【作者】 李建兰；

【导师】 黄树红；

【作者基本信息】 华中科技大学 ， 热能工程， 2008， 博士

【摘要】 科学的维修决策是发电设备安全、可靠、经济运行的重要保证。针对目前优化维修正在我国电力行业中推广应用的现状及其在应用过程中存在的问题,本文对可靠性点检、故障诊断、状态预测、状态评价、风险评估、维修决策等优化维修过程中的关键技术进行了系统研究,并研究开发了发电设备优化维修决策支持系统。针对目前点检制在中国发电厂实施过程中存在的盲目性点检等问题,本文提出了发电设备可靠性点检(RSI——Reliability Spot Inspection),并建立了可靠性点检的模型。RSI将点检过程分为静态规划和动态调整两个阶段。静态规划从系统固有功能特性的角度出发,对设备进行加权失效模式分析、后果和致命度分析以及性能分析后,根据设备在系统中固有的风险度来规划点检,既保证对重要设备的充分监测,又减少不必要的点检。动态调整在设备运行期间定期对设备进行状态评估,根据设备当前的运行风险对点检规划进行调整,避免了点检过剩和点检不足。与传统的点检制相比,RSI保证了设备必要的可靠性,同时节约了电厂的各项资源成本,提高了点检的效率和经济性。重要度分析是可靠性点检和优化维修决策的基础,本文提出了一个新的设备重要度指标——综合权重。综合权重综合考虑了发电设备在系统中的固有结构和设备的基本功能,以及设备的实际运行状态,全面体现了设备在系统中的实际重要度。为了解决特征信息的不一致性和冗余问题,在分析讨论旋转机械振动特点的基础上,本文采用信息融合和粗糙集方法提出了一个振动故障诊断模型,该模型利用振动信号信息熵构造了一个Rough决策系统,约简后推导出振动故障的决策规则,不仅可以实现对完全一致振动故障信息的诊断,同时,利用Rough因子还可以实现对完全不一致振动故障信息的诊断,有效提高了对振动故障模式的识别能力。同时,针对复杂故障模式的发动设备,提出了基于模糊数学的故障诊断模型,通过计算改进贴近度对故障模式进行模糊识别,从而实现对该类设备的故障诊断。对于发电设备运行参数受外界因素干扰而出现波动的问题,本文提出了改进的GM(1,1)模型,该模型通过引入一个m点均值算子解决了原始序列的波动问题,并通过建立等维新息模型,缩小灰平面,实现对具有波动性质状态序列的有效预测。此外,本文提出并定义了一个灰色位关联度以反映比较序列之间在空间相对距离以及几何形状上的接近程度,是一个表征序列之间联系的量化指标。基于灰色位关联度,本文建立了一个发电设备状态评价的灰色模型,该模型通过计算被评估设备的状态参数序列与额定状态参数序列之间的灰色位关联度,实现对发电设备状态的定量评价,特别是较好地解决了个别参数严重偏离额定值时对设备状态的评价难题。通过对发电设备的风险分析,本文定义了设备运行脆弱度的概念,建立了发电设备运行风险评估模型。该模型充分考虑了设备状态、故障后果、运行趋势等因素的影响,通过对发电设备运行脆弱度模糊规则量化处理,构造脆弱度的三维曲面图来实现对发电设备的风险评估。在重要度分析的基础上,本文通过逻辑树分析实现了发电设备维修方式的优化,为设备故障模式确定最有效的维修方式。同时,建立了一个发电设备维修时刻优化模型,在考虑发电单位成本、销售电价、发电机组发电量、设备维修费用、机组停运损失、延长设备检修间隔而节省维修成本等因素的基础上,提出以发电机组运行效益最大为目标函数,解决了预知性检修中最佳维修时刻的确定问题。最后,根据以上研究成果,研究开发了发电设备优化维修支持系统,该系统包括设备管理、可靠性点检管理、故障诊断、设备状态评价、设备运行风险评估、维修决策、维修管理等模块,实现了一个完整的优化维修流程。该系统在电厂的实际应用中取得了较好的效果。更多还原

