The Concept of Virtual Laboratory and PIL Modeling with REX Control System

Transcription

1 st International Conference on Process Control (PC) June 6 9, 2017, Štrbské Pleso, Slovakia The Concept of Virtual Laboratory an PIL Moeling with REX Control System Stepan Ozana VSB-Technical University of Ostrava Faculty of Electrical Engineering an Computer Science Czech Republic, Ostrava, stepan.ozana@vsb.cz Tomas Docekal VSB-Technical University of Ostrava Faculty of Electrical Engineering an Computer Science Czech Republic, Ostrava, tomas.ocekal@vsb.cz Abstract This paper eals with the escription of concept of a virtual laboratory esigne for eucational purposes an eployment of processor-in-the-loop (PIL) concept with the use of the REX Control System. It also presents the case stuy regaring control of a simple inverte penulum on the cart. The below mentione techniques represent moern, accessible an aaptive teaching methos supporting a istance eucation. I. INTRODUCTION State-of-the-art new information technologies provie alternative tools for improving teacher-stuent interaction. A web-base, multi-user virtual lab is also possible without necessity of reprogramming the eperimentation interface coe [1]. It helps to stuy ynamic simulations, it supports user interactivity, the generation of new eperiments, an the opportunity to practice with classical or avance control strategies applie for ifferent systems. The virtual laboratory is an interactive environment for creating an conucting simulate eperiments: a playgroun for eperimentation [2]. The iea of virtual laboratory is very clearly escribe in [3]: virtual laboratory is a computer-base activity where stuents interact with real apparatus via a computer link, yet the stuent is remote from that apparatus. We shoul istinguish the latter case from a computer-controlle eperiment, where a stuent will irectly control an apparatus in his or her vicinity via a computer interface, see Fig. 1 (aopte an moifie from original sources [3]). process process Fig. 1. Distinction between a computer-controlle eperiment an a remote controlle eperiment The reference [3] also introuces a efinition of virtual lab in technical sense: a virtual laboratory is one where the stuent interacts with an eperiment or activity which is intrinsically remote from the stuent or which has no immeiate physical reality. The latter part of this efinition may seem to imply that virtual laboratory can have no physical reality behin it at all. However, the entire concept of virtual laboratories supports as close connection to reality as possible, having the real apparatus present at the laboratory. The reference [4] efines the key iea behin this concept as a esign an evelopment of laboratories which have real harware set-ups an can be virtually accesse by users (stuents or researchers) via the Internet. Once a virtual laboratory has been set up, users will be able to conuct actual eperiments from remote computers anywhere in the worl 24/7 as if they were working in actual laboratories. However, some of the references istinguish between two terms: a virtual laboratory (no presence of real physical setup) an a remote laboratory. REX-base solution propose allow both of these concepts, however the paper is primarily aime at a virtual lab. Assuming a high-fielity mathematical moel of controlle process, the concept of virtual laboratory can be moifie in the way that real apparatus may be replace by this mathematical moel. This approach closely relates to the techniques of real-time simulations such as HIL (harware-in-the-loop) or PIL (processor-in-the-loop). The latter will be emonstrate with the REX Control system, being aopte for the real-time synchronization. There are several solutions mentione in literature base on ifferent technologies. In [9], authors escribe the solution Remote Electronic Lab (REL), which uses the LabVIEW for interaction with real eperiment an for its simulation via Internet Eplorer browser (ver. 5.0) with ActiveX Control an Visual Basic scripts. Similar solution is escribe in [10], [11] an [12] where the eperimental part is carrie out in LabVIEW. The solution in [13] is base on MATLAB, introucing two alternatives for remote access: MATLAB Web Server an WinCon software. Other virtual labs [11], [12], [14] use stanar web browser with various components (Java applets, JavaScript, HTML, Flash, XML) for virtual laboratory purposes. The possibility of connection with wiely use Moole learning management system is mentione in [11] /17/$31.00 c 2017 IEEE 98

