資源目錄里展示的全都有，所見即所得。下載后全都有,請放心下載。原稿可自行編輯修改=【QQ：401339828 或11970985 有疑問可加】 本科畢業(yè)設(shè)計外文文獻(xiàn)翻譯 （ 學(xué)生姓名： 學(xué) 院： 系 別： 專 業(yè)： 班 級： 指導(dǎo)教師： Layered clustering multi-fault diagnosis for hydraulic piston pump Efficient diagnosis is very important for improving reliability and performance of aircraft hydraulic piston pump, and it is one of the key technologies in prognostic and health management system. In practice, due to harsh working environment and heavy working loads, multiple faults of an aircraft hydraulic pump may occur simultaneously after long time operations. However, most existing diagnosis methods can only distinguish pump faults that occur individually. Therefore, new method needs to be developed to realize effective diagnosis of simultaneous multiple faults on aircraft hydraulic pump. In this paper, a new method based on the layered clustering algorithm is proposed to diagnose multiple faults of an aircraft hydraulic pump that occur simultaneously. The intensive failure mechanism analyses of the five main types of faults are carried out, and based on these analyses the optimal combination and layout of diagnostic sensors is attained. The three layered diagnosis reasoning engine is designed according to the faults' risk priority number and the characteristics of different fault feature extraction methods. The most serious failures are first distinguished with the individual signal processing. To the desultory faults, i.e., swash plate eccentricity and incremental clearance increases between piston and slipper, the clustering diagnosis algorithm based on the statistical average relative power difference (ARPD) is proposed. By effectively enhancing the fault features of these two faults, the ARPDs calculated from vibration signals are employed to complete the hypothesis testing. The ARPDs of the different faults follow different probability distributions. Compared with the classical fast Fourier transform-based spectrum diagnosis method, the experimental results demonstrate that the proposed algorithm can diagnose the multiple faults, which occur synchronously, with higher precision and reliability. Highlights The simultaneously occurred five main faults of aircraft hydraulic pump are diagnosed. Four diagnostic sensors' layout is obtained based on failure mechanism analyses. Three layered diagnosis reasoning engine is designed. Statistical average relative power difference algorithm diagnoses desultory faults. Layered clustering diagnosis method demonstrates superior reliability and accuracy.Diagnosis reasoning engine; Statistical average relative power difference (ARPD) 1. Introduction In the lifecycle of aircraft, the costs of maintenance and logistics occupy a majority of expenses according to the statistical data, so the prognostic and health management (PHM) system emerges in the last few years to guarantee the high reliability, safety and maintainability of aircraft [1]?and?