外文翻譯--電磁閥和液力偶合器的改善的控制

1、<p><b>　　附錄A 譯文</b></p><p>　　電磁閥和液力偶合器的改善的控制</p><p><b>　　摘要</b></p><p>　　本文針對一個電磁閥連接的液力偶合器的操作法，是一種新、改善的控制方法。通過輸送機動力學模擬對該方法進行了驗證，在本文將介紹驗證結果。</p>&

2、lt;p><b>　　1 介紹</b></p><p>　　液力偶合器是自1987 年在南非被用在處理大塊固體生產上。但是, 最初的高期望未被履行, 主要由于缺乏對控制系統(tǒng)聯(lián)結的限制的理解的一些應用。液力偶合器能通過各種各樣的方式完成它的性能[1]，并且對工程師來說是扣人心弦工作，因為唯一的限制也許是某人的假想。扭矩控制最常見的方法依靠電磁閥，電磁閥啟動并在被預先確定的極限內由PLC

3、系統(tǒng)監(jiān)督整體性能。扭矩特征曲線呈鋸牙樣式。上部和下部的扭矩極限依照具體要求確定。</p><p>　　在多數(shù)場合這個技術是滿足的, 然而, 有不希望的副作用的源頭譬如那些與所謂的“抨擊”控制技術相關的。盡管這些困難, 但電磁閥是一種有效耐用的設備, 也適合原材料處理產業(yè)的環(huán)境。</p><p>　　隨后人們企圖運用這個具體設備挖掘它的全部潛能，并在期間用一臺輸送機的改善的控制技術研制一個適

4、當?shù)目刂葡到y(tǒng)。</p><p><b>　　2 背景知識</b></p><p>　　經(jīng)過一段時間研究，我們采用控制參量調整分離開關實現(xiàn)連續(xù)控制，在電子學和農業(yè)領域執(zhí)行試驗并且在區(qū)域內通過遞象開關方式實施系統(tǒng)。PUTTER 和Gouws[2]方法，保持相互依賴的參量是可能的，譬如在溫室里使用通過風扇、加熱器和噴水隆頭的分離行動使溫度和濕氣保持平衡。當接受溫室里高變化動

5、力學差異時, 對電磁閥和液力偶合器的應用開發(fā)和研究一個相似的概念是有意思的。</p><p>　　2.1 控制系統(tǒng)算法</p><p>　　近年來，控制系統(tǒng)在現(xiàn)代控制技術上應用有許多種, 譬如模糊邏輯控制[4] 、神經(jīng)網(wǎng)絡[5]，它是作為對非線性和多變量控制問題的解決辦法。盡管所有成功例子, 許多典型的技術, 譬如比例-積分-導數(shù)(PID)控制。PID控制比智能控制技術的優(yōu)勢是少處理器容量

6、和所需時間。但是, 雖然已解決一些問題[6], 但PID 控制器仍然保留一個難題, 尤其是在系統(tǒng)中有不可接受的實驗錯誤方法。PID控制的缺點是調節(jié)系統(tǒng)非線形性能力受限制。</p><p>　　電磁閥和液體偶合器組合應用PID控制技術。通過控制閥門的開關, 電動機/連軸器扭矩能保持不變。系統(tǒng)失效時間是足夠短的，但對系統(tǒng)非線形性調節(jié)不明顯, 在這些情況下，可應用PID控制技術，描述控制等式得如下[7]：</p&

7、gt;<p><b>　?。?-1）</b></p><p>　　E —誤差，區(qū)別在凝固點和被觀察的價值之間；</p><p>　　T —所需扭矩變動量[Nm]；</p><p><b>　　Kc —比例增量；</b></p><p>　　Td —派生導數(shù)時間常數(shù)[s]；</p&

8、gt;<p>　　Tj —累積時間常數(shù)[s]；</p><p>　　Ts —樣品時間常數(shù)[s]；</p><p><b>　　n —樣品數(shù)字；</b></p><p>　　公式1-1與常規(guī)PID方法相反，它相對于當前的扭矩(△T)確定所需變化量。這是一種快而方便的確定輸出量的方式, 因為沒有數(shù)字微分或積分。</p>