2 II. REX-BASED CONCEPT A. Basic Information About REX The REX Control System is a family of software proucts for automation projects. It can be use it in all fiels of automation, robotics, measurements an feeback control [5]. Main features of the REX Control System are as follows: Graphical programming without han-coing Programming control units on a stanar PC or laptop User interface for esktop, tablet an smartphone Wie family of supporte evices an input-output units Inustry-proven control algorithms Easy integration in business IT infrastructure (ERP/BMS) REX Control system supports both computer-controlle an remote controlle eperiments, incluing MIL, SIL, PIL, HIL as escribe in [6]. A typical case stuy concerning virtual lab an PIL technique is emonstrate in net subsections. B. General PIL Concept Versus Virtual Laboratory Concept The basic iea of PIL is to provie possibility to test functionality of the esigne control system with the plant moel instea of the real plant. The control system with its algorithms is running on so calle embee target. Uner real conition it controls the real plant via I/O interface. PIL technique makes it possible to substitute a plant moel instea, proviing that all I/O signals are emulate at a communication line as escribe in Fig. 2. The embee target is usually connecte to a so calle host computer so the entire block scheme can be epresse by Fig. 3. I/O PLC/PAC/uPC/softPLC/IPC (embee target) I/O (emulate) I/O comm. interface (real plant) Fig. 2. Iea of PIL simulation (plant moel) Fig. 3. Block scheme of PIL simulation components The basic iea of a virtual laboratory is to provie almost the same feeling an results as if one was in the laboratory in-situ. Reference [7] states: the avances in computer graphics an multimeia technology allow construction of virtual physical environments that are representative, etaile an realistic. Original virtual lab efinition supposes presence of a real setup connecte to control system via I/O signals an accessibility of the variable through the network. However, a real setup may require some special conitions an there is always a potential risk of its malfunction. Having a precise mathematical moel aopte for PIL technique, it is possible to use it instea of the real setup while the concept of virtual lab can stay unchange. C. PIL Concept with REX Definition of PIL claims that simulation of the plant moel is synchronize with the control algorithms running on the embee target at full rate (as fast as possible). It is in accorance with one of the reasons why PIL is esigne for, namely choice of appropriate sampling perio an verification if the neee performance of embee target is achieve. However, the real-time synchronization on both sies (embee target an simulation of the plant moel) oes not clash the efinition. It is secure by ReCore - a runtime core running on the target evice (Linu IPC, WinPAC, Raspberry Pi etc.). It hanles timing an eecution of all algorithms an provies various services. The iniviual tasks are eecute using preemptive multitasking. From REX point of view, the following configurations cover these typical situations, see Fig. 4: two targets (A,B) linke via a communication line, both with REX run-time, A consiere as embee target hosting control algorithms (C), B consiere as a plant moel (S) (with the monitor or touch-screen attache to simulate real feeling similar to observation of a real plant)-etermine for situations where REX control system is use both for control an for simulating a mathematical moel of the real plant two targets (A,B) linke via a communication line, A consiere as embee target hosting control algorithms (C) with no nee of REX run-time of any type (PLC, IPC,...), B consiere as a plant moel (S) (with the monitor or touch-screen attache to simulate real feeling similar to observation of a real plant)-etermine for situations where control algorithms (C) are implemente in non- REX platform an shoul be teste with a mathematical moel (S) of the real plant mae create in REX two targets (A,B) linke via a communication line, A consiere as embee target with REX run-time hosting control algorithms (C), B consiere as a plant moel (S) implemente on PC an Simulink (with the monitor attache to simulate real feeling similar to observation of a real plant)-etermine for situations where control algorithms (C) are implemente in some REX platform an shoul be teste with a mathematical moel (S) of the real plant create in Simulink (+ aitional Simscape-family blocks for physical moeling) or any other simulation software environment 99

