田?穎,王文豪,楊利明,邵文婷
基于刀具磨損狀態(tài)識別的加工參數(shù)多目標(biāo)優(yōu)化
田?穎,王文豪,楊利明,邵文婷
(天津大學(xué)機(jī)械工程學(xué)院,天津 300072)
在數(shù)控銑削過程中,刀具磨損對機(jī)床主軸能耗影響很大,同時(shí)與刀具加工能力直接相關(guān),需適時(shí)調(diào)整加工參數(shù)以適應(yīng)不同磨損狀態(tài),保證多目標(biāo)綜合最優(yōu).針對此問題,基于刀具磨損狀態(tài)識別,根據(jù)不同磨損時(shí)期給出相應(yīng)的加工參數(shù)優(yōu)化策略.首先,以刀具壽命周期內(nèi)的主軸功率為基礎(chǔ),采用粒子群優(yōu)化算法(particle swarm optimization,PSO)建立考慮刀具磨損的能耗模型,平均誤差低于5%,并基于與刀具磨損的強(qiáng)相關(guān)性,以主軸功率作為單一指標(biāo)識別刀具磨損狀態(tài).進(jìn)一步,為獲取影響加工成本的動(dòng)態(tài)性指標(biāo)——刀具剩余使用壽命(remaining useful life,RUL)及相應(yīng)的主軸功率,利用人工神經(jīng)網(wǎng)絡(luò)(artificial neural network,ANN)建立刀具磨損退化模型,通過描述主軸功率隨時(shí)間的變化來隱含退化過程,該模型擬合度達(dá)0.992.最終,綜合考慮能耗、刀具及時(shí)間成本,設(shè)計(jì)多目標(biāo)優(yōu)化函數(shù),并通過遺傳算法(genetic algorithm,GA)搜索函數(shù)最小值,給出不同磨損時(shí)期的最優(yōu)加工參數(shù)及優(yōu)化策略.此加工參數(shù)多目標(biāo)優(yōu)化融入了主軸功率與刀具RUL等動(dòng)態(tài)指標(biāo),可依據(jù)主軸功率在線識別刀具磨損狀態(tài),進(jìn)而調(diào)整加工參數(shù)以適應(yīng)不同磨損時(shí)期,降低加工成本.結(jié)果表明:采用此方法在線調(diào)整加工參數(shù),可平均降低24.258%的綜合成本,具備有效性與實(shí)用性.
加工參數(shù);多目標(biāo)優(yōu)化;刀具磨損;主軸功率;刀具剩余使用壽命
近年來,綠色制造成為熱點(diǎn)研究方向.世界能源供需不平衡現(xiàn)象不斷加劇[1].在多數(shù)工業(yè)國家中,制造業(yè)能耗約占全國的30%~50%[2].生產(chǎn)制造分多個(gè)層次:設(shè)備單元、多設(shè)備系統(tǒng)、工廠單元、多工廠系統(tǒng)以及全球生產(chǎn)鏈[3].設(shè)備單元是基礎(chǔ),尤其是作為重要制造設(shè)備的數(shù)控機(jī)床,其節(jié)能潛力巨大.因此,通過優(yōu)化加工參數(shù)促進(jìn)數(shù)控機(jī)床的節(jié)能減排,對于推進(jìn)制造業(yè)的綠色發(fā)展具有重要意義.
