汽車變速器殼體零件加工工藝與夾具設計【含CAD圖紙、說明書、三維模型】,含CAD圖紙、說明書、三維模型,汽車,變速器,殼體,零件,加工,工藝,夾具,設計,cad,圖紙,說明書,仿單,三維,模型 畢業(yè)設計（論文）任務書 姓名 畢業(yè)設計（論文）題目 汽車變速器殼體零件加工工藝與夾具設計 校內(nèi)指導教師 學 歷 職 稱 教授 校外指導教師 職稱或職務 單 位 無 主要任務及要求： 1.汽車變速器殼體零件三維幾何造型：使用 Pro/Engineer 軟件對汽車變速器殼體零件進行幾何造型。2.汽車變速器殼體零件加工工藝與夾具設計對零件進行加工工藝分析、編寫加工工藝路線、編寫工藝卡片，設計一副夾具。切削用量的計算與夾具的三維模型以及夾具的裝配圖、零件圖。應用計算機設計、計算、繪圖，設計說明書字數(shù)不少于 1.5 萬字，參考文獻不得少于 30 篇 且需要有一定數(shù)量的英文參考文獻，翻譯一篇英文文獻，譯文字數(shù)不少于 2000 單詞。進度安排： 第 1 周：收集資料、撰寫開題報告、外文翻譯 第 2~5 周：總體加工工藝路線方案設計、主要參數(shù)計算、殼體零件的建模 第 6~9 周：制定加工工藝路線、編寫工藝卡片 第 10~14 周：繪制二維圖紙、撰寫畢業(yè)論文 第 15 周：對論文材料進行修改以及準備答辯材料 第 16 周：答辯 指導教師簽字： 年 月 日 系（部）意見： 同 意 系（部）主任簽章： 年 月 日 開始日期 交稿日期 附注： 利用機器學習技術對不同冷卻/潤滑條件下的車削操作進行預測建模，以實現(xiàn)可持續(xù)制造 Djordje Cica a,? , Branislav Sredanovic a , Sasa Tesic a , Davorin Kramar b 關鍵字：機器學習，可持續(xù)加工，加工力，切割力，切割壓力 摘要 可持續(xù)制造是當前工業(yè)中最重要和最具挑戰(zhàn)性的問題之一。為了減少切削液帶來的負面影響，機械加工工業(yè)界正在不斷發(fā)冷卻和潤滑切削的技術和系統(tǒng)區(qū)域，同時保持加工效率。在本研究中基于三種回歸的機器學習技術，即多項式回歸（PR），支持向量回歸（SVR）和高斯過程開發(fā)了回歸（GPR）來預測加工力，切削能力和切削壓力在AISI 1045的轉變中。在預測模型的開發(fā)中，切削的加工參數(shù)速度，切削深度和進給速度被視為控制因素。由于冷卻/潤滑技術嚴重影響加工性能，預測模型發(fā)展的質量特征在最小量潤滑（MQL）和高壓冷卻液（HPC）切削條件下進行。通過統(tǒng)計誤差分析方法評估開發(fā)模型的預測準確性。還對基于回歸的機器學習技術的結果與以下方法之一進行了比較：最常用的機器學習方法，即人工神經(jīng)網(wǎng)絡（ANN）。最后，一個利用基于神經(jīng)網(wǎng)絡算法的元啟發(fā)式方法來執(zhí)行高效兩種切削環(huán)境的工藝參數(shù)的多目標優(yōu)化。 1. 介紹 切削液傳統(tǒng)上用于金屬切削操作提高刀具壽命，表面質量以及整個加工工藝生產(chǎn)率。 但是，由于存在潛在的危害有害化學物質，切削液會產(chǎn)生負面影響。此外，使用切削液代表了相當大的總制造成本 并證明了切削液約占總加工量的7％至17％費用。如今，常規(guī)水冷是最常見的冷卻/潤滑技術，用于提高加工性能。但是，切削液消耗量大，功率大消耗，冷卻/潤滑能力差，浪費過多管理以及與人類健康和環(huán)境有關的問題是這方面最重要的地方。 全球產(chǎn)量的增長，因此增長切削液的應用引起了有關冷卻/潤滑系統(tǒng)的經(jīng)濟和環(huán)境方面切割區(qū)。因此，最近幾次冷卻/潤滑開發(fā)技術以實現(xiàn)可持續(xù)制造通過減少或消除切削液。目前，負壓低的最廣泛使用的冷卻/潤滑技術對環(huán)境和操作人員健康的影響包括：干式切削，低溫冷卻，最少潤滑（MQL），高壓冷卻液（HPC）或用作冷卻/生物可降解油的潤滑液。除了更多經(jīng)濟和環(huán)境可持續(xù)發(fā)展的這些新技術與傳統(tǒng)洪水相比，效率更高冷卻。 據(jù)報道表面有明顯改善質量、工具壽命、生產(chǎn)率、總成本等。 由于制造操作消耗大量的事實競爭性制造業(yè)需要的能源數(shù)量高效節(jié)能的加工工藝，最大程度減少負面影響對環(huán)境以及降低成本。 例如，根據(jù)周期等。每年全球消費總量的20％能源用于制造業(yè)， 除了環(huán)保和清潔的加工過程，廢物管理和減少，節(jié)能代表了以下方面的重要指標之一：可持續(xù)生產(chǎn)。 因此，行業(yè)正在尋找消耗更少能源的替代制造方法。 加工過程的特征是存在大量高度相關的參數(shù)。 由于高金屬切削現(xiàn)象的復雜性和非線性，這非常復雜甚至無法制定適當?shù)姆治龇椒ㄊ褂没谶^程物理原理的傳統(tǒng)方法進行建模。如今，趨勢正在轉向預測模型這些過程使用機器學習方法。 在過去幾十年，人工神經(jīng)網(wǎng)絡（ANN）和多元回歸已經(jīng)成功地實施了各種預測干燥和常規(guī)條件下的車削質量特征冷卻液的供給，例如表面粗糙度，切削力，刀具磨損，比切削力，切削功率和切削溫度。 如今，學術研究和產(chǎn)業(yè)努力正在針對消除挑戰(zhàn)，或至少最大限度地減少切削液的使用量，同時保持工藝效率科學。 但是，涉及機器學習技術在估計中的應用在環(huán)保工藝中的加工響應，例如作為MQL和HPC輔助的加工環(huán)境制定兩個模型，即ANFIS和基于ANN的模型進行預測主切削力，進給力和被動力的關系。 三種不冷卻和潤滑條件（洪水，MQL和HPC），切削深度，進給速度，切削速度用作變量用于切削力分量建模。他們得出結論，兩者模型可以有效地用于預測切削力分量在車削作業(yè)中。建議的人工神經(jīng)網(wǎng)絡和支持用于表面粗糙度，切割的向量回歸（SVR）模型Ti-車削時的溫度和切屑系數(shù)預測6Al-4V合金。切削速度，進給速度，切削條件（干燥和HPC），而旋轉力是輸入變量。 他們展示了兩種方法都可以成功地用于預測加工響應。建模的工具壽命和表面HPC環(huán)境下100Cr6鋼硬車削時的粗糙度使用ANN和ANFIS。 實驗加工數(shù)據(jù)如飼料，切割速度和時間在本研究中用于訓練和評估兩個模型。 兩者獲得的估計結果模型與實驗結果進行比較，非常好遵守協(xié)議。Mia和Dhar開發(fā)了基于ANN的EN 24T硬車削中表面粗糙度的預測模型在干燥和高壓冷卻液噴射加工環(huán)境下的鋼材。除切削條件外，切削速度，進給速度和材料硬度用作輸入變量。 不同的人工神經(jīng)網(wǎng)絡架構和幾種訓練方法被用來確定最佳的預測模型。Mia和Dhar 制定表面粗糙度的兩個預測模型，即支持向量回歸中的回歸和響應面方法（RSM）AISI 1060鋼在干燥和HPC條件下使用。切割速度進給速度和材料硬度被視為輸入變量用于模型制定。 結果表明兩種方法都可以用來預測干車削時的粗糙度值，而在HPC中，支持向量回歸模型優(yōu)于RSM輔助轉向。ANN模型切削溫度，取決于切削速度，進給速度，深度切割工件材料（C-60、17CrNiMo4和42CrMo4）并切割環(huán)境（干燥，潮濕和HPC），準確度為97.3％。