3 Fig. 4. Typical REX PIL Configurations D. Virtual Laboratory Concept with REX The following features of REX Control system etermine the possibility to use it for the concept of virtual laboratories: it is capable of generating an performing web-base an customize visualizations it contains its own embee webserver, therefore the only thing one nee to perform remote eperiment is a web browser the web visualization can be passwor-protecte an secure from technical point of view there is no ifference between computer-controlle an remote controlle eperiment it is multiplatform an it allows wie possibilities to attach various I/O moules Apart from web access, REX also allows remote access by winows-base ReView iagnostic tool proviing hierarchical information about the running control algorithm. III. NOVELTY OF THE SOLUTION, METHODOLOGY OF USAGE, APPLICABILITY The most important novelty of this solution lies in: inepenence on platform ecellent price/performance ratio no nee of etra SW equipment ease of use, transfer to other platforms, integrate uner one roof approach With this propose approach, both control algorithms an moel of the plant can run on PC, minipc (such as Raspberry), PLC or IPC, moreover uner ifferent operating systems. Moel of the plant is suppose to be create in REX control system, but control algorithms can be implemente in various SW environments running on various HW platforms, while communication between these two parts is establishe via stanar protocols such as RS, I 2 C, Mobus, Profinet an others. As for visualization, the only tool neee to isplay visualization is a stanar web browser running on PC, smartphone, tablet, or touchscreen. Moreover, there are more types of visualization at isposal. The basic visualization can be auto-generate via a single-click base on the block scheme create in REX environment (ReDraw). It is isplaye within a web browser, showing the same block structure as in the source scheme, isplaying live values of inputs an outputs of the blocks an allowing setting blocks parameters on the fly. In case of more avance visualization screens, so calle ReHMI Designer is at isposal. It makes it possible to use both embee an custom graphical elements, an let them be easily boun to particular REX variables. Composition of the entire visualization is carrie out in the same environment (ReDraw) as both all algorithms an visualization are integrate into one common project. Last but not least, transfer between SIL (software in the loop, thus moeling control algorithms an the plant in the same environment running on the single target) an PIL is etremely easy. Methoology of usage of the concept escribe in this paper is as follows: setting up the mathematical moel of the plant an control algorithms REX implementation of the mathematical moel of the plant using REX blocks implementation of control algorithms on a chosen platform (it is also REX-base platform, SIL technique is very useful to try before final PIL) implementation of PIL technique (emulating all physical signals to communication protocol) implementation of web-base visualization, bouning of iniviual visualization variables to REX signals testing PIL moel, changing control algorithms, changing plant s parameters on the fly transfer of the solution towars control of real plant The propose solution is especially useful for eucation process ue to favourable licensing an technical capabilities. It is wiely applicable in analysis an synthesis of control systems. IV. CASE STUDY: CONTROL OF SIMPLE INVERTED PENDULUM ON THE CART Control of the simple inverte penulum on the cart has been chosen as the case stuy in this paper for more reasons. First of all, it is a challenging an interesting task, also attractive for stuents an researchers. It is a very goo eample of feeback control system, combining classical algorithms (PID) an more avance algorithms base on so calle moern control theory. The topic of inverte penulum control has been treate so many times, much more than any other physical eucational moel. This paper oes not escribe the etails of control approach, but yet it shows some novelty. It very clearly shows the concept of the mathematical moel an control algorithm through inputs from two encoer signals (penulum angle, cart position) an a PWM signal representing manipulate value brought to power electronic unit an DC motor. The physical setup use for this case stuy is shown as its 3D moel in Fig

