付茂文 沈丹峰 趙剛 尚國飛 柏順偉
摘 要:為更好地控制經紗張力,提高系統動態響應性能減小抖振,開發了一種神經反步分數階快速終端滑模控制器(RBF-BCFOFTSMC),通過動力學分析建立了織機送經系統的時變數學模型。同時,推導了一種新的反步分數階快速終端滑模控制方法。針對織機織造過程中系統總干擾上界的未知性和系統時變性的特點,設計了自適應律來估計外部干擾的上界值,設計神經網絡參數自適應律來逼近真實的系統狀態,并利用李雅普諾夫穩定性證明系統的合理性。通過其與傳統滑模控制(SMC)和神經PID控制(RBF-PID)在仿真實驗和實際工況下的對比,結果表明:RBF-BCFOFTSMC在經紗張力控制方面不僅減小了抖振,并且具有較高的魯棒性和響應性能。
關鍵詞:經紗張力;分數階;反步;滑模控制;神經網絡
中圖分類號:TS103;TP183;TP273
文獻標志碼:A
文章編號:1009-265X(2023)04-0130-09
收稿日期:2022-12-15
網絡出版日期:2023-03-21
基金項目:國家自然科學基金項目(51805402)
作者簡介:付茂文(1996—),男,山東泰安人,碩士研究生,主要從事送經系統張力控制方面的研究。
通信作者:沈丹峰, E-mail: dfshen@xpu.edu.cn
經紗張力的穩定性對于織機生產不同花紋和提高織造效率具有舉足輕重的作用,無論是張力過大或者過小都會降低織造質量[1],嚴重時還有可能導致紗線崩裂從而停車的現象,需要人工干預后才能再重新織造,加大了人力成本和降低了織造效率。由于送經系統存在電機振動和綜框等運動,想要保持較高的張力穩定性,完成更高質量和不同的織造要求,設計一種魯棒性較高和響應性能較快的控制器迫在眉睫。
為了提高送經系統的張力穩定性,許多智能控制算法被提出用于實際控制系統中,如自適應PID控制[2-3]、反步控制[4-5]、滑模控制[6-7]和神經網絡控制[8-9]等,以提高系統控制性能。崔征山等[10]設計擴張狀態觀測器來對系統擾動進行在線估計,并將估計到的擾動補償到滑模控制器中,很好地應對了系統中運動產生的強擾動,但擴張狀態觀測器的引入增加了控制器的調參難度。黃道敏等[11]將分數階理論融合到積分滑模控制中,設計指數趨近律,并且為估計外部擾動添加擾動估計項,該控制策略具有較快的收斂速度,對于非線性的系統魯棒性較強。鄧檳檳等[12]設計了一種新的快速終端滑模控制方法,經過了有限時間穩定性證明,誤差可在短時間內快速收斂,提高了控制系統的跟蹤精度。梁相龍等[13]將神經網絡和指令濾波融合到滑模控制算法中,指令濾波器用來信號的估計和噪聲處理,通過梯度下降算法來自適應更新網絡權值系數,對于系統的不確定性和外部干擾具有很強的魯棒性。還有一部分學者也對反步控制進行了研究,Razmi等[14]設計了一種針對參數不確定性和外部干擾的控制策略,采用神經網絡自適應更新滑模面的系數,增加了系統的瞬態和穩態性能。Chen等[15]為了提高反步控制的收斂速度和跟蹤性能,設計了具有更多自由度的分數階反推控制器,并采用模糊神經網絡估計系統的不確定性,采用指數調節機制補償估計誤差。熊蕊[16]設計一種改進神經網絡反步控制策略,利用神經網絡逼近外部未知狀態,利用自適應律更新神經網絡的參數,實現了系統的高精度控制。Fu等[17]提出了一種自適應神經反步動態表面控制算法,采用動態表面來優化反步控制算法,采用神經網絡來逼近系統的動力學模型,最終通過實驗證明了控制方法的有效性。通過上述研究和分析可知,送經系統是強非線性系統,采用高效非線性控制算法有利于提高張力穩定性,滑模控制因其自身存在的抖振和奇異問題,在非線性系統的應用中往往需要改進或者與其他算法融合,將分數階理論融合到滑模控制中并證明有限時間穩定性,可實現滑模控制的性能提升,但是在滑模控制中存在的一些狀態變量是不容易測量的,因此將反步控制算法應用到滑模控制中簡化控制量。送經系統在運行過程中是時變的,根據傳統的數學建模方法得到的模型信息較難反應織機真實的狀態,考慮到上述研究采用的神經網絡估計外部未知狀態得到了很好的效果,故本文將神經網絡引入到反步快速終端滑模控制中來估計未知的建模信息,進一步提高算法對系統的控制性能。