制定了基于ANN的表面預測模型MQL輔助硬車削的粗糙度模型，其中切削輸入速度，進料速度和MQL流速。 他們的結果表明人工神經(jīng)網(wǎng)絡模型可以保持97.5％的準確性。米婭等利用SVR預測平均表面相對于主軸轉速，進給速度，深度的粗糙度參數(shù)MQL輔助車削中脈沖之間的切割和時間間隔的變化硬度鋼。他們的結果表明，開發(fā)的模型能夠以95.04％的準確度預測輸出響應。 阿巴斯等開發(fā)了表面粗糙度的回歸模型干濕和納米流體時的功耗和功耗轉向AISI 1045。Nouioua等利用回應曲面方法和人工神經(jīng)網(wǎng)絡技術來尋找最佳預測車削時表面粗糙度和切削力的預測X210Cr12鋼根據(jù)切削速度，進給速度和切削量的不同在干、濕和MQL加工條件下的深度，人工神經(jīng)網(wǎng)絡被發(fā)現(xiàn)比響應面方法學模型更好預測切削參數(shù)。 機器學習技術已被廣泛應用于對車削中不同加工響應的預測。 然而，提出的文學模型主要涉及干濕切割。 此外，很少有可利用的信息提供有關切割能量的預測以及不同冷卻/潤滑條件下的切削壓力 條件。這項研究提出了預測模型的發(fā)展加工力，切削能量和切削壓力 使用三種基于回歸的機器學習技術進行轉彎（多項式回歸，支持向量機和高斯過程回歸）以及人工神經(jīng)網(wǎng)絡。與其他呈現(xiàn)的作品相反，這里是對選定作品的估計針對不同的冷卻/潤滑進行了機加工響應條件。特別是，該研究涵蓋了MQL和HPC加工條件。 選定的機器學習技術是而且用于比較加工評估為了確定最佳方法，根據(jù)模型的準確性和能力。 另外，多目標優(yōu)化問題也在進行中。 2. 實驗細節(jié) 以棒材形式提供的AISI 1045（C45E）鋼的直車削直徑120毫米，300毫米的勃林格車床開發(fā)了8千瓦的主軸功率硬質合金刀片SNMG 1204 08 NMX。 在這項研究中，重點是應用于各種冷卻/潤滑技術在加工中。 因此，實驗是在不同的條件下進行的加工環(huán)境，即MQL和HPC的實驗設置中附帶了MQL和HPC系統(tǒng)在機加工試驗中。 對于MQL輔助車削，將切削液供應給噴涂噴槍以30毫升/小時的速度與壓縮空氣混合（3 bar）在噴槍的混合室中。 然后的混合物 噴槍在切割區(qū)域提供空氣和切削液噴嘴位于距刀尖30 mm的位置，成90度角和30，分別從切削刃和后刀面開始。 在HPC輔助車削過程中，切削液的供應量為恒定壓力為50 bar，流速為2 l / min垂直于切削刃的0.4毫米噴嘴（直徑）低與切削工具前刀面成角度（約5–6）。 噴嘴定位在距刀尖30毫米處，目的是達到相當接近工具-芯片接觸區(qū)的程度，并減少噴嘴對流動屑的干擾。 除了不同的加工環(huán)境外，三切削參數(shù)，例如切削速度（v），切削深度（a）和進給速度（f），這些參數(shù)的范圍是根據(jù)建議選擇的刀具制造商的意見，并按照之前學習。 此外，參數(shù)范圍也擴大了為了獲得更高的生產(chǎn)率并研究機加工在不同加工環(huán)境中的響應。 切削力的三個組成部分，即主切削測量力（Fc），進給力（Ff）和被動力（Fp）使用Kistler測力計9259A。 測量鏈還包括一個電荷放大器（奇石樂5001），頻譜分析儀（HP3567A）和用于數(shù)據(jù)采集的個人計算機和分析。 加工力（FR），切削力（Pc）和切削壓力（Ks）由以下公式計算： 獲得的實驗數(shù)據(jù)分為兩個數(shù)據(jù)-集，即用于模型開發(fā)的訓練數(shù)據(jù)集（75％整個數(shù)據(jù)集）和用于模型驗證的測試數(shù)據(jù)集（占數(shù)據(jù)的25%的整個數(shù)據(jù)集）。 因此，隨機選擇了27組實驗試驗用于模型構建，其余9個測試模型性能的數(shù)據(jù)集。 相同的數(shù)據(jù)分區(qū)該方案用于MQL和HPC加工條件。 Predictive modeling of turning operations under different cooling/ lubricating conditions for sustainable manufacturing with machine learning techniques Djordje Cica a,? , Branislav Sredanovic a , Sasa Tesic a , Davorin Kramar b Abstract Sustainable manufacturing is one of the most important and most challenging issues in present industrial scenario. With the intention of diminish negative effects associated with cutting fluids, the machining industries are continuously developing technologies and systems for cooling/lubricating of the cutting zone while maintaining machining efficiency. In the present study, three regression based machine learning techniques, namely, polynomial regression (PR), support vector regression (SVR) and Gaussian process regression (GPR) were developed to predict machining force, cutting power and cutting pressure in the turning of AISI 1045. In the development of predictive models, machining parameters of cutting speed, depth of cut and feed rate were considered as control factors. Since cooling/lubricating techniques significantly affects the machining performance, prediction model development of quality characteristics was performed under minimum quantity lubrication (MQL) and high-pressure coolant (HPC) cutting conditions. The prediction accuracy of developed models was evaluated by statistical error analyzing methods. Results of regressions based machine learning techniques were also compared with probably one of the most frequently used machine learning method, namely artificial neural networks (ANN). Finally, a metaheuristic approach based on a neural network algorithm was utilized to perform an efficient multi-objective optimization of process parameters for both cutting environment. 