4 The entire state-space escription of inverte penulum (IP block in Fig. 8) can be escribe by Eq. (2)-Eq. (5). ẋ 1 = 2 (2) Fig. 5. 3D moel of inverte penulum on the cart A. Moeling of Control of Inverte Penulum on the Cart There are more approaches an equations escribing a simple penulum moel. Physical setup shown in Fig. 5 can be aopte for ientification in the form epresse by Fig. 6. where ẋ 2 = g l sin 1 b l l u cos 1 (3) ẋ 3 = 4 (4) ẋ 4 = u (5) 1 [ra] penulum angle 2 [ra s 1 ] penulum angular spee 3 [m] cart position 4 [m s 1 ] cart spee u [m s 2 ] cart acceleration Actuator (DC motor an electronic unit for a spee control) can be consiere as a part of the system with its ynamic approimate by a fast 1 st orer system. Though mathematical moel gives the information about cart an penulum spee, PIL concept reflects the number of variables being at isposal uner real conitions, therefore only cart position an penulum angle are measure an thus representing the two outputs, see Fig. 7. The entire moel of the system consiere for PIL technique an virtual laboratory is then shown in Fig. 8. RPi/PLC/PC u* 1 v* s I/O R e al System 1 K(t) v sp C 4 S 3 Fig. 7. Control of the real system of inverte penulum on the cart Fig. 6. Situation scheme for analytical ientification It is possible to erive Eq. (1) for a penulum movement where l ϕ g sin ϕ + b ϕ u cos ϕ = 0 (1) l [m] penulum length to the mass center MP ϕ [ra] penulum angle g [m s 2 ] gravity acceleration b [m s 1 ] amping coefficient u [m s 2 ] cart acceleration In orer to get the full state-space escription of the system, Eq. (1) can be etene by two state variables representing position an spee of the cart, supposing u as controlle input. The time constant τ has been ientifie eperimentally. It represents ynamical behavior of the inner spee loop implemente as a fast PI controller. Note that control system hosting control algorithms is ientical both in control of the real system (Fig. 7) an its PIL representation (Fig. 8). Remaining unmeasure states 2 an 4 representing cart an penulum spee are approimate by numeric erivatives. RPi/PLC/PC K(t) C RPi+TouchScreen Mobus u* 1 v* v sp s + - 1/ 4 Inner spee loop moel S u 1 v s Fig. 8. PIL representation of control of inverte penulum on the cart IP

5 B. Implementation of PIL moel with Raspberry Pi3 C. Web Visualization for Virtual Laboratory The concept escribe by Fig. 8 has been implemente in REX Control system running on two RPi3 single-boar computers in accorance with Fig. 3. Fig. 9 shows the picture of PIL testing stan. Upper part of this picture shows Rpi3 as embee target hosting control algorithms for inverte penulum while the visualization touch-screen attache to the 2n Rpi3 (beyon the touch-screen) serves as the mathematical moel accoring Eq. (2)-Eq. (5). These computers use Mobus communication protocol for the ata echange. Web visualization contains both tet an visual information about position of the cart an penulum ro, using ynamic graph components an tet boes. Visualization of the movement of both cart an ro positions are isplaye in real time. It is compose of two objects representing the cart an the ro. Firstly rotation property of the ro is boun for the 1 REX variable. Then it is groupe with the cart object, an translation of this group is assigne to 3 REX variable. This is possible ue to ReHMI capabilities of REX Control system, allowing sequential animations an thus there is no nee to perform eternal conversion between particular coorinate systems. Apart from feeling or real eperimentation for a remote user, it is also important for the virtual laboratory to have the possibility of ata eport an archiving. ReHMI esigner makes it possible to use TRND web component capable of selecting signals to be isplaye, viewing the current trens or pausing the simulation in orer to analyse the time waveforms. The eport of raw values can be ae to the HTML source coe or easily via ReView remote client. The eperiment is performe via the switch responsible for swing-up, regulation in the upright position an swing-own. The etail screenshot of the website in a web browser corresponing to Fig. 9 showing regulation in upright position is given in Fig. 11. Fig. 9. Photo of the testing stan for PIL moel of control of inverte penulum on the cart Fig. 11. Web visualization screenshot Fig. 12 isplays the ata eport representing penulum angle an cart position uring the swing-up an steay states. V. C ONCLUSION Fig. 10. Photo of the of inverte penulum on the cart Once the control algorithms are verifie by PIL simulation, they can be eploye for the real system shown in Fig The concepts of PIL an virtual laboratory introuce in this paper have been ocumente for the particular case stuy of the most typical control laboratory eperiment, that is a simple inverte penulum on the cart. Description of control algorithms was beyon the scope of this paper. The main relevance of ocumente approach in case of inverte penulum moel is the thorough esign an test of swing-up algorithms as it is quite ifficult to compute a functional control strategy concerning constraints given by physical reality, mainly the limits of the range in -ais an a possible saturation limit for a spee controller.