在數(shù)控機(jī)床層面,針對加工參數(shù)優(yōu)化的研究取得很多進(jìn)展.李愛平等[4]針對粗/精加工過程,以低能耗為優(yōu)化目標(biāo),在多個(gè)邊界約束條件下,進(jìn)行加工參數(shù)的局部優(yōu)化與全局優(yōu)化.Sardinas等[5]以相互沖突的指標(biāo)——材料去除率與代表切削損傷的分層因子為優(yōu)化目標(biāo),實(shí)現(xiàn)雙目標(biāo)參數(shù)優(yōu)化.Solimanpur等[6]考慮多個(gè)指標(biāo)包括生產(chǎn)成本、生產(chǎn)時(shí)間和表面粗糙度.Yang等[7]則以加工時(shí)間、成本和利潤率作為優(yōu)化目標(biāo),在刀具耐用度、表面粗糙度與切削力等約束條件下,獲取最優(yōu)加工參數(shù).Yan等[8]權(quán)衡銑削過程中的加工質(zhì)量、生產(chǎn)效率以及加工可持續(xù)性.文獻(xiàn)[9-10]則以加工時(shí)間與比能為優(yōu)化目標(biāo),考慮更全面的約束:功率上限、表面質(zhì)量、刀具耐用度與切削梨耕能量約束等,面向能量優(yōu)化加工參數(shù).
隨著研究的深入,機(jī)床能耗還會受刀具狀態(tài)的影響,能耗建模逐漸考慮刀具磨損因素[11-12].張翔[13]以影響因子的形式,將刀具磨損退化規(guī)律引入能耗模型,綜合考慮能耗和時(shí)間成本,實(shí)現(xiàn)加工參數(shù)的雙目標(biāo)優(yōu)化.Tian等[14]以減少生產(chǎn)碳排放、加工時(shí)間與能耗為目的,考慮刀具磨損因素,建立量化關(guān)系模型.Bhushan[15]則以加工參數(shù)與刀尖半徑為自變量,建立能耗模型與刀具壽命模型,最大限度地減少功耗的同時(shí),延長刀具壽命,獲取加工參數(shù)的最佳水平.
現(xiàn)存的加工參數(shù)優(yōu)化方法中,盡管將刀具磨損引入能耗模型,但在制定優(yōu)化目標(biāo)時(shí),未考慮刀具磨損的動(dòng)態(tài)變化對加工成本的影響,沒有融入動(dòng)態(tài)性指標(biāo),導(dǎo)致參數(shù)優(yōu)化與調(diào)整缺乏動(dòng)態(tài)性,難以適應(yīng)不同的刀具磨損狀態(tài).在加工過程中,刀具磨損狀態(tài)與機(jī)床主軸能耗動(dòng)態(tài)相關(guān),其直接影響加工成本,將該因素融入到加工參數(shù)優(yōu)化過程中,有助于實(shí)現(xiàn)數(shù)控加工的精益能效與健康管理,對于調(diào)整加工參數(shù)與指導(dǎo)生產(chǎn)具有重要意義.因此,本文不僅在能耗模型中引入刀具磨損,同時(shí)以剩余使用壽命(remaining useful life,RUL)作為刀具動(dòng)態(tài)健康指標(biāo),融入到機(jī)床能耗、刀具以及時(shí)間成本指標(biāo)中,實(shí)現(xiàn)加工參數(shù)的多目標(biāo)優(yōu)化,并基于不同磨損時(shí)期制定優(yōu)化策略,用以在線調(diào)整加工參數(shù),使得多目標(biāo)綜合最優(yōu).
圖1?本文框架
機(jī)床能耗有4大模塊:基礎(chǔ)系統(tǒng)、輔助系統(tǒng)、進(jìn)給系統(tǒng)和主傳動(dòng)系統(tǒng)[16],如圖2所示.前3個(gè)模塊的能耗不受刀具磨損的影響,而主傳動(dòng)系統(tǒng)能耗與刀具狀態(tài)直接相關(guān),其主要指主軸電機(jī)能耗.
圖2?機(jī)床能耗組成
完整加工過程有3個(gè)階段:啟動(dòng)、空載與切削階段[17].在重點(diǎn)研究的切削階段,主軸功率分為空載與切削功率.空載功率僅與主軸轉(zhuǎn)速有關(guān),而切削功率則受加工參數(shù)與刀具磨損的共同影響.
相對其他傳感器信號,機(jī)床主軸功率具有獲取便捷、易于后處理、能適應(yīng)多種工況、建模成本低等多種優(yōu)勢,其既與刀具健康狀態(tài)相關(guān),也是重要能效指標(biāo),可用于優(yōu)化加工參數(shù).因此,對主軸功率進(jìn)行建模是非常重要的.