[2]. The main attraction of PHM is that it not only could monitor the system's health and diagnose the faults in real time but also could predict the remaining useful life and provide the maintenance support comparing with conventional fault diagnosis technology [3]. After the successful application in the Joint Strike Fighter (JSF) program [4], PHM has been widely used in the utility systems and engine systems of aircrafts so as to reduce lifecycle cost and realize the autonomic logistics support [5]. Efficient diagnosis of aircraft hydraulic power system is one of the key technologies in PHM systems. Among aviatic utility systems, the hydraulic power system plays an important role of providing high pressure fluid to manipulate rudders and braking systems of airplane. Generally, three or four redundant hydraulic power systems are adopted in aerial design. As a key component in hydraulic power system, the engine driven pump must be reliable and safe during the flight processes. The data-driven diagnosis method treats the objective system as a “black box”, and analyzes the external measuring data to realize the fault diagnosis. For the aircraft hydraulic pump, its structure is very complex, and the relationship between the internal parameters is highly nonlinear. There are also strong couplings among various fault features. Therefore, accurate mathematical model of aircraft hydraulic pump is difficult to be established, and as a result, data-driven diagnosis method is commonly used for aircraft hydraulic pumps. Until now, much work has been done on fault diagnosis of hydraulic pumps. Zhao et al. [6] presented a method based on intermittent chaos and sliding window symbol sequence statistics to detect the early fault of one single piston loose shoes of hydraulic piston pump on a hydraulic tube tester. The method analyzed the pump discharge pressure through the time history diagram of Duffing oscillator, and used the sliding window symbol sequence statistics method to identify the state of Duffing oscillator to realize the early fault diagnosis. Gao et al. [7]?and?[8] found that the pump discharge pressure is an informative variable and carries lots of information to support a sensitive and reliable fault diagnosis of hydraulic pump. The wavelet analysis methods were used to improve the capability of diagnosing the health conditions of piston pump. The experimental results showed that these methods can identify several types of faults when the faults occur individually. Chen et al. [9] presented a method for the fault