模糊模型后件参数的对比实验,最后通过数值优化方法得到模糊模型预测控制的最优解,在大范围负荷变化条件下,将其与传统PID控制作对比。仿真结果显示,动量随机梯度下降算法在后件参数辨识的对比实验中,震荡幅度小,下降速度更快,且精度更高;在模糊模型具有非线性对象信息足够多的基础上,模糊模型预测控制能够相较于传统PID控制灵活性更强,超调小,并且符合系统约束条件。As the proportion of new energy power generation continues to rise,the flexibility of thermal power generation is particularly important for the safe and stable operation of the grid.The improvement of the control performance of thermal power units is an important technical method.Compared with traditional control algorithms,the controller using model predictive control technology not only has good performance,but also can strictly follow the physical constraints of the nonlinear system,and can also adapt to various types of models.The T-S fuzzy model is established using k-means++clustering and momentum stochastic gradient descent algorithm,and the comparison experiment of the momentum stochastic gradient descent algorithm and the standard stochastic gradient descent algorithm for identifying the parameters of the consequent part of the T-S fuzzy model is given,and finally through the numerical optimization method obtains the optimal solution of the quadratic programming problem of the fuzzy model predictive control,and also compares it with the traditional PID control under the conditions of a wide range of load changes.The simulation results show that,the momentum stochastic gradient descent algorithm in the comparison experiment of the consequent part identification of the T-S fuzzy model,under the oscillation amplitude,the descending speed is faster and the accuracy is higher.On the basis that the fuzzy model has enough non-linear object information,the fuzzy model predictive control can be more flexible than the traditional PID control,with less overshoot,and easily meet the system constraints.

关 键 词：模型预测控制 模糊模型 锅炉汽轮机系统 辨识算法 k-means++聚类算法 随机梯度下降 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]