在加工參數(shù)優(yōu)化方面,能耗建模是核心工作之一.能耗建模不僅要考慮加工參數(shù),還有其他因素如工件材料硬度、刀具磨損以及溫度等.刀具磨損直接影響加工能耗與表面質(zhì)量,嚴(yán)重時(shí)會導(dǎo)致機(jī)床停機(jī),影響生產(chǎn)效率.
目前,考慮刀具磨損的能耗模型多數(shù)針對切削功率,主要包括基本指數(shù)[14]、二次多項(xiàng)式組合[18]、分段函數(shù)[19]及其他形式.第一種形式最常用,該模型系數(shù)較少、建模簡單,可降低實(shí)驗(yàn)與時(shí)間成本.
本文采用PSO算法優(yōu)化模型系數(shù)以建立能耗模型.PSO源于鳥類的捕食行為[20].該算法適合在動(dòng)態(tài)、多目標(biāo)優(yōu)化環(huán)境中尋優(yōu),具有優(yōu)秀的計(jì)算速度與全局搜索能力[21].PSO通過粒子群的迭代更新,搜索誤差函數(shù)最小值,篩選出全局最優(yōu)粒子,優(yōu)化模型系數(shù).標(biāo)準(zhǔn)PSO的粒子更新過程[21-22]為
以能耗模型建立主軸功率與刀具磨損的映射,獲取不同磨損下的主軸功率.本文考慮到功率信號的獲取便捷、易于分析以及便于實(shí)現(xiàn)健康與能效的綜合評價(jià)等優(yōu)勢,以主軸功率作為單一指標(biāo)識別刀具磨損狀態(tài).相關(guān)性分析表明主軸功率與刀具磨損具有強(qiáng)相關(guān)性,并能適應(yīng)多種工況,結(jié)果詳見實(shí)驗(yàn)部分,其證明了該磨損狀態(tài)識別方法的可行性.故本文利用主軸功率在線識別刀具磨損狀態(tài).
本文通過描述主軸功率隨時(shí)間的變化規(guī)律,來隱含刀具磨損退化的規(guī)律,建立刀具磨損退化模型.如圖3所示,在建立能耗模型的基礎(chǔ)上,結(jié)合刀具磨損退化模型,獲取刀具剩余使用壽命指標(biāo),即
式中與分別為退化模型中與對應(yīng)的加工時(shí)間,與分別為磨損閾值與當(dāng)前狀態(tài)對應(yīng)的主軸功率.
在單工況下,功率隨時(shí)間變化一般符合二次、三次或指數(shù)函數(shù)的規(guī)律.而在多工況下,難以建立統(tǒng)一形式的函數(shù)模型.近年來,隨著ANN算法的迅速發(fā)展,建立適應(yīng)多種工況的退化模型成為可能.
圖4?ANN結(jié)構(gòu)
本文采用單隱藏層的ANN結(jié)構(gòu)建立退化模型.主軸功率信號是監(jiān)測對象,其與加工參數(shù)共同作為網(wǎng)絡(luò)輸入,加工時(shí)間則作為網(wǎng)絡(luò)輸出.退化模型當(dāng)中的不僅是時(shí)間變量,其還代表著當(dāng)前刀具磨損狀態(tài)在全壽命周期中的所處位置,尤其是后者,對于刀具RUL非常重要.
在多種工況下,通過能耗模型獲取主軸功率與刀具磨損的映射,用于識別刀具磨損狀態(tài);通過刀具磨損退化模型,獲取影響加工成本的動(dòng)態(tài)指標(biāo).本文分別從能耗成本、刀具成本及時(shí)間成本考慮,設(shè)計(jì)相應(yīng)的成本指標(biāo),即優(yōu)化目標(biāo).