degradation assessment of the water hydraulic piston motor, which has similar structure as common oil hydraulic piston pump. The method was based on wavelet packet analysis and Kolmogorov–Smirnov test to analyze the impulsive energy of the vibration signals. The different piston conditions were detected. Zhao et al. [10] modified the traditional neighborhood rough set model and applied it in the fault feature extraction of hydraulic pump. The vibration signals of two different pump casings were analyzed in order to classify two kinds of faults. Wang and Chen [11] presented an integrated diagnosis method based on the wavelet transform, rough set and neural network. The wavelet function was used to extract fault features from vibration signals of hydraulic pump, and the diagnosis knowledge was acquired by the rough set. Then all diagnosis knowledge was used as input for the partially linearized neural network (PNN), and the output of the PNN was the possibility grades of different kinds of faults. Liu et al. [12] used the wavelet package analysis to eliminate the noise and extract the fault characteristics from original pump vibration signals. Then the improved Elman neural network was adopted to realize the mapping between the fault feature vectors and the failure modes. Hancock and Zhang [13] analyzed pump outlet pressure to detect defects of piston pumps. They created a fuzzy-neural network classifier to sort feature signals, which were preprocessed by using Fast Fourier Transform (FFT) and wavelet packet analyses. Zhou et al. [14] presented an information fusion application to identify pump malfunctions in terms of spectrum analysis and limit checking results on pump discharge pressure and oil temperature of the return line after the relief valve. Dong and He [15] built and trained a novel integrated framework based on the hidden semi-Markov model (HSMM) to realize the diagnosis and prognostic of hydraulic pumps. The vibration signals of pumps were processed by using wavelet packet decomposition and the wavelet coefficients were used as the inputs to the HSMMs. The different wear conditions of pumps were successfully classified and the remaining useful life was predicted as well. Except for the above methods, there are some other interesting diagnostic methods of hydraulic pumps and motors. Pietola and Varrio [16] investigated the thermography in the condition monitoring, fault diagnosis and predictive maintenance of fluid power components and systems. The shortages and limitations of the thermal imaging in the fault diagnosis of fluid power were discussed. Jardine et al. [17] used a proportional hazards modeling statistical approach to monitor a hydraulic motor health condition of a mine haul truck loader by regularly