抽水蓄能机组的集成故障诊断非线性预测控制研究

项目名称： 抽水蓄能机组的集成故障诊断非线性预测控制研究

项目编号： No.51479076

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 水利工程

项目作者： 李超顺

作者单位： 华中科技大学

项目金额： 84万元

中文摘要： 抽水蓄能机组水泵水轮机全特性呈现强非线性，存在S区和驼峰区等不稳定区域，且机组启停频繁、工况变化剧烈，导致机组同时面临控制品质恶化和控制系统故障的风险。为提高机组控制品质并保障控制系统可靠性，以预测控制和多模型故障检测理论为基础，以复杂系统在线建模为突破点，提出具有故障自诊断和容错控制特点的抽水蓄能机组集成故障诊断非线性预测控制方法。首先针对复杂系统预测模型精度和求解效率的均衡问题，推导基于超平面原型聚类和递推卡尔曼滤波的在线模糊辨识方法，建立抽水蓄能机组调速系统模糊预测模型；研究基于模糊模型的抽水蓄能机组非线性预测控制方法，通过在线离线优化计算和快速矩阵推求解决控制实时性问题；以此为基础，建立基于多模型与扩展卡尔曼滤波的故障诊断模型并重构控制轨迹，阐明非线性预测控制模型对故障诊断信息的响应关系。研究成果将完善非线性预测控制理论，并为提高我国抽水蓄能机组的运行水平提供重要的理论依。

中文关键词： 抽水蓄能机组；水泵水轮机；非线性预测控制；容错控制；集成故障诊断

英文摘要： Characteristic curves of pump-turbine contain S zone and hump zone, and show strong nonlinearity. In operation, Pumped storage unit (PSU) need to frequently start and stop with working condition varying violently. These factors increase the risk of deterioration of control qualtiy and fault of the control system. In order to improve the control quality and ensure the safety of PSU, we propose a novel noniear predictive control (NPC) for PSU, integrating fault tolerant control and fault diagnosis, on the basis of model predictve control, fuzzy system identification and mutiply model based fault detection. Firstly ,in order to improve the accuracy and efficiency of modeling of NPC, on-line fuzzy model identification method based on hyper-plane clustering and recursive kalman filter will be studied, and the fuzzy model of control target of PSU will be built. Secondly, NPC method based on precise fuzzy modeling of control target of PSU will be deduced and the stability will be proved theoritically. The realtime performance of the NPC alorithm will be improved by the way of optimally seperating on-line and off-line caculation and implementing rapid matrix compation. Finally, in the frame of model predictive control, a fault detection and diagnosis model based on multiply model theory and state observer technique will be proposed to detect fault and contribute to the rebuilding of control trajectory. The work of this part will try to illuminate the response relationship of predictive control and fault diagnosis information. Research results of this project will imporve the NPC theory and provide theoritical basis and technical support for operation of PSU.

英文关键词： pumped storage unit;pump-turbine;nonliear predictive control;fault tolerant control;integrated fault diagnosis