9、<p>　　數(shù)字n 是當前的測量誤差, n - 1是早先測量，n - 2是n-1的前一個。測量間的時間是重要的。太短的采樣時間能引起設備過分循環(huán), 而太長采樣的時間能引起超載和不穩(wěn)定[8]。常數(shù)Kc、Tj、Td最初根據(jù)知名的ZIEGLER-NICHOLS方法在理論上確定[6]。在控制器的性能上這些常數(shù)的每個作用用史密斯描述了[3], 被總結如下[7]：</p><p>　　Kc值小產生超載現(xiàn)象但穩(wěn)定性好

10、, 但當Kc值大時減少超載現(xiàn)象但增加設備循環(huán)。</p><p>　　Tj值小消除恒定值誤差, 但導致控制設備迅速循環(huán)。反過來, Tj 值大導致產生恒定誤差。</p><p>　　Td值小導致超載大, 當Td值大提高反應時間時, 導致穩(wěn)定性提高。</p><p>　　雖然閥門只能關起, 但如果通過控制開關開啟、停滯時間的線性地變化的閥來控制開關是可能的。公式1-2，△

11、T被轉換成使用率, 被定義如下：</p><p><b>　　使用率：</b></p><p>　　△T = （1-2）</p><p>　　—交換閥門的時間() [s]；</p><p>　　—各樣品間在一時間間隔期間閥門交換的最大時間的時間(tmax = )[s]；&l

12、t;/p><p>　　實際上，轉換能以不同方式完成，涉及△T與閥門操作的時間之間從一個簡單的比例關系到一相當復雜關系，其中要考慮扭矩/油流量變化率。假想的發(fā)展或者選擇通過幾個因素治理，其中由軟件控制和硬件限制的簡單化是二個最明顯部分的。</p><p>　　對于這種具體應用，閥門交換每秒不可以超過5次或5赫茲。最大工作循環(huán)(tmax)的最佳值要根據(jù)指定的標準或動態(tài)模仿實驗經(jīng)反復試驗確定。<

13、;/p><p>　　機模擬實驗校核改善概念的性能：測試概念并通過輸送機模擬校核。第一套由計算機編程建造的輸送機模型是以兩年之前用這種材料的詳細設計而研制的。輸送機是通過液力偶合器驅動，它連接著三項換向閥。有關系統(tǒng)和輸送機的詳細的信息可能都來自[9]。偶合器的應用指標與在[11]中描述的差不多，比如在南非很有代表性的那種液力偶合器。</p><p>　　在重要的零件圖中，需要分析啟動階段的轉矩。

14、合格的運行結果是通過調節(jié)運行周期和控制設置來實現(xiàn)的。這樣可以進一步探索啟動階段速度。為了完成進一步的研究工作，制作的模型將拓展到輸送機的高配置的動力裝置。在這張圖表上涉及到了8.5m長帶有頂部和尾部驅動裝置的輸送機。盡管是帶有鏟斗的系統(tǒng)操作，但是在這些模擬裝置中輸送機的模型是與液力偶合器連接在單向閥口處。</p><p>　　2.2 輸送機動態(tài)模型的結果</p><p>　　3.3km長鋼

15、絲輸送機扭矩率116.0Nm/s，在開始階段描繪的圖的坡度大，尤其在泵剛工作的前6秒之間。從而，大約延遲12秒后，控制系統(tǒng)開始啟動，存在兩個問題：</p><p>　　1）最初，低扭矩傳輸不允許輸送機機立即加速度，但通過軟件的調整這種情況就可以，正如模擬實驗的那樣，注意速度變化和控制開始的時間不少于在電機啟動后的8秒，雙向泵啟動后的6秒。</p><p>　　2）油從偶合器卸荷的速度是有限