analyzing oil samples. Although oil analysis has been proven as a good method for timing oil changes and even fault detections, it has two drawbacks of the difficulty of obtaining a representative oil sample and temporal requirements for sampling, testing and evaluating results. Treuhaft et al. [18] presented a real-time monitoring approach of pistons and slippers wear condition of an axial piston pump by utilizing radioactive tracer technology. Through producing characteristic radionuclides (isotopes) in the slippers and pistons, the wear in an operating pump can be estimated by continuously monitoring the gamma-ray activity in the circulating fluid under various operating conditions. The methods listed above can realize correct diagnosis under certain conditions. Some of the methods can distinguish only one type of fault, some can distinguish various kinds of faults. According to the literature, all kinds of faults that can be diagnosed should occur individually, that is, the hydraulic pump should only have a single fault at a given moment. However, in real applications, the aircraft hydraulic pump is exposed to very harsh working environments with severe vibrations and high temperatures. The working load of aircraft hydraulic pump is very heavy under pressures as high as 28?MPa to even 35?MPa. Under these conditions, multiple faults on aircraft hydraulic pump are very likely to occur simultaneously. In comparison to diagnosis of single fault, diagnosis of multiple faults is much more challenging. The vibration signals and the discharge pressure of pump carry abundant information, which can be easily chosen as the diagnostic signal when only one fault occurs. However, when multiple faults occur simultaneously, the fault features will mix together, and the sensor signals will be polluted by the environmental noises. At this time, the problems of how to choose appropriate diagnostic sensors and how to extract effective fault features are essential to the diagnosis of multiple faults. The solutions to these problems are rarely found in published literatures. This paper proposes a new method to diagnose simultaneous multiple faults of an aircraft hydraulic pump by using the layered clustering algorithm. The intensive fault mechanism analysis of the five common faults of aircraft hydraulic pump is carried out. As a result the optimal combination and layout of diagnostic sensors is obtained. The layered reasoning engine is designed according to the fault risk priority number (RPN) and the characteristics of fault feature extraction methods. In the third diagnosis layer, the features of the two progressive faults are very weak and similar. In order to distinguish these two faults, the clustering diagnosis algorithm based on the statistical average relative power difference (ARPD) is proposed. This ARPD-based algorithm is compared with the classical Fast Fourier Transform (FFT)-based diagnosis method in terms of reliability and accuracy. （譯文） 分層聚類多液壓柱塞泵的故障診斷 高效的診斷對提高飛機(jī)的可靠性和液壓柱塞泵的性能是非常重要的，它是一個在預(yù)測與健康管理系統(tǒng)的關(guān)鍵技術(shù)。