1. Introduction Cutting fluids are traditionally used in metal cutting operations to improve the tool life, surface quality as well as entire machining process productivity. However, cutting fluids have negative effects on the human health and environment due to presence of potentially harmful chemicals . In addition, the use of cutting fluids represents a considerable amount of total manufacturing costs . Weinert et al. demonstrates that the estimated cost of the cutting fluids is around 7 to 17% of the aggregate machining costs. Nowadays, conventional flood cooling is the most common cooling/lubricating technique used to improve machining performance. However, high cutting fluid consumption as well as power consumption, poor cooling/lubrication capability, excessive waste management and problems related to human health and environmental issues are some of the most important disadvantages of this method. The growth of global production and consequently the increase of cutting fluids application caused intensive research concerning economic and environmental aspects of systems for cooling/lubricating the cutting zone. Thus, recently several cooling/lubrication techniques were developed in order to achieve sustainable manufacturing by reducing or eliminating of cutting fluids. Currently, the most widely used cooling/lubricating techniques with low negative effect on the environment and human operator’s health are: dry cutting, cryogenic cooling, minimum quantity lubrication (MQL), high-pressure coolant (HPC), or utilization as a cooling/ lubricating fluid the biodegradable oils. Apart from being more economical and environmentally sustainable, these new technologies are also more efficient as compared with traditional flood cooling. Considerable improvements have been reported in surface quality, tool life, productivity, total costs, etc. Due to fact that manufacturing operations consume significant amounts of energy, competitive manufacturing industries require energy efficient machining processes to minimize negative effect on the environment as well as to reduce costs. For instance, according to Zhou et al. about 20% of overall consumed annual worldwide energy is used in manufacturing. In addition to eco-friendly and clean machining process, waste management and reduction, conserving energy represents one of the important indicators of sustainable production. Therefore, industries are searching for alternative manufacturing methods in which less energy is consumed . Machining processes are characterized by the presence of a large number of highly correlated parameters. Due to the high complexity and nonlinearity of metal cutting phenomena, it is very complicated or even impossible to formulate an adequate analytical model using traditional methods based on the process’s physics. Nowadays, the trends are towards predictive modeling of these processes using machine