發(fā)布時間：2018-05-20 04:29

  本文選題：TRT + 神經(jīng)網(wǎng)絡��； 參考：《上海交通大學》2012年碩士論文

【摘要】：機械設備是企業(yè)生產(chǎn)的物質(zhì)基礎,是生產(chǎn)力的重要組成部分。在工業(yè)生產(chǎn)中,隨著時間的推移,各種設備必然會產(chǎn)生各種形式的磨損,以及導致設備精度和效率的降低,從而使產(chǎn)品質(zhì)量下降,嚴重的還會造成設備事故,為此開展對設備故障及性能診斷技術的課題研究意義重大。 故障及性能診斷是一門快速發(fā)展的交叉學科,它集測試技術、軟件工程、計算機技術、信號處理、模式識別、人工智能、決策科學、信息科學等眾多現(xiàn)代科學技術于一體,成為既注重理論研究,又重視實際應用的現(xiàn)代工程科學,并逐步形成一個體系完整、理論嚴謹且具有重大工程意義的新學科。從當前我國大多數(shù)企業(yè)對設備故障的處理體制來看,基本上都采用了習慣的“定期維修”和“事后維修”兩種方式。定期維修雖能發(fā)現(xiàn)一些早期故障,防止一部分突發(fā)事故的發(fā)生,但會因為對不需要維護的設備過頻更換零部件而造成過剩維修,形成一些不必要的浪費。對于事后維修而言,任何異常工況和突發(fā)故障導致的停機檢修和生產(chǎn)節(jié)拍的停頓,都必然造成生產(chǎn)工序的積壓,嚴重地影響生產(chǎn)計劃的順利完成。因此需要在生產(chǎn)過程中及早地對設備進行故障診斷,做到“先人一步發(fā)現(xiàn)故障,先故障之前消除隱患”。 本課題以萊蕪鋼鐵股份公司能源動力廠5#高爐煤氣余熱余壓能量回收透平發(fā)電裝置(Blast-Furnace Top pressure Recovery Turbine Unit,簡稱TRT)的工況監(jiān)測與故障診斷為研究內(nèi)容,研究的目的是要根據(jù)TRT在各種工況下表現(xiàn)出來的振動、噪聲、溫度、液壓、轉(zhuǎn)子、轉(zhuǎn)速、氣味、泄露等所有規(guī)律特征信息去綜合分析和識別設備工作狀態(tài)、故障類型和故障的嚴重程度,最終得到對修復故障有重要指導作用的診斷結(jié)論。本文在分析TRT常見的故障機理基礎上,深入研究了神經(jīng)網(wǎng)絡技術的原理方法和應用技術特點,結(jié)合故障特點找出與其相對應的特征量,構(gòu)建了TRT故障診斷系統(tǒng)的神經(jīng)網(wǎng)絡模型,并針對TRT透平轉(zhuǎn)子故障樣本進行了神經(jīng)網(wǎng)絡訓練,基于VB軟件實現(xiàn)了TRT系統(tǒng)的故障診斷界面開發(fā)。本文應用故障診斷技術實現(xiàn)了對機組的保護,避免了因高爐頂壓瞬間增大而導致的不必要停車,為保證高爐TRT長期安全、穩(wěn)定、順行以及提高經(jīng)濟效益等提供了有力支持。
[Abstract]:Mechanical equipment is the material basis of enterprise production and an important part of productivity. In industrial production, with the passage of time, various kinds of equipment will inevitably produce various forms of wear and tear, and will lead to the reduction of equipment precision and efficiency, thus leading to a decline in product quality and serious equipment accidents. Therefore, it is of great significance to research the technology of equipment fault and performance diagnosis. Fault and performance diagnosis is a rapidly developing interdisciplinary subject. It integrates testing technology, software engineering, computer technology, signal processing, pattern recognition, artificial intelligence, decision science, information science and so on. It has become a modern engineering science which pays attention to both theoretical research and practical application, and gradually forms a new discipline with complete system, rigorous theory and great engineering significance. According to the current system of handling equipment failures in most enterprises in our country, two methods of "regular maintenance" and "afterwards maintenance" are basically adopted. Although periodic maintenance can find some early faults and prevent some sudden accidents from happening, it will cause excessive maintenance because of the excessive replacement of parts and components for the equipment that does not need maintenance, resulting in some unnecessary waste. For the maintenance after the event, any abnormal working conditions and sudden failures will inevitably cause the backlog of production procedures, which will seriously affect the smooth completion of production planning. Therefore, it is necessary to diagnose the equipment as early as possible in the process of production, so as to "find the fault first and eliminate the hidden trouble before the failure". In this paper, the monitoring and fault diagnosis of blast furnace gas residual heat residual pressure recovery turbine generator unit, Blast-Furnace Top pressure Recovery Turbine Unit, is taken as the research content in Laiwu Iron and Steel Co., Ltd. The purpose of the study is to comprehensively analyze and identify the working state of the equipment according to the characteristic information of vibration, noise, temperature, hydraulic pressure, rotor, speed, smell, leakage and so on, which are shown by TRT under various working conditions. Finally, the fault type and the severity of the fault can be used as an important guide for fault diagnosis. On the basis of analyzing the common fault mechanism of TRT, this paper deeply studies the principle, method and application technical characteristics of neural network technology, finds out the corresponding characteristic quantity according to the fault characteristic, and constructs the neural network model of TRT fault diagnosis system. The neural network training for the fault sample of TRT turbine rotor is carried out, and the fault diagnosis interface of TRT system is developed based on VB software. In this paper, the fault diagnosis technology is used to protect the unit and avoid the unnecessary stop caused by the instantaneous increase of the top pressure of the blast furnace, which provides a strong support for ensuring the long-term safety, stability, smooth running of the blast furnace TRT and improving the economic benefit.
【學位授予單位】：上海交通大學
【學位級別】：碩士
【學位授予年份】：2012
【分類號】：TH165.3

【引證文獻】

相關碩士學位論文 前1條

1 吳軍;火炮狀態(tài)智能診斷技術研究[D];南京理工大學;2013年

，

本文編號：1913206