在實踐中，由于惡劣的工作環(huán)境和沉重的工作負(fù)荷，飛機(jī)液壓泵多故障可能同時長時間操作之后發(fā)生的。然而，大多數(shù)現(xiàn)有的診斷方法只能識別泵故障發(fā)生單獨。因此，需要開發(fā)新的方法來實現(xiàn)同時多故障對飛機(jī)液壓泵的有效診斷。 在本文中，基于分層聚類算法提出了一種飛機(jī)液壓泵同時發(fā)生多個故障診斷新方法。密集的失效機(jī)理分析的五種主要類型的故障進(jìn)行了分析，并在此基礎(chǔ)上優(yōu)化組合，達(dá)到診斷傳感器的布置。三層診斷推理機(jī)是根據(jù)故障的風(fēng)險優(yōu)先數(shù)和不同的故障特征提取方法的特點設(shè)計。最嚴(yán)重的故障信號處理與個人首先識別。到的故障，即，斜盤的偏心率和增量間隙增大柱塞和滑靴之間，聚類診斷算法基于統(tǒng)計平均相對功率差（ARPD）提出了。通過有效提高這兩種故障的故障特征，計算出的ARPDS振動信號來完成的假設(shè)檢驗。不同的故障ARPDS遵循不同的概率分布。基于快速傅里葉變換光譜診斷方法比較經(jīng)典，實驗結(jié)果表明，該算法能夠多故障診斷，可同時發(fā)生，具有更高的精度和可靠性。 同時發(fā)生的飛機(jī)液壓泵的五個主要的故障診斷。四診斷傳感器的布局是基于失效機(jī)理的分析得到。三層診斷推理機(jī)的設(shè)計。統(tǒng)計平均相對功率差分算法的故障診斷。分層聚類診斷方法顯示了卓越的可靠性和準(zhǔn)確性。 診斷推理引擎；統(tǒng)計平均相對功率差（ARPD） 1、簡介 在飛機(jī)的生命周期，維護(hù)和物流成本占據(jù)統(tǒng)計數(shù)據(jù)的大部分費用，所以與健康管理（PHM）的預(yù)后系統(tǒng)出現(xiàn)在過去的幾年里，為了保證高可靠性，安全性和飛機(jī)[ 1 ]和[ 2可維護(hù)性]。PHM主要的吸引力在于它不僅可以監(jiān)測系統(tǒng)的健康和實時故障診斷也是可以預(yù)測的剩余使用壽命，提供與傳統(tǒng)故障診斷技術(shù)[ 3比較維護(hù)支持]。在聯(lián)合攻擊戰(zhàn)斗機(jī)（JSF）計劃的成功應(yīng)用后[ 4 ]，PHM已廣泛應(yīng)用于本系統(tǒng)和飛機(jī)發(fā)動機(jī)系統(tǒng)以降低生命周期成本和實現(xiàn)自主物流支持[ 5 ]。 飛機(jī)液壓系統(tǒng)有效的診斷是一個在PHM系統(tǒng)的關(guān)鍵技術(shù)。航空實用系統(tǒng)中，液壓動力系統(tǒng)中提供高壓力流體操縱舵和制動系統(tǒng)飛機(jī)的重要作用。一般來說，三或四的多余的液壓動力系統(tǒng)在空中的設(shè)計采用了。在液壓動力系統(tǒng)的關(guān)鍵部件，發(fā)動機(jī)驅(qū)動泵必須在飛行過程中是可靠的和安全的過程。數(shù)據(jù)驅(qū)動的診斷方法對目標(biāo)系統(tǒng)作為一個“黑盒子”，并分析了外部測量數(shù)據(jù)來實現(xiàn)故障診斷。對飛機(jī)液壓泵，它的結(jié)構(gòu)非常復(fù)雜，和內(nèi)部參數(shù)之間的關(guān)系是高度非線性的。也有強(qiáng)烈的耦合之間的各種故障特征。因此，飛機(jī)液壓泵精確數(shù)學(xué)模型難以建立，并且作為一個結(jié)果，數(shù)據(jù)驅(qū)動的診斷方法通常用于飛機(jī)液壓泵。 直到現(xiàn)在，許多工作已經(jīng)在液壓泵故障診斷。趙等人。[ 6 ]提出了一種基于混沌的方法和滑動窗口的符號序列統(tǒng)計檢測單柱塞松鞋在水壓試驗機(jī)液壓柱塞泵早期故障。該方法分析了泵的排出壓力通過Duffing振子的時間歷程圖，并使用滑動窗口的符號序列統(tǒng)計的方法來確定Duffing振蕩器實現(xiàn)早期故障診斷狀態(tài)。Gao等人。[ 7 ]和[ 8 ]發(fā)現(xiàn)泵出口壓力是一個內(nèi)容豐富的變量進(jìn)行信息支持一個敏感的液壓泵故障診斷的可靠的多。小波分析方法被用來提高柱塞泵的健康狀況診斷的能力。實驗結(jié)果表明，這些方法可以識別幾種類型的故障發(fā)生故障時單獨。Chen等人。[ 9 ]提出的水液壓柱塞馬達(dá)故障退化評估方法，它具有類似的結(jié)構(gòu)，常見的液壓柱塞泵。該方法是基于小波包分析和Kolmogorov–斯米爾諾夫檢驗分析的振動信號的脈沖能量。不同的活塞的條件下進(jìn)行檢測。趙等人?！?0】修正傳統(tǒng)的鄰域粗糙集模型，并將其應(yīng)用于液壓泵的故障特征提取。兩個不同的泵殼體振動信號進(jìn)行分析，以兩種故障的分類。王、陳[ 11 ]提出了一種基于小波變換的綜合診斷方法，粗糙集和神經(jīng)網(wǎng)絡(luò)。小波函數(shù)被用來從液壓泵振動信號的故障特征提取和診斷知識，采用粗糙集獲取。然后所有的診斷知識作為部分線性化神經(jīng)網(wǎng)絡(luò)（PNN），輸入和輸出的概率是不同類型的故障的可能性等級。劉等人。[ 12 ]利用小波包分析消除噪聲和原始泵振動信號中提取故障特征。將改進(jìn)的Elman神經(jīng)網(wǎng)絡(luò)來實現(xiàn)故障特征向量和故障模式之間的映射。漢考克和張[ 13 ]分析泵出口壓力檢測缺陷的活塞泵。他們創(chuàng)造了一個模糊神經(jīng)網(wǎng)絡(luò)分類器的分類特征信號，并利用快速傅立葉變換（FFT）預(yù)處理和小波包分析。