learning methods. Over the last few decades, artificial neural networks (ANN) and multiple regression have been successfully implemented in the prediction of various quality characteristics in turning under dry and conventional coolant supply, such as surface roughness , cutting force , tool wear , specific cutting force , cutting power and cutting temperature . Nowadays, academic research as well as industrial efforts are being directed towards the challenge of elimination or, at least, minimization of cutting fluids use, while preserving process effi- ciency. However, there are significantly less studies dealing with application of the machine learning techniques in estimating machining responses in environmentally friendly processes, such as MQL and HPC-assisted machining environments. Cica et al. formulate two models, namely, ANFIS and ANN-based, for prediction of main cutting force, feed force and passive force. Three different cooling and lubricating conditions (flood, MQL and HPC), depth of cut, feed rate, cutting speed were used as the variables for cutting force components modeling. They concluded that both models can be used effectively to predict the cutting force components in turning operations. Mia et al. proposed ANN and upport vector regression (SVR) models for surface roughness, cutting temperature and chip coefficient prediction when turning of Ti- 6Al-4V alloy. Mia et al. utilized the SVR for the prediction of average surface roughness parameter with respect to spindle speed, feed rate, depth of cut and time gap between pulsing in MQL assisted turning of high hardness steel. Their results show that the developed model is able to predict the output responses with 95.04% accuracy. Abbas et al. [31] developed the regression models for the surface roughness and power consumption under dry, wet and nanofluid MQLassisted turning of AISI 1045. Nouioua et al. utilized response surface methodology and ANN technique to search for optimal prediction of predicting surface roughness and cutting force in turning of X210Cr12 steel according to cutting speed, feed rate and cutting depth under dry, wet and MQL machining conditions. ANN were found to be better than the response surface methodology model in the prediction of cutting parameters. Machine learning techniques have been extensively utilized in the prediction of different machining responses in turning. However, presented models in literature mainly dealt with dry and wet cutting. Furthermore, a very few utilizable information is provided regarding the prediction of the cutting energy and as well as the cutting pressure under different cooling/lubricating conditions. This study presents a prediction models development of machining force, cutting energy and cutting pressure in turning using three regression based machine learning techniques (polynomial regression, support vector machine and Gaussian process regression) as well as artificial neural networks. Contrary to other presented works, here the estimation of selected machining responses was carried out for different cooling/lubricating conditions. In particular, the study covered the MQL and HPC machining conditions. Selected