摘要
发电设备状态预测作为运行维修工作站（O&M）的重要组成部分，接受来自各种监测与诊断系统或现场运行人员的状态报告，预测设备未来的发展趋势，并根据预测结果评价设备状态、合理安排维修时间与维修项目，这对于发电设备的运行与维修决策具有重大的意义。近年来，基于神经网络预测在许多领域中的成功应用，为发电设备的状态预测技术的发展开拓了新的途径。
本论文在分析神经网络算法的基础上，进一步阐述了基于神经网络的状态预测模型、设备发展规律及状态预测相关技术，结合发电设备运行特征提出一种组合式预测模型，并开发了一套的发电设备状态预测软件。
给水泵是电站单元机组的重要设备之一，它的失效将带来巨大的经济损失。开展给水泵状态预测的研究，提高给水泵运行的安全性、可靠性及状态维修，都具有重要的现实意义。因此，本论文将基于神经网络状态预测技术的理论成果应用于主给水泵及液力耦合器状态预测，为其它发电设备开展状态预测提供了一种的思路与方法。
As the important integrant of the Operating and Maintenance Workstation, the condition forecasting of power generating equipment accepts condition report coming from various monitoring and diagnosing system or personnel on the site, forecasts the developing trend of equipment, evaluates the condition of equipment and arranges the schedule and items of maintenance rationally according to the results of forecasting, which has great signification to maintenance decision-making of power generating equipment. In recent yeas, the successful using of neural network based forecasting in many fields has exploited new way for the development of condition forecasting technology of power generating equipment.
This paper analyzed the basic problems of neural network based condition forecasting of power generating equipment, the developing rule of equipment and concrete realization method of condition forecasting on the basis of analyzing the neural network algorithms, put forward combination forecasting model according to operating characteristic of power generating equipment and developed a set of condition forecasting software of power generating equipment.
The feed water pump of boiler is one of important equipment of boiler unit. The invalidation of it will bring large economic loss. It has important real signification to study the condition forecasting of feed water pump, advance the safety and reliability of operation of feed water pump and to develop condition maintenance to feed water pump. Therefore, this paper applied the study fruit of neural network based condition forecasting technology to the condition forecasting of master feed water pump and hydraulic coupling. This provides new idea or method for condition forecasting of other power generating equipment.
引文
[1]Michael Callahan. Hidden Features of CMMS, PLANT ENGEERING, 1997 MARCH
[2]Larry Stoddard. Make CMMS work for you, POWER, 1996 SEP/OCT 140(8)
[3]EPRI Reports.CMMS Selection at Wisconsion P.S.: Computerized Maintenance Management System,TR-108938,Oct 1997
[4] EPRI Reports. Operations and Maintenance Workstation:Interface Tool to Plant Operations and Maintenance Data.TR-109729,V1,Mar 1998
[4]EPRI Reports. Operations and Maintenance Workstation:Interface Tool to Plant Operations and Maintenance Data.TR-109729,V2,Mar 1998
[5]EPRI Reports. CMMS Selection at Cinergy:(Computerized Maintenance Management System).TR-109529,Dec 1997
[6]黄雅罗等.发电设备状态检修.北京：中国电力出版社，2000
[7]王东生等.混沌、分形及其应用.合肥：中国科技大学出版社，1995
[8]G.Sugiharaetal. Nonlinear forecasting as a way of distinguishing chaos from measurement error in time series. Nature, 1990,344(19):734-741
[9]F.Takens. Detecting strange attractors in turbulence. Lecture Notes in Mathematics, 1981:366-381
[10]C.Cempeletal. A new method of symptom based processing for machinery condition recognition and forecasting. Mechanical Systems and Signal Processing,1995,9(2):129-137
[11]钟掘等.关于机械系统非线性故障诊断若干思考.第五届全国机械设备故障诊断学术会议，1996
[12]姜建东.机械测试信号的分形评估[博士学位论文].西安：西安交通大学，1997
[13]张曾科.模糊数学在自动化技术中的应用.北京：清华大学出版社，1997
[14]吴余华.基于HOS的机械设备故障诊断及剩余寿命评定问题的研究[博士学位论文].杭州：浙江大学，1995
[15]曹朗等.用确定性方法预测工业总产值.预测，1995，（2）：52－56
[16]T.Hill.人工神经网络预测和决策模型.预测，1995，（4）：68－72
[17]李勇等.汽轮发电机组故障预测方法.汽轮机技术，1996，38（1）：17－20
[18]王宏伟等.旋转机械运行状态的神经网络预测研究.第五届全国机械设备故障诊断学术会议，1996
[19]陈大禧等.神经网络预测性能优化.第五届全国机械设备故障诊断学术会议，1996
[20]蔡正国等.神经网络用于非线性时间序列预报.第四届全国机械设备故障诊断
学术会议，1994
[21]徐金梧等.运用神经网络方法预测故障的趋势.第四届全国机械设备故障诊断学术会议，1994
[22]李书明.制造设备智能诊断与状态预测技术的研究[博士学位论文].天津：天津大学，2002
[23]陈耀武.旋转机械状态预测监测系统与故障诊断、状态预测方法研究[博士学位论文].杭州：浙江大学，1998
[24]R.Lamedicaetal. A neural network based technique for short-term forecasting of anomalous load periods. IEEE Trans. On Power System,1996,11(4):1749-1756
[25]张国华.基于神经网络的短期负荷预测[硕士学位论文].北京：华北电力大学，2000.12
[26]王东生等.混沌、分形及其应用.合肥：中国科技大学出版社，1995
[27]伊恩.斯图尔特.混沌之数学.上海：上海远东出版社，1995
[28] G.Cybenkoetal. Approximation by superposition of sigmoidal
function.Mathematics of Control Signals and Sys.1989,(2):303-314
[29]雷鸣等.非线性序列建模与预测神经网络法.华中理工大学学报，1993，21（1）：47－52
[30]A.S.Dehdashtietal. Forecasting of hourly load by pattern recognition-a determination approach. IEEE Trans. On Power Apparatus and Sys.,1982,101(9):3290-3294
[31]P.Tenti. Forecasting foreign exchange rates using recurrent neural networks. Applied Artificial Intelligence.1996,(10):567-581
[32]R.Katayama. Fusion technology of fuzzy and chaos theory and its application. Advances in Fuzzy System, Vol.2, 1995:241-270
[33]陈晓龙.火电站锅炉给水泵系统的振动监测及其预知维修系统的研究[博士学位论文].北京：清华大学，1991
[34]蔡镜等.“玻璃企业市场预测与分折”预测支持系统研究与设计. 山东轻工业学院学报，1998，12（2），64－74
[35]刘志俭等.MATLAB外部程序接口(6.x).北京：科学出版社，2002
[36]杜江等.面向对象预测支持系统动态结构研究.数学杂志，1998，18（增刊），117－120