1 張力數學模型
織機送經系統的織軸結構簡化原理圖[18]如圖1所示。圖1中,T(t)為經紗動態張力,r0為經軸初始半徑,r1(t)為經軸實時半徑,M1(t)為經軸電機電磁轉矩,v1(t)為機上經紗線速度,a(t)為機上經紗加速度。伴隨著織機織造過程不斷進行,經軸和卷軸半徑不斷變化,又因經紗柔性特點,因此送經系統具有強耦合性和強非線性。
3.2 實驗
為驗證上述控制方法在實際工況中的有效性,將3種控制方法應用到搭建的實驗平臺中,分別在張力跟蹤精度上和控制器輸入上進行對比。如圖4所示為STM32和FPGA聯合控制的實驗平臺,兩者之間通過SPI通信傳輸數據,由FPGA采集信號并傳遞給STM32完成算法運算,其中實驗平臺模擬了織軸卷徑的變化、電機振動和綜框運動,綜框運動的模擬由軸承外徑周圍凸起的轉動來實現。實驗分別在設定張力為1.56 N和2.56 N下進行,如圖5為3種控制器的張力跟蹤效果和控制器輸出情況,表2為3種控制器的實驗性能指標。
在16 s的運行過程中織軸卷徑變化了1 mm,在此期間送經電機輸出合適的轉速保持張力穩定。由圖5分析可知,由于綜框模擬運動和電機振動等原因,3種控制器都上下波動穩定到某一狀態,其中RBF-BCFOFTSMC控制器MAXE最小,跟蹤精度優于其他兩種控制器在最靠近張力設定值附近波動。3種控制器的輸出轉速都隨織軸卷徑變化和干擾等不斷波動,在設定張力1.56 N和2.56 N下的波動幅度分別為4.41、5.93、5.48和4.42、5.95、5.50,
其中RBF-BCFOFTSMC波動幅度最小,穩定性更好。滑模控制的自適應切換部分和神經網絡能夠實時估計外部干擾和未建模動力學,使得RBF-BCFOFTSMC控制器隨著織軸卷徑變化實時調整,輸出高精度的控制律保持張力穩定,總結可知RBF-BCFOFTSMC跟蹤性能和抗干擾性能較好,具有較高的魯棒性。
通過聯合FPGA和ARM開發出的送經系統張力控制實驗平臺,不僅控制簡單、成本較低,可對于新型控制算法進行穩定性驗證,擺脫了測試過程需在真實織機中運行的依賴,極大地減少了經紗張力的檢測門檻和技術難度,為控制算法應用到實際工況中的紗線張力檢測和調節提供了新的思路。該實驗平臺通過送經與卷取電機的配合來完成經紗送出,在此基礎上對提出的控制算法進行有效性驗證,但是該實驗平臺與真實織機還存在一定差異,對于其中的一些其他運動也只是采用模擬的方式,后續研究還有必要在該實驗基礎上追加實驗,證明其他運動對送經系統張力的影響。
4 結 論
織機織造過程中,張力過小容易導致出現粗紗或冒紗,降低織物平整度,而當張力過大時,又會導致紗線斷裂停車,從而降低織造效率,由于送經系統時變性和各種干擾的存在,常規控制算法存在超調嚴重和穩態精度低等問題,因此為改善織機送經系統的張力穩定性,設計了一種神經反步分數階快速終端滑模控制算法。首先建立了織機送經系統的數學模型,采用自適應律實時更新控制器參數達到對外部干擾估計的效果,將RBF神經網絡介入滑模控制中,逼近送經系統的真實系統狀態,以此得到更為精確的數學模型。采用反步控制和滑模控制相結合的方法避免了更多系統變量的使用,簡化了滑模控制的控制律,引入分數階理論給控制器帶來更多的自由度,通過李雅普諾夫函數驗證了控制器的有限時間收斂性和穩定性。最終通過仿真和實驗證明RBF-BCFOFTSMC控制器具有較高的張力穩定性,提高了系統的控制精度。所設計的控制器改善了織機的送經系統控制水平,對于經軸上的紗線退繞下來進入綜框運動時的張力精度具有提高作用,有利于減少紗線斷頭現象,增強送經量的恒定水平,對于提高織機的生產效率和胚布質量具有較高的意義。
參考文獻:
[1]馬宏帥,趙世海.基于線性自抗擾控制的放卷張力控制系統[J].現代紡織技術,2019,27(1):87-92.