此后,以主軸功率作為單一指標(biāo)在線識別刀具磨損狀態(tài),判斷刀具磨損時(shí)期,在線調(diào)整加工參數(shù)為相應(yīng)的最優(yōu)水平,使得多目標(biāo)綜合最優(yōu).
圖5?GA運(yùn)算流程
本文使用加州大學(xué)伯克利分校BEST實(shí)驗(yàn)室銑削數(shù)據(jù)集[24].先建立能耗模型,以主軸功率作為識別刀具磨損狀態(tài)的單一指標(biāo).其次,利用ANN建立刀具磨損退化模型,并獲取動(dòng)態(tài)健康指標(biāo)RUL.最后,進(jìn)行加工參數(shù)多目標(biāo)優(yōu)化,并根據(jù)結(jié)果做出評價(jià).
在銑削過程中,采集主軸電機(jī)電流信號,并有間隔地測量刀具側(cè)面磨損量.加工中心為MC-510V,刀具為70mm的6面銑刀,安裝KC710型刀片,工件尺寸為483mm×178mm×51mm.選取其中部分實(shí)驗(yàn),變量為切削深度與進(jìn)給速度,如表1所示.實(shí)驗(yàn)其他信息可參考該數(shù)據(jù)集[24].
表1?實(shí)驗(yàn)參數(shù)
Tab.1?Details of experimental parameters
設(shè)定參數(shù):切削寬度5mm,電壓380V,功率因數(shù)0.8.截取原始信號中間一半數(shù)據(jù)為有效數(shù)據(jù),采用集合經(jīng)驗(yàn)?zāi)B(tài)分解(ensemble empirical mode decomposition,EEMD)降噪,并提取交流電流的均方根特征設(shè)為電流幅值,進(jìn)而求取主軸功率,即
本文為了方便分析,從主軸電機(jī)電流中提取主軸功率.但在實(shí)際應(yīng)用中,主軸功率可由機(jī)床內(nèi)部接口或便捷式功率儀來采集,信號獲取簡單.此外,相較于電流信號,功率信號作為重要的能耗指標(biāo),其在實(shí)際生產(chǎn)當(dāng)中應(yīng)用更廣泛.
將偏離主趨勢的數(shù)據(jù)替換為前后數(shù)據(jù)的平均值,進(jìn)行單調(diào)性修正,并關(guān)注刀具磨損量小于0.6mm的數(shù)據(jù).實(shí)驗(yàn)涉及切削深度與進(jìn)給速度兩個(gè)變量,其他參數(shù)為常量,歸納到常系數(shù)中.為防止實(shí)驗(yàn)數(shù)據(jù)的前后順序影響建模,將數(shù)據(jù)隨機(jī)打散.通過PSO優(yōu)化模型參數(shù),得到以下能耗模型,即
該模型的平均誤差為4.442%,低于5%.如圖6所示,實(shí)驗(yàn)數(shù)據(jù)均勻分布在理想輸出線附近,驗(yàn)證了模型的有效性.
本文進(jìn)行主軸功率與刀具磨損的相關(guān)性分析,計(jì)算兩者之間的皮爾遜相關(guān)系數(shù).分析結(jié)果如圖7所示,在多種工況下,主軸功率與刀具磨損的相關(guān)系數(shù)均大于0.9,由此證明以主軸功率作為單一指標(biāo)識別刀具磨損狀態(tài)的可行性.因此,本文結(jié)合能耗模型獲取不同磨損下的主軸功率,進(jìn)而通過監(jiān)測主軸功率來
圖7?相關(guān)性分析結(jié)果
識別磨損狀態(tài).以第一組實(shí)驗(yàn)為例,如圖8所示,利用主軸功率可有效識別刀具磨損狀態(tài).
圖8?刀具磨損狀態(tài)監(jiān)測結(jié)果
圖9?ANN訓(xùn)練過程
圖10?ANN建模結(jié)果
初始磨損設(shè)為0.05mm,閾值設(shè)為0.4mm,通過能耗模型獲取主軸功率初始值與閾值,并帶入ANN模型,結(jié)合式(6)計(jì)算刀具RUL.如圖11所示,刀具RUL的預(yù)測值與實(shí)驗(yàn)值具有很高的一致性,由此表明通過退化模型,可準(zhǔn)確預(yù)測刀具RUL,便于將其融入成本指標(biāo)中.