16、的，也許不夠抵制輸送機機的動態(tài)反應，因此，在開始的初始階段注意某種的程度超速。</p><p>　　盡管已提出一些問題，但啟動開始速度的變化的情況比希望從控制操作的分散模塊上解決問題更有深遠意義。</p><p>　　進一步測試, 運用不同的速度曲線或者控制常數(shù)能得到更好的結果。</p><p>　　整體的結果是合理的，但是，值得注意是在某種程度上尾部驅動比頭部的性

17、能好。</p><p>　　對于每個驅動裝置，在開始的12秒到22秒之間提高扭矩或者張力是可行的，再一次調整設備或者工作因數(shù)，能改善結果。</p><p><b>　　3 結論</b></p><p>　　基于模擬的結果能得一下結果：</p><p>　　1）電磁閥的控制概念在很大程度上已實現(xiàn)，把閥的分散運動變成精確的、

18、連續(xù)的輸出是可能的。</p><p>　　2）新系統(tǒng)允許電磁閥的應用可根據(jù)原策略的速度，然而，對于扭矩來說控制可能不精確，尤其在開始啟動階段。</p><p>　　3）保證控制參數(shù)的優(yōu)化，模擬實驗的結果表示選擇不當?shù)目刂茀?shù)會導致不好的性能結果。</p><p>　　4）如所示，這個概念能被應用于中長型輸送機中，尾部驅動不易限制，然而，在這種情況上，只有對開始的扭矩

19、進行測試。</p><p>　　5）PID控制是傳統(tǒng)的控制技術，它廣泛應用于加工工業(yè)中，并能解決不同的控制問題。</p><p><b>　　附錄B 外文文獻</b></p><p>　　Improved control of a solenoid valve and drain coupling</p><p><

20、;b>　　Summary</b></p><p>　　The paper looks at a new, improved control concept for a fluid coupling operating in conjunction with a solenoid valve.</p><p>　　Verification of the method ha

21、s been done by means of conveyor dynamic simulations and some of the results are presented in the paper.</p><p>　　1.Introduction</p><p>　　Drain fluid couplings have been available in South Af

22、rica since 1987 and have found acceptance in the bulk solids handling industry. However, initial high expectations have not always been fulfilled, mainly due to some applications which showed a lack of understanding of

23、 the coupling’s limits of performance and/or deficient control systems. The drain coupling can be assisted in its performance by various means [1] and in this respect is exciting to work with for an engineer, as the onl

24、y limit</p><p>　　This technique is sufficient in most instances, but, however, may be the source of undesirable side effects such as those associated with so called “bang-bang” control techniques. Despite t

25、hese difficulties, the solenoid valve remains a cost effective and robust device, well suited to the environment of materials handling industry.</p><p>　　Subsequently an attempt has been made to utilize this

26、 specific device to its full potential and to develop a suitable control system which would allow improved control of a conveyor during start up.</p><p>　　2. Background Information</p><p>　　Fo

27、r some time research has been performed into continuous control by means of discrete on/off adjustment of control parameters. Trials were performed and systems implemented in areas as far afield from bulk solid handing

28、as switch mode power electronics and agriculture. As an example of the potential of this approach PUTTER and Gouws [2] stated that it was possible to maintain a pre-determined level of interdependent parameters such as t

29、emperature and humidity in a greenhouse by means of disc</p><p>　　2.1 Control System Algorithm</p><p>　　The control system arena has, in recent years, been filled with publications on modern c

30、ontrol techniques, such as fuzzy logic control [4] and neural networks [5] as solutions to non-linear and multi-variable control problems. Despite all the success stories, more classical techniques, such as Proportional

31、-Integral-Derivative Control (PID). An advantage of PID-control, is that less processor capacity and time is required, than with the intelligent control techniques. However, even though many</p><p>　　The s

32、olenoid valve and drain coupling combination lends itself to the application of PID control. By controlling the switching of the valve, the motor/coupling torque could maintain a pre-determined pattern. The system dead

33、time is short enough not to contribute significantly to system non-linearity, in which case PID-control may be applied. The control equation is described as follows [7]:</p><p><b>　　[1]</b></p

34、><p><b>　　where:</b></p><p>　　E -error, difference between set point and observed value</p><p>　　T- required torque change [Nm]</p><p>　　Kc -proportional gain