MA Hongshuai, ZHAO Shihai. Unwinding tension control system based on linear auto disturbance rejection control[J]. Advanced Textile Technology, 2019, 27(1): 87-92.
[2]KARAHAN O. Design of optimal fractional order fuzzy PID controller based on cuckoo search algorithm for core power control in molten salt reactors[J]. Progress in Nuclear Energy, 2021, 139: 103868.
[3]姜磊.智能梳棉機自調勻整控制系統設計開發[J].現代紡織技術,2020,28(3):89-96.
JIANG Lei. Design and development of autoleveling control system for intelligent carding machine[J]. Advanced Textile Technology, 2020, 28(3): 89-96.
[4]FANG Y M, FEI J T, YANG Y Z. Adaptive backstepping design of a microgyroscope[J]. Micromachines, 2018, 9(7): 338.
[5]徐子琴,雷明.風擾動下固定翼無人機指令濾波反步著陸控制[J].計算機仿真,2022,39(9):55-62.
XU Ziqin, LEI Ming. Command filtered backstepping landing control of fixed-wing unmanned aerial vehicle considering wind disturbance[J]. Computer Simulation, 2022, 39(9): 55-62.
[6]ZAIHIDEEM F, MEKHILEF S, MUBIN M. Robust speed control of PMSM using sliding mode control (SMC): Areview[J]. Energies, 2019, 12(9): 1669.
[7]陶慧,艾朋偉.改進滑膜控制雙降壓式逆變器的動力學特性[J/OL].電力系統及其自動化學報:1-8[2022-10-21].DOI:10.19635/j.cnki.csu-epsa.001115.
TAO Hui, AI Pengwei. Dynamic characteristics of double buck inverter with improved sliding mode control [J]. Proceedings of the CSU-EPSA: 1-8[2022-10-21].DOI:10.19635/j.cnki.csu-epsa.001115.
[8]WANG H Q, LIU S W, YANG X B. Adaptive neural control for non-strict-feedback nonlinear systems with input delay[J]. Information Sciences, 2020, 514: 605-616.
[9]李建偉,張磊安,黃雪梅,等.基于改進徑向基神經網絡的風電葉片模溫串級PID控制算法[J].太陽能學報,2022,43(3):330-335.
LI Jianwei, ZHANG Lei′an, HUANG Xuemei, et al. Cascade PID control algorithm for wind turbine blade mold temperature based on improved RBF neural network [J]. Acta Energiae Solaris Sinica, 2022, 43(3): 330-335.
[10]崔征山,周揚忠,張競,等.基于滑模和擴張狀態觀測器的雙繞組無軸承磁通切換電機轉子懸浮控制策略研究[J].儀器儀表學報,2022,43(6):269-279.
CUI Zhengshan, ZHOU Yangzhong, ZHANG Jing, et al. Research on rotor suspension control strategy of dual-winding bearingless flux-switching permanent magnet machines based on sliding mode control and extended state observer[J]. Chinese Journal of Scientific Instrument, 2022, 43(6): 269-279.
[11]黃道敏,韓麗君,唐國元,等.水下機械手分數階積分滑模軌跡跟蹤控制方法研究[J].中國機械工程,2019,30(13):1513-1518.
HUANG Daomin, HAN Lijun, TANG Guoyuan, et al. Fractional integral sliding mode control for trajectory tracking of underwater manipulators[J]. China Mechanical Engineering, 2019, 30(13): 1513-1518.
[12]鄧檳檳,尚偉偉,張彬,等.6自由度繩索牽引并聯機器人的快速終端滑模同步控制[J].機械工程學報,2022,58(13):50-58.
DENG Binbin, SHANG Weiwei, ZHANG Bin, et al. Fast terminal sliding mode control with synchronization error for a 6-dof cabel-driven parallel robot[J]. Journal of Mechanical Engineering, 2022, 58(13): 50-58.
[13]梁相龍,姚建勇.基于神經網絡的機電伺服系統非線性控制[J].控制與決策,2023,38(4):1008-1014.