圖11?RUL預(yù)測結(jié)果
將刀具整個(gè)壽命周期分為4個(gè)磨損時(shí)期,并給出切削深度與進(jìn)給速度的限制范圍,如表2所示.
表2?刀具壽命周期中的4個(gè)時(shí)期
Tab.2?List of four periods in the tool life cycle
每個(gè)指標(biāo)在優(yōu)化結(jié)果中的占比情況如圖13所示,能耗與時(shí)間成本占主要比重,刀具成本占比最小,時(shí)間成本占比最大.由此說明,在多目標(biāo)優(yōu)化過程,刀具成本指標(biāo)成為主要優(yōu)化目標(biāo),其次是能效成本,最后是時(shí)間成本.對于多目標(biāo)優(yōu)化權(quán)重的配比,可以根據(jù)決策者的需求,在優(yōu)化目標(biāo)函數(shù)中,在對應(yīng)指標(biāo)前面配置相應(yīng)的權(quán)值系數(shù).
圖12?適應(yīng)度進(jìn)化曲線
圖13?每個(gè)成本指標(biāo)在優(yōu)化結(jié)果中的占比
表3?參數(shù)優(yōu)化結(jié)果
Tab.3?Results of optimization parameters
本文建立了考慮刀具磨損的能耗模型,并基于與刀具磨損的強(qiáng)相關(guān)性,以主軸功率作為單一指標(biāo)識別刀具磨損狀態(tài).其次,利用ANN建立刀具磨損退化模型,通過描述主軸功率隨時(shí)間的變化規(guī)律來隱含磨損退化過程,用以獲取刀具RUL及其相應(yīng)的主軸功率等動(dòng)態(tài)性指標(biāo).進(jìn)一步,綜合考慮能耗、刀具及時(shí)間成本,設(shè)計(jì)多目標(biāo)優(yōu)化函數(shù),將動(dòng)態(tài)指標(biāo)融入其中,并采用GA給出不同磨損時(shí)期的最優(yōu)加工參數(shù)及優(yōu)化策略.以主軸功率在線識別刀具磨損狀態(tài),結(jié)合不同磨損時(shí)期將加工參數(shù)調(diào)整為最優(yōu)水平,綜合成本總體降低24.258%.
此加工參數(shù)優(yōu)化方法考慮了刀具磨損對加工成本的動(dòng)態(tài)影響,將影響加工成本的動(dòng)態(tài)指標(biāo)融入優(yōu)化目標(biāo),便于制定動(dòng)態(tài)優(yōu)化策略.采用此方法可在線調(diào)整加工參數(shù),以適應(yīng)不同磨損狀態(tài),保證多目標(biāo)最優(yōu),降低綜合成本,具備有效性與實(shí)用性.
本文采用了ANN算法,但由于沒有大量實(shí)驗(yàn)數(shù)據(jù),會導(dǎo)致模型泛化能力不足,從而影響優(yōu)化結(jié)果及評價(jià)的準(zhǔn)確性.在加工參數(shù)優(yōu)化過程中,考慮刀具健康的動(dòng)態(tài)影響,對于機(jī)床能效的精益優(yōu)化與管理具有很大的研究意義.
[1] Yoon H S,Kim E S,Kim M S,et al. Towards greener machine tools—A review on energy saving strategies and technologies[J]. Renewable and Sustainable Energy Reviews,2015,48:870-891.
[2] Park C W,Kwon K S,Kim W B,et al. Energy consumption reduction technology in manufacturing—A selective review of policies,standards,and research[J]. International Journal of Precision Engineering and Manufacturing,2009,10(5):151-173.