35、 </p><p>　　Td- derivative time constant [s]</p><p>　　Tj- integral time constant [s]</p><p>　　Ts- sample time constant [s]</p><p>　　n- sample number.</p><p>

36、;　　Contrary to the conventional PID approach Eq. (1) determines the change required relative to the current torque (△T) and not the amount of torque required as a function of the error. This is a quicker and more conve

37、nient way to determine the output, since no numeric integration or differentiation is required.</p><p>　　The number n refers to the current error measurement, n –1 to the previous measurement and n- 2 to th

38、e one before that. The time between measurements is critical. A too short sampling time can result in excessive equipment cycling, while a too long sampling time can result in overshoot and instability [8]. The constant

39、s Kc,Tj Td are originally determined theoretically according to the well-known ZIEGLER-NICHOLS method [6]. These values serve as a starting point from where further fine tuning </p><p>　　A small value of Kc

40、 produces large overshoot but gives good stability, while larger values of Kc reduce the overshoot but increase equipment cycling.</p><p>　　-Small values of Tj eliminate constant errors quickly, but result

41、in rapid cycling of control equipment. In turn, large values of Tj cause constant errors to occur.</p><p>　　-A small value of Td causes large overshoot, while a large value of Td increases the reaction t

42、ime, which results in increased stability.</p><p>　　Although the valve can only be switched on or off, it is still possible to control the switching as if it were a linearly varying valve by controlling the

43、 time for which it is switched on or off. By applying Eq. (2), △T is converted to a duty cycle, which is defined as follows:</p><p>　　Duty cycle = [2]</p><p><b>　　where:</b></p&g

44、t;<p>　　-time for which the valve is switched on () [s]</p><p>　　-time between each sample which also represents maximum time for which the valve can be switched on during one interval (tmax = )[s].&l

45、t;/p><p>　　In fact the conversion may be done in several ways ranging from a simple proportional relationship between △T and time of valve operation to a rather complex one where rates of torque/oil flow chan

46、ge are taken into account. Development and/or selection of the concept may be governed by several factors of which simplicity of the control software and limitations of the hardware are the two most obvious ones.</p&

47、gt;<p>　　For this specific application, the valve may not be switched more than 5 times per second or 5 Hz. The optimum length of the maximum duty cycle (tmax) may be determined by trial and error according to spe

48、cified criteria or by dynamic simulation.</p><p>　　Conveyor Simulations Verifying the Performance of the Developed Concept: The concept has been tested and verified by means of conveyor dynamic simulations.

49、 The first set of simulations utilised a computer model of a conveyor which was a subject of a detailed design investigation by Dynamika Materials Handing two years ago. The conveyor was supplied with drain couplings w

50、orking in conjunction with a three way valve. Detailed information about the system and the conveyor may be found in [9]. T</p><p>　　The torque based starting strategy was analysed in significant detail. Ac

51、ceptable performance results were obtained by adjusting the duty cycles and control settings. This led to further exploration of a velocity based starting strategy.</p><p>　　To complete the investigation, t

52、he simulations were extended to conveyors of greater length and higher installed power. Some of the graphs presented in this paper refer to an 8.5 km long overland conveyor with head and tail drives [10]. Although the a

53、ctual system operates with scoop couplings, in these simulations the conveyor was modeled with drain couplings operating in conjunction with solenoid valves.</p><p>　　2.2 Results of the Conveyor Dynamic Si

54、mulations</p><p>　　Torque was ramped at a rate of 116.0 Nm/s which is suitable for a 3.3 km long conveyor with steel cord belting (2100 Nm should be reached in 15 sec). The resulting ramp is too steep for t

55、he coupling during the initial stages, specifically during the initial 6 seconds of pump operation. Consequently, the control system comes into action with an approximate delay of 12 seconds.</p><p>　　Two

56、problems are apparent :</p><p>　　Initial low torque delivery does not allow instant acceleration of the conveyor. This is possible to rectify by adjustments in the software as is the case for the simulatio

