基于概率模型的热力机组性能诊断与仿真  

Probabilistic Model Based Performance Diagnosis of Thermal Power Plants and Its Simulation

作　　者：李利平[1] 张春发[2] 牛玉广[1] 马进[1] 

机构地区：[1]华北电力大学自动化学院,保定071003 [2]华北电力大学能源动力学院,保定071003

出　　处：《动力工程》2007年第4期564-568,共5页Power Engineering

基　　金：国家自然科学基金(50576022)

摘　　要：从复杂系统的不确定性本质出发,利用热力机组的大量实际运行数据,建立系统的概率模型用于系统在线性能诊断和运行决策。以能耗率的准确计算为基础,以已知或不可控的边界条件为证据,通过概率推理得到其它状态参数取值的置信区间并结合工程实践的经验进行决策调整而达到优化运行的目的。概率模型有效扩展了传统的基于热力学模型诊断方法,特别适应于现场环境下采用常规仪表测量参数的在线检测与辅助决策。通过系统全仿真动态模型验证了概率模型的有效性。Sensing the inherent uncertainty of complex systems, a probability graph＇ s model of the system, which makes use of abundant actual operational data, is constructed, to be used for online performance diagnosis and operational decisions. With exact calculation of energy consumption rate as a basis, and with known or uncontrollable boundary conditions as evidence, the objective of optimal operation can be attained, by ascertaining the probabilistic reasoning based confidence interval for determining the remaining state parameters, and then by referring to workable practical experiences, operational decisions can herewith be timely adjusted. Probability models are effective expansions of, thermodynamic model based traditional ways of diagnosis and are especially suitable for online diagnosis, and helpful for decision making according to parameter readings of conventional instruments. Simulation results with dynamic models of complete systems have validated the effectiveness of the model.

关 键 词：自动控制技术 复杂系统 概率模型 性能诊断 优化决策 仿真模型 

分 类 号：TK39[动力工程及工程热物理—热能工程]