基于多模型动态矩阵预测的冷却剂平均温度控制  

Simulation research on coolant average temperature control based on multi-model dynamic matrix predictive control algorithm

作　　者：钱虹[1] 金蔚霄 

机构地区：[1]上海电力学院自动化工程学院,上海200090

出　　处：《热力发电》2015年第11期98-103,108,共7页Thermal Power Generation

基　　金：上海市重点实验室资助项目(04DZ05901)

摘　　要：由于核反应堆堆芯R棒棒位与冷却剂平均温度动态关系的非线特性,现有核电机组固定负荷控制系统的堆芯核功率难以快速跟踪机组负荷变化,对此,建立了表征R棒棒位与冷却剂平均温度动态特性关系的非参数模型组,基于动态矩阵预测控制算法设计了多模型控制器,并采用MATLAB/SIMULINK软件构建了冷却剂平均温度多模型动态矩阵预测控制系统,与核电站仿真模型连接构成仿真平台。利用该平台,在高功率运行状态下基于核电机组仿真模型数据,将多模型动态矩阵预测控制器与传统PI控制器进行了仿真对比。2 000s内冷却剂平均温度跟踪性能评价指标和仿真结果表明,在机组负荷升、降过程中多模型动态矩阵预测控制器作用下的冷却剂平均温度跟踪能力显著优于传统PI控制器。Due to nonlinear dynamic characteristics relationship between the R rod position and coolant average temperature in nuclear reaction core,the current fixed load control system of nuclear power plant has difficulties in making the nuclear power in reactor core trace the unit load variation rapidly.Against this problem,a nonparametric model set of dynamic characteristic relationships between the R rod position and coolant average temperature were established.On the basis of the dynamic matrix predictive control algorithm,the model controller was designed.Moreover,using the MATLAB/SIMULINK,the multi-model dynamic matrix predictive control system for coolant average temperature was built up and then connected to the simulation model for nuclear power station to form the simulation platform.In high power operation state,by using the simulation model data in nuclear power plants,the multi-model dynamic matrix predictive controller and the conventional PI controller were compared to verify the effectiveness and superiority of the designed multi-model dynamic matrix predictive controller.The evaluation indexes of tracing effect of coolant average temperature within 2 000 sand the simulation results show that,the tracing effect of this multi-model dynamic matrix predictive controller is dramatically better than that of the conventional PI controller during unit load ascension and descending process.

关 键 词：核电机组 核反应堆 堆芯 R棒棒位 冷却剂 平均温度 多模型 动态矩阵预测控制 

分 类 号：TM623[电气工程—电力系统及自动化] TP391.4[自动化与计算机技术—计算机应用技术]