【Abstract】 Scientific decision making is the key guarantee for equipment’s safety, reliability and economic running. According to the actuality of maintenance basis optimization (MBO) being popularized in Chinese electric power industry and the problems during its application, the key technology during MBO, which are reliability spot inspection (RSI), fault diagnosis, condition prediction and estimation, risk evaluation and maintenance decision making, are researched, and the MBO system for equipment of steam power plant is also developed in this thesis.Aim at the problem of blindness during implement of spot inspection in steam power plant, RSI is proposed and the model of RSI is built. RSI is made of two phases which are static programming and dynamic regulation. From the point of system’s inherent structure function, static programming takes weighted fault mode, effect and criticality analysis (WFMECA) to equipments, and programs spot inspect tasks according to equipment’s inherent importance, which assures enough supervision on important equipments and avoids un-essential spot inspection tasks. Dynamic regulation is to estimate equipment’s condition periodically and regulate spot inspect tasks according to equipment’s present running risk, which avoids over spot inspection and deficiency spot inspection. Compared with traditional spot inspection, RSI can both assure the necessary reliability of equipment and save cost, thus, it improves the efficiency and economy of spot inspection. Importance analysis is the basis of RSI and maintenance decision making, a new equipment’s importance index that is synthetic weigh is proposed. Synthetic weigh reflects the practical importance of equipment in system by synthesizing the equipment’s inherent structure, basis function and running station.In order to deal with the inconsistent and redundance information, a model of vibration faults diagnosis based on Rough Set and information fusion is brought forward based on the analysis of vibration characteristic of rotary machine. The model constructs a Rough decision making system by using vibration information entropy and deduces decision making rules by reduction, which not only realizes the diagnosis for unification vibration imformation, but also for the inconsistent vibration imformation by a Rough factor. So, the model improves the identification of vibration fault mode. At the same time, a fault diagnosis model for electric equipment with complex fault modes is proposed too. The model realizes the fuzzy fault modes identification for this kind equipment by calculating the approach degrees between fault mode and standard modes.An improved GM(1,1) model is brought forward to disposing the problem of fluctuant character parameters caused by outside interference. The model translates the fluctuant sequence to a exponential rule sequence by a operator of m points, and constructs an identical demensions and a new information model to reduce the grey space, thus, it realizes the effective prediction for fluctuant sequence. Furthermore, a new concept of grey space relation is advanced and defined. Grey space relation reflects the approximation degree of two sequences both in distance and shape, which is a quantity index that denotes the relation between sequences. A grey model of condition evaluation for electric equipment based on grey space relation is constituted. By calculating the grey space relation between condition parameter sequence and rating condition parameter sequence of evaluated equipment, the approximation degree of equipment condition and rating condition is obtained, thus, the fix quantity evaluation of equipment is realized, especially, the model preferably evaluates the equipment condition when a few parameter deviate the rating condition parameter badly.According to the equipment’s risk analysis, equipment’s running vulnerability is defined and risk evaluation model is built. The model combines the factors of equipment’s condition, failure sequent and running tendency, and constructs the vulnerability’s 3-demensions drawing by quantifying the vulnerability’ fuzzy rules to realize the equipment’s risk evaluation. Base on the importance analysis, the maintenance mode optimization for equipment is realized by logic analysis, which provides the most effective maintenance mode for equipment’s fault. At the same time, a maintenance time optimization model is built. The model takes the unit’s benefit as object function by synthesizing the factors of generating unit cost, electricity selling price, unit generating electricity amount, maintenance cost, break down loss, and saved maintenance cost because of prolonging maintenance interval, then makes the optimization maintenance time for prediction maintenance. Finally, MBO system has been developed according to the above research. The system consists of equipment management module, RSI management module, fault diagnosis module, equipment’s condition prediction and estimation module, equipment’s running risk evaluation module, maintenance decision making module and maintenance management module, which realizes a whole MBO procedure. Some of the modules are developed and applied with good effect.更多还原