周等人。[ 14 ]提出了一種信息融合應(yīng)用識別泵故障的頻譜分析和極限檢查泵的排出壓力和回油管油溫度安全閥后的結(jié)果。東和他[ 15 ]建立和培養(yǎng)一個新的集成框架，基于隱半馬爾可夫模型（HSMM）實現(xiàn)了液壓泵的診斷和預(yù)后。泵的振動信號進(jìn)行小波包分解和小波系數(shù)的處理被用來作為HSMMs輸入。泵磨損條件的不同進(jìn)行分類和預(yù)測以及剩余使用壽命。 除了以上方法外，還有一些其它的有趣的診斷方法液壓泵和馬達(dá)。pietola和馬里厄[ 16 ]研究的熱狀態(tài)監(jiān)測中，液壓元件和系統(tǒng)的故障診斷與預(yù)測維修。在液壓故障診斷熱成像的不足和局限性進(jìn)行了討論。怡和等。[ 17 ]用比例風(fēng)險統(tǒng)計的定期分析油樣監(jiān)測礦井運輸卡車裝載機(jī)液壓馬達(dá)的健康狀況的建模方法。盡管石油分析已被證明是定時換油和故障檢測的一種好方法，它有兩個缺點，獲得有代表性的油樣和采樣時間要求的困難，測試和評估結(jié)果。特里哈富特等人。[ 18 ]提出了一個活塞和拖鞋實時監(jiān)測方法的軸向柱塞泵磨損條件下利用放射性示蹤技術(shù)。通過生產(chǎn)特征放射性核素（同位素）的拖鞋和活塞，在一個工作泵的磨損可以通過連續(xù)監(jiān)測伽馬射線活動在循環(huán)流體在不同操作條件下估計。 上述方法可以實現(xiàn)正確的診斷在一定條件下。有些方法可以區(qū)分，只有一種類型的故障，一些可以區(qū)分各種故障。根據(jù)文獻(xiàn)，各種故障的可診斷應(yīng)單獨出現(xiàn)，那就是，液壓泵應(yīng)該只在某一時刻有一個單一的故障。然而，在實際應(yīng)用中，飛機(jī)液壓泵暴露于嚴(yán)重的振動和高溫非常惡劣的工作環(huán)境。飛機(jī)液壓泵的工作負(fù)載很重的下壓力高達(dá)28兆帕至35兆帕。在這些條件下，對飛機(jī)液壓泵多故障同時出現(xiàn)，很有可能。 在單一故障診斷比較，多故障診斷是更具挑戰(zhàn)性的。振動信號和泵的排出壓力攜帶了豐富的信息，可作為診斷信號時，只有一個故障發(fā)生時。然而，當(dāng)多個故障同時發(fā)生故障的特點，將混合在一起，和傳感器信號受環(huán)境噪聲污染。在這個時候，的關(guān)鍵問題是如何選擇適當(dāng)?shù)脑\斷傳感器和如何提取有效故障特征是多故障診斷的基礎(chǔ)。這些問題的解決方案是很少在文獻(xiàn)中發(fā)現(xiàn)的。 本文提出了采用分層聚類算法診斷飛機(jī)液壓泵多故障的一種新方法。對飛機(jī)液壓泵常見故障五密集的故障機(jī)理進(jìn)行分析。作為一個結(jié)果的優(yōu)化組合和診斷傳感器的布置。分層推理機(jī)是根據(jù)故障風(fēng)險優(yōu)先數(shù)（RPN）設(shè)計和故障特征提取方法的特點。在第三個診斷層，這兩個漸進(jìn)性故障的特點是非常弱的，類似的。為了區(qū)分這兩種故障診斷算法的聚類，基于統(tǒng)計平均相對功率差（ARPD）提出了。本文基于arpd算法相比，經(jīng)典的快速傅立葉變換（FFT）為基礎(chǔ)的診斷方法的可靠性和準(zhǔn)確性方面。 本科生畢業(yè)設(shè)計(論文)開題報告 學(xué)生姓名 學(xué) 號 班 級 指導(dǎo)教師 職 稱/學(xué) 位 講師/碩士 題目名稱 150MPa手動液壓泵設(shè)計及運動仿真 題目類型 裝備裝置類 題目的意義、目的： 超高壓液壓手動泵具有壓力高、重量輕、結(jié)構(gòu)緊湊、操作簡單等特點，特別是于野外無電源情況下使用。廣泛應(yīng)用于液壓千斤頂，液壓拉伸器，液壓拔輪器，液壓螺栓，液壓螺母等手動打壓系統(tǒng)。本課題通過對超高壓液壓手動泵的系統(tǒng)設(shè)計使學(xué)生對所學(xué)知識全面復(fù)習(xí)并能夠綜合應(yīng)用。 設(shè)計（研究）主要內(nèi)容及方案： 內(nèi)容：手動液壓泵配置包括：超高壓壓力表、超高壓軟管、溢流閥、單向閥、油箱及快速接頭的手動泵組；即為全套的手動超高壓系統(tǒng)。高壓軟管，兩端安裝高壓快速接頭；便于在使用過程迅速的安裝于拆卸。裝配完成后就可以打壓。雙速操作手動液壓泵，手柄行程比單速操作時減少78%手柄省力，降低操作者勞動強(qiáng)度.可鎖手柄和輕巧的結(jié)構(gòu)便于攜帶。油箱容量大，可適用于較多型號的油缸和工具。內(nèi)置安全閥，用于手動液壓泵過載保護(hù)。 方案：柱塞式液壓泵是靠柱塞在柱塞腔內(nèi)的往復(fù)運動，改變柱塞腔容積實現(xiàn)吸油和排油的。當(dāng)手柄運動時帶動低壓泵高壓泵，同時供油，液壓油充滿工作腔后壓力逐漸升高，當(dāng)壓力升高到調(diào)定壓力時，溢流閥開啟，低壓泵的液壓油回到油箱。此時，液壓油基本充滿工作腔，由高壓柱塞繼續(xù)小流量地供油提高壓力當(dāng)壓力達(dá)到額定工作壓力時，安全閥開啟，系統(tǒng)保持最高的工作壓力。在裝配完畢后，進(jìn)入裝配環(huán)境，首先泵體“固定”，手桿和泵體用“銷釘”連接裝配，手桿和滑套用“圓柱”連接裝配，滑套和活塞桿用“銷釘”連接裝配，活塞桿和活塞用“剛體”連接裝配。裝配完畢后，進(jìn)入“應(yīng)用程序”中的“機(jī)構(gòu)”，選“電機(jī)”，“運動軸”選手桿上銷孔的軸線，再進(jìn)入“機(jī)構(gòu)分析”，設(shè)置“終止時間”20，“幀數(shù)”10，把手桿的水平位置設(shè)為“初始位置”，完成仿真。 工作進(jìn)度安排（具體）： 第1 周 收集相關(guān)資料、寫開題報告 第2 周 復(fù)習(xí)液壓傳動中關(guān)于液壓泵的相關(guān)基礎(chǔ)知識 第3 周 初步計算和進(jìn)行相應(yīng)的校核 第4 周 學(xué)習(xí)pro-e相關(guān)知識 第5-7周 完成實體化建模并進(jìn)行運動仿真 第8-9周 用autocad完成二維圖 第10-12周 編寫說明書并整理畢業(yè)設(shè)計全部文件，準(zhǔn)備答辯 指導(dǎo)教師意見： 指導(dǎo)教師簽名： 年 月 日