[3] Duflou J R,Sutherland J W,Dornfeld D,et al. Towards energy and resource efficient manufacturing:A processes and systems approach[J]. CIRP Annals-Manufacturing Technology,2012,61(2):587-609.
[4] 李愛平,鮑?進(jìn),李?聰,等. 基于低能耗的平面端銑削粗/精加工參數(shù)全局多目標(biāo)優(yōu)化[J]. 中國機(jī)械工程,2015,26(14):1888-1894.
Li Aiping,Bao Jin,Li Cong,et al. Global optimization with multi-targets for rough and finish end-milling parameters based on minimum energy performance[J]. China Mechanical Engineering,2015,26(14):1888-1894(in Chinese).
[5] Sardinas R Q,Reis P,Davim J P. Multi-objective optimization of cutting parameters for drilling laminate composite materials by using genetic algorithms[J]. Composites Science and Technology,2006,66(15):3083-3088.
[6] Solimanpur M,Ranjdoostfard F. Optimisation of cutting parameters using a multi-objective genetic algorithm[J]. International Journal of Production Research,2009,47(21):6019-6036.
[7] Yang W,Guo Y,Liao W. Multi-objective optimization of multi-pass face milling using particle swarm intelligence[J]. The International Journal of Advanced Manufacturing Technology,2011,56(5/6/7/8):429-443.
[8] Yan J,Li L. Multi-objective optimization of milling parameters—The trade-offs between energy,production rate and cutting quality[J]. Journal of Cleaner Production,2013,52(8):462-471.
[9] 周麗蓉. 數(shù)控機(jī)床能耗建模與面向能量的加工參數(shù)優(yōu)化[D]. 濟(jì)南:山東大學(xué)機(jī)械工程學(xué)院,2018.
Zhou Lirong. Research on Modeling Energy Consumption of CNC Machine Tools and Energy Oriented Machining Oarameters Optimization[D]. Jinan:School of Mechanical Engineering,Shandong University,2018 (in Chinese).
[10] 王太勇,孫熙冉,田松齡,等. 可重構(gòu)機(jī)床多目標(biāo)優(yōu)選方法[J]. 天津大學(xué)學(xué)報(bào)(自然科學(xué)與工程技術(shù)版),2021,54(9):881-889.
Wang Taiyong,Sun Xiran,Tian Songling,et al. Multi-objective optimization selection method for the reconfigurable machine tool[J]. Journal of Tianjin University(Science and Technology),2021,54(9):881-889(in Chinese).
[11] Wang Q,Zhang D,Tang K,et al. A mechanics based prediction model for tool wear and power consumption in drilling operations and its applications[J]. Journal of Cleaner Production,2019,234:171-184.
[12] Zhao G Y,Liu Z Y,He Y,et al. Energy consumption in machining:Classification,prediction,and reduction strategy[J]. Energy,2017,133(8):142-157.
[13] 張?翔. 考慮刀具磨損的機(jī)床能耗建模及優(yōu)化方法研究[D]. 哈爾濱:哈爾濱工業(yè)大學(xué)機(jī)電工程學(xué)院,2016.
Zhang Xiang. Research on Energy Consumption Modeling and Optimization of Machine Tools Considering Tool Wear[D]. Harbin:School of Mechatronics Engineer-ing,Harbin Institute of Technology,2016(in Chinese).
[14] Tian C,Zhou G,Zhang J,et al. Optimization of cutting parameters considering tool wear conditions in low-carbon manufacturing environment[J]. Journal of Cleaner Production,2019,226:706-719.
[15] Bhushan R K. Optimization of cutting parameters for minimizing power consumption and maximizing tool life during machining of Al alloy SiC particle composites[J]. Journal of Cleaner Production,2013,39:242-254.
[16] Deng Z,Zhang H,F(xiàn)u Y,et al. Optimization of process parameters for minimum energy consumption based on cutting specific energy consumption[J]. Journal of Cleaner Production,2017,166(11):1407-1414.
[17] Liu F,Xie J,Liu S. A method for predicting the energy consumption of the main driving system of a machine tool in a machining process[J]. Journal of Cleaner Production,2015,105(10):171-177.