machine learning techniques are moreover used for comparative assessment of machining responses in order to determine the best approach according to model accuracy and capability. In addition, multi-objective optimization problem was also carried out. 2. Experimental details Straight turning of AISI 1045 (C45E) steel supplied as bars 120 mm in diameter and 300 mm long in a lathe Boehringer that develops a spindle power of 8 kW have been carried out by standard carbide inserts SNMG 1204 08 NMX. In this study, focus is placed on the application of various cooling/lubricating techniques in machining. Therefore, the experiments are conducted under different machining environments, namely MQL and HPC. The MQL and HPC systems were attached in the experimental setup during the machining trials. For MQL assisted turning, cutting fluid was supplied to spray gun at the rate of 30 ml/h, which is mixed with compressed air (3 bar) in the mixing chamber of spray gun. Then the mixture of air and cutting fluid is supplied at the cutting zone by spray gun nozzle located 30 mm away from tool tip, at an angles of 90 and 30, from the cutting edge and clearance face, respectively. During HPC assisted turning, the cutting fluid was supplied at a constant pressure of 50 bar and flow rate of 2 l/min through 0.4 mm nozzle (diameter) normal to the cutting edge at a low angle (about 5–6) with the cutting tool rake face. The nozzle was positioned 30 mm away from the tool tip with purpose to achieve fairly close to the tool-chip contact zone as well as to reduce the interference of the nozzle with the flowing chips. Apart from different machining environments, three cutting parameters, that are cutting speed (v), depth of cut (a) and feed rate (f), were also selected as control factors. Referring to Table 1, the three levels of cutting speed, three levels of depth of cut and four levels of feed rate generate 36 number of experimental runs for each of machining environment. The ranges of these parameters were selected based on the recommendations of the cutting tool manufacturer and in accordance with previous studies. Moreover, the parameter ranges were also extended in order to achieve higher productivity and to investigate machining responses in different machining environments. The three components of the cutting force, namely, main cutting force (Fc), feed force (Ff) and passive force (Fp), were measured using the Kistler dynamometer type 9259A. The measurement chain further includes a charge amplifier (Kistler 5001), spectrum analyzer (HP3567A) and personal computer for data acquisition and analysis. The machining force (FR), cutting power (Pc) and cutting pressure (Ks) are computed from the following equations: The obtained experimental data were divided into two data - sets, namely training data set for model development (75% of the entire data set) and test data set for model validation (25% of the entire data set). Thus, 27 sets of randomly selected experimental trials were used for model construction, leaving the remaining 9 sets of data to test model performance. Identical data partition scheme was utilized for MQL and HPC machining conditions.