57、ns, it can be noted that the velocity ramp and controlling action starts some 8 seconds after the motors and 6 seconds after the coupling’s pumps were energized.</p><p>　　The rate at which oil is discharged

58、 from the coupling is limited and may be insufficient to counteract the dynamic reaction of the conveyor and as a result, a certain degree of overspeed can be noted in the initial stages of the start up.</p><

59、p>　　Despite the mentioned problems, it may be stated that the results of the velocity ramp start up are significantly better than can be expected from a discrete type of control operation.</p><p>　　Furthe

60、r tests, utilizing different velocity curves and/or refined tuning of the control constants may produce better results.</p><p>　　The overall results are acceptable. However, it can be noted that the perfo

61、rmance of tail end drive is somewhat better than that at the head end.</p><p>　　For each drive it is possible to notice a period of increased oscillations of torque/tensions between the period of 12 and 22

62、 seconds in the start-up. Once again by adjusting settings and/or the length of the duty cycle, the results can improve. </p><p>　　3. Conclusions</p><p>　　Based on the results of simulations the

63、 following conclusions can be made:</p><p>　　—The objectives which were set for the control concept of the solenoid valve have been achieved to a large extent. It was possible to convert the discrete action

64、 of the valve into a precise, continuous output.</p><p>　　—The new system allows the application of the solenoid valve for velocity based starting strategy. However, control may not be as precise as for a t

65、orque based strategy, specifically during initial stages of the start-up.</p><p>　　—Care must be taken in fine tuning of the control parameters. Simulation results have shown that badly selected control pa

66、rameters may lead to poor performance results.</p><p>　　—As was shown, the concept may be applied both to very long and medium length conveyors. Existence of the tail drive is not a restriction for the appl

67、ication. However, only torque based starting strategy has been tested in this case.</p><p>　　PID-control is a classic control technique which is widely applied in the process industry and which has provid

68、ed good solutions to diverse control problems, provided that it is well-tuned</p><p>　　References</p><p>　　[1] CTREBSKI, M.: Torque control using drain fluid coupling; bulk solids handling Vol.

69、16 (1996) No. 1, pp. 39-41.</p><p>　　[2] PUTTER, E. and GOUWS, J.: An automatic environmental controller for greenhouses; Elektron (Journal for South African Institute for Electronic Engineers) August 1996,

70、 pp.66-67.</p><p>　　[3] SMITH, C.L.: Digital Computer Process Control; lntext Educational Publishers, New York, 1972.</p><p>　　[4] ZADEH, L.A.: Similarity relations and fuzzy orderings; Informa

71、tion and Science, Vol.3,1971, pp. 177-200.</p><p>　　[5] KOSKO, B.: Neural Networks and FLZZY Systems - A Dynamical Systems Approach to Machine Intelligence Prentice-Hall International Editions, New Jersey1

72、992.</p><p>　　[6] FRANKLIN, G.F., POWELL, D.J., and EMAMINAEINI, A.: Feedback Control of Dynamic Systems: Addison-Wesley Publishing, Company, 1986.</p><p>　　[7] PUTTER, E.: Multi-variable Contr

73、ol Techniques for Greenhouses; Master’s Thesis (Electrical and Electronic Engineering), Rand Afrikaans University 1996, South Africa.</p><p>　　[8] MACDONALD, R.D., HAWTON ,J., and HAYWARD G.L.: A proporti

74、onal integral derivative control system for heating and ventilating livestock buildings; J. of CSAE Vol.31,1989, No. 1, pp. 45-49.</p><p>　　[9] PRETORIUS, J.L.: Design of a long conveyor system with combi

75、ned ascending and descending sections; bulk solids handing Vol.16(1996) No. 2, pp. 195-201.</p><p>　　[10]FAUSREACH, R. and OTREESKI, M.: The Syferfontein overland conveyor system at the Sasol Secunda Plant

76、 in South Africa; bulk solids handling Vol.13 (1993) No. 2, pp. 289 - 295.</p><p>　　[11]PAGE, J.L. et al.: Design of long overland conveyor with tight horizontal curves; bulk solids handling Vol.14 (1994) N