[18] Wan T,Chen X,Li C,et al. An on-line tool wear monitoring method based on cutting power[C]//2018 IEEE 14th International Conference on Automation Science and Engineering(CASE). Munich,Germany,2018:205-210.
[19] Shi K N,Zhang D H,Liu N,et al. A novel energy consumption model for milling process considering tool wear progression[J]. Journal of Cleaner Production,2018,184(5):152-159.
[20] Noushabadi A S,Dashti A,Raji M,et al. Estimation of cetane numbers of biodiesel and diesel oils using regression and PSO-ANFIS models[J]. Renewable Energy,2020,158:465-473.
[21] 包子陽,余繼周. 智能優(yōu)化算法及其MATLAB實(shí)例[M]. 北京:電子工業(yè)出版社,2016.
Bao Ziyang,Yu Jizhou. Intelligent Optimization Algorithm and MATLAB Examples[M]. Beijing:Publishing House of Electronics Industry,2016(in Chinese).
[22] Shi Y. A modified particle swarm optimizer[C]// IEEE World Congress on Computational Intelligence. Anchorage,AK,USA,1998:69-73.
[23] Zhou Y,Xue W. Review of tool condition monitoring methods in milling processes [J]. The International Journal of Advanced Manufacturing Technology,2018,96:2509-2523.
[24] Agogino A,Goebel K. Milling data set[EB/OL]. https://ti.arc.nasa.gov/tech/dash/groups/pcoe/prognostic-data-repository/,2007-07-14.
Multi-Objective Optimization of Machining Parameters Based on Tool Wear Condition
Tian Ying,Wang Wenhao,Yang Liming,Shao Wenting
(School of Mechanical Engineering,Tianjin University,Tianjin 300072,China)
Computer numerical control(CNC)milling is a machinery process,in which tool wear has a significant impact on the energy consumption of the machine tool spindle motor,which is directly related to the capacity of tool. Thus,it is necessary to adjust machining parameters in time,adapt to different wear conditions,and ensure a detailed multi-objective optimization approach. In response to this issue,based on tool wear conditions,optimization strategies of machining parameters have been used to adjust the machining parameters online in different wear periods.At first,energy consumption model considering tool wear was developed by the particle swarm optimization(PSO),which has a mean error of less than 5%. Based on the strong correlation with tool wear,machine tools’ spindle power was used as a single indicator to evaluate tool wear conditions. Furthermore,to obtain dynamic indicators that affect costs,such as tool remaining useful life(RUL) and corresponding spindle power,the tool wear degradation was modeled using an artificial neural network(ANN),describing spindle power changes over time to indicate tool degradation process. The degree of model fit reaches 0.992 in this process. Finally,energy consumption,tool,and time cost have been weighed to design a multi-objective optimization function. Besides,the genetic algorithm(GA)was used to search optimal machining parameters,and obtain optimization strategies for different wear periods. Dynamic indicators,such as spindle power and tool RUL,have been incorporated into this multi-objective optimization of machining parameters. Tool wear conditions were identified online based on spindle power to determine wear periods,and then machining parameters were adjusted to the corresponding optimal level. The results show that this method incorporates dynamic indicators toadapt to tool wear conditions,and it decreases the total costs by an average of 24.258%,showing that the method is effective and practical.
machining parameters;multi-objective optimization;tool wear;spindle power;tool remaining useful life
10.11784/tdxbz202007073
TG659
A
0493-2137(2022)02-0166-08
2020-07-30;
2020-11-15.
田?穎(1977—??),女,博士,副教授.Email:m_bigm@tju.edu.cn
田?穎,tianying@tju.edu.cn.
國家自然科學(xué)基金資助項(xiàng)目(51975407).
the National Natural Science Foundation of China(No. 51975407).
(責(zé)任編輯:王曉燕)
天津大學(xué)學(xué)報(bào)(自然科學(xué)與工程技術(shù)版)2022年2期