6 Transition between a virtual laboratory an a remote laboratory (control of the real system an its PIL moel) is easily possible ue to the possibility of connecting I/O interfaces to the various supporte HW platforms, such as Aruino boar which is currently use to control the real apparatus of inverte penulum on the cart (Fig. 10) or Monarco HAT for RPi [8]. The propose solution is currently teste using a ifferent embee platform hosting control algorithms, namely the PLC SIMATIC S7 that can be easily connecte to the Rpi3 hosting a mathematical moel of the plant by S7 Comm - a special river evelope by REX Controls, s.r.o. Current virtual laboratory may be etene by a visual feeback provie by a web camera broacasting the stream to the website together with the visualization. Consiering the iea of replacing real apparatus by its mathematical moel, it is possible to host several virtual eperiments within one piece of harware with a single REX run-time use for the moels. The presente case stuy is suppose to become a template base on which other similar moels are planne to be create in the Laboratory of Control Systems at Department of Cybernetics an Biomeical Engineering. As there are several subjects that eal with control theory in ifferent levels, it is planne to aopt the iea for classical PID algorithms an algorithms base on moern control theory with the use of current physical eucational linear an nonlinear SISO an MIMO moels (helicopter, ball on beam, ball on plate, ball on spool, rotary inverte penulum, magnetic levitation). Fig. 12. Penulum angle an cart position uring the swing-up Unsuccessful attempts uring control esign applie in the real system without PIL simulation may result in a amage of the real setup. For a ouble or a triple penulum, esign an verification of control algorithms with the use of PIL or HIL become a necessity. Virtual laboratory is base on moern HTML5 technology, allowing all the features require for this concept. As it allows secure multi-user access with customizable ifferent levels of user rights, it becomes efficient both for stuents an researchers. However, it is obvious that certain time slots for particular users shoul be consiere an introuce to the system for efficient remote work to avoi unwante interference cause by possible concurrent actions. A very important feature of the propose solution is a possibility of introucing customizable interactivity of web visualization components. For eample, the graphical objects can be raggable, or assigne with other actions allowable by Javascript events. Having applie for this case stuy, it is possible to simulate a isturbance in the control circuit by ragging the cart or the ro, corresponing to a manual touch of the penulum in upright position uner real conitions. Moreover, important parameters playing a crucial role in the system ynamics (such as gravity, ro length, friction coefficient) can be entere via web eit boes, see Fig. 11. ACKNOWLEDGMENT This work was supporte by the project SP2017/158, Development of algorithms an systems for control, measurement an safety applications III of Stuent Grant System, VSB-TU Ostrava. REFERENCES [1] Sanchez, J., et al. (2002). Virtual an remote control labs using Java: A qualitative approach. Ieee Control Systems Magazine 22(2): [2] [ laboratory] [3] [ [4] Nejeljko, P. (2000). Virtual laboratory for automatic control an supervision-challenges an opportunities. Proceeings of UN-ECE International Conference on Technology Transfer for Economic Development, Zagreb, Croatia. [5] [ [6] [ [7] Granao, E., et al. (2007). A Web-base virtual laboratory for teaching automatic control. Computer Applications in Engineering Eucation 15(2): [8] [ [9] Auer, Michael E. Virtual lab versus remote lab. 20TH Worl Conference on Open Lerning An Distance Eucation [10] Neic, Zorica, Jan Machotka, an Anrew Nafalski. Remote laboratories versus virtual an real laboratories. Vol. 1. IEEE, [11] Valera, Angel, et al. Virtual an remote control laboratory evelopment. IEEE Control Systems 25.1 (2005): [12] Chaos, Dictino, et al. Virtual an remote robotic laboratory using EJS, MATLAB an LabVIEW. Sensors 13.2 (2013): [13] Guimares, Eliane, et al. REAL: A virtual laboratory for mobile robot eperiments. IEEE Transactions on eucation 46.1 (2003): [14] Cheng, K. W. E., et al. Virtual laboratory evelopment for teaching power electronics. Power Electronics Specialists Conference, pesc IEEE 33r Annual. Vol. 2. IEEE,