LIANG Xianglong, YAO Jianyong. Nonlinear control of mechatronic servo system based on neural network [J]. Control and Decision,2023,38(4):1008-1014.
[14]RAZMI H, AFSHINFAR S. Neural network-based adaptive sliding mode control design for position and attitude control of a quadrotor UAV[J]. Aerospace Science and Technology, 2019, 91: 12-27.
[15]CHEN S Y, LI T H, CHANG C H. Intelligent fractional-order backstepping control for an ironless linear synchronous motor with uncertain nonlinear dynamics[J]. ISA Tran-sactions, 2019, 89: 218-232.
[16]熊蕊.考慮瞬態性能的工業機器人雙臂反步控制方法[J].現代制造工程,2022(8):53-59.
XIONG Rui. Back-stepping control method of industrial robot dual-arm considering transient performance[J]. Modern Manufacturing Engineering, 2022(8): 53-59.
[17]FU C Y, HONG W, LU H Q, et al. Adaptive robust backstepping attitude control for a multi-rotor unmanned aerial vehicle with time-varying output constraints[J]. Aerospace Science and Technology, 2018, 78: 593-603.
[18]XU G W, ZHOU R X, LIU W, et al. The equivalent sliding mode tension control of carbon fiber multilayer diagonal loom[J]. International Journal of Control, Auto-mation and Systems, 2019, 17(7): 1762-1769.
Neural backstepping fractional order fast terminal sliding mode control of warp tension
FU Maowen1, SHEN Danfeng1, ZHAO Gang2, SHANG Guofei 1, BAI Shunwei1
(1.School of Mechanical and Electrical Engineering, Xi'an Polytechnic University, Xi'an 710048, China;
2.Shaanxi Changling Textile Mechanical & Electronic Technological Co., Ltd., Baoji 721013, China)
Abstract:
With the progress of current computer technology and modern control methods and theories, the textile field has been fully developed in the past decade, gradually realizing intelligence and advancement. However, the domestic textile industry still has the problems of low competitiveness and high labor costs. Looms in textile machinery need to be closely integrated with electromechanical equipment. High-quality looms apply more advanced algorithms to looms on the basis of continuous pursuit of higher weaving efficiency and fabric quality, reducing the degree of manual intervention. The performance of the let-off mechanism, a direct tension control mechanism, determines the speed and efficiency of the loom spindle. Studying the let-off system and developing a more efficient control algorithm or structure is an important factor to improve the performance of the loom, which meets the national economic needs and social significance of China.
In order to enhance the matching degree between the let-off mechanism and weaving requirements of looms, the key control algorithm of the let-off mechanism is designed, which is combined with modern control theory to improve the robustness and stability of the control algorithm. This research aims to develop a neural backstepping fractional order fast terminal sliding mode controller (RBF-BCFOFTSMC) to control the warp tension. Firstly, the time-varying mathematical model of the let-off system of the loom was established through dynamic analysis. In order to improve the dynamic response performance of the system and reduce chattering, a new backstepping fractional order fast terminal sliding mode control method was derived. Since there are disturbances such as motor vibration and heald frame motion in the weaving process of the loom, and the upper bound of the total disturbance of the system is unknown, an adaptive law was designed to estimate the upper bound of the external disturbance. The time-varying characteristics of the system make the controller have unmodeled and modeling uncertainties. The neural network parameter adaptive law was designed to approximate the real system state, and the Lyapunov stability was used to prove the rationality of the system. In order to verify the effectiveness of the designed control strategy, it was compared with traditional sliding mode control (SMC) and neural PID (RBFPID) in simulation experiments and actual working conditions. The results show that RBF-FOTSMC not only reduces chattering in warp tension control, but also has high robustness and response performance.
Through the research, the algorithm design and experiment of the let-off control system have been successfully completed, which has greatly improved the control effect, robustness and stability of the system. However, as the loom let-off system is a complex control system, more research needs to be supplemented in the future from two main points. First, it is necessary to study the influence of heald frame, weft insertion, beating up and other movements on loom tension, and analyze the influencing factors for corresponding tension compensation. Second, the adopted hardware needs to be optimized. If the controller with faster processing speed can be replaced, the high-speed and advanced level of the loom will be improved.
Keywords:
warp tension; fractional order; backstepping; sliding mode controller; neural network