Design and Development of a Generic, Platform Independent MoSART Environment

Transcription

1 Design and Development of a Generic, Platform Independent MoSART Environment Musti K. S. Sastry, Richard Balgobin Abstract This paper describes a new approach for the development of Modeling, Simulation, Animation and Real Time Control (MoSART) Environment. Using this approach MoSART environments can be developed on different operating systems, and exploit the power of computer networks as well to present simulations to a remote client computer. Two dynamic systems: the Inverted Pendulum on Cart system and the DC Motor Speed control system have been considered for demonstrating the effectiveness of this approach. The environment is based on Microsoft Windows, Visual C++ and MATLAB/Simulink. The environment interface permits users to select the simulation parameters such as: dynamic system, feedback controller, input signal applied and magnitude of input. The developed environment can be a useful tool for students and professionals in the field of Automatic Controls. Keywords Modeling-Simulation-Animation and Real Time Control (MoSART), Control Systems, Dynamic Systems, Controllers, Simulation Software. I I. INTRODUCTION N recent times the need for simulation and visualization software in the area of control system has greatly increased. Generally, any modern control simulation software comes with preloaded necessary mathematical models of physical systems, so that users can observe the operation, performance of the system for different input conditions. Further, users can design customized controllers for the systems to achieve desired response under control to varying conditions of disturbances. There are very few such software packages are available for specific systems such as flight control. However, such simulation software is quite expensive and often has high hardware requirements [5]. The concept of Modeling, Simulation, Animation and Real Time Control (MoSART) environments is to create a simulation environment which utilizes existing software such as MATLAB and Simulink [1] and this has been presented by number of authors thus far. By using the popular development software MATLAB/ Simulink and Microsoft Visual Studio, it is possible develop MoSART environments at much reduced budgets which can be made to Musti Sastry is with the Department of Electrical and Computer Engineering, University of the West Indies, St Augustine, Trinidad and Tobago (phone: ; fax: ; Musti.Sastry@sta.uwi.edu). Richard Balgobin is with the Department of Electrical and Computer Engineering, University of the West Indies, St Augustine, Trinidad and Tobago (phone: ; fax: ; richardbalgobin@gmail.com work on stand-alone computers as well on remote computers. Additionally the MATLAB and Simulink tools are excellent tools which can fully utilize the physical mathematical models. MoSART environments also include the animation capability; hence users can view an animation of the system under control and visualize what the system s physical response will look like. Simulation software such as this can be used to determine the feasibility of controller implementation on a system. These environments are also ideal for learning purposes, as students will be able to design controllers and observe the effects of typical disturbances on the designed system. Several authors have reported the development of MoSART environments over the last decade. The next section provides a summary of such developments. II. SUMMARY OF THE MOSART ENVIRONMENTS DEVELOPED SO FAR The concept of control system design and simulation has been implemented using various approaches, mainly using C++ or Visual C++ in combination with MATLAB [],[3],[6]-[10]. The environment developed by Rodriguez [8] demonstrated that the developed environments were more accurate representations of the control systems that the general purpose software available at that time because of their system specific nature. Gamino [3] investigated the power of interaction between the different programming languages of MATLAB and C++ to complete a control task. The work indicated that although control tasks could be accomplished by the use of only one programming language such as MATLAB or C++, the integration of these two languages allowed the specific strengths of each language to be maximized. Lim [6] showed development of a vehicle based MoSART environment, similar to that of Rodriguez s model [8] with extensive mathematical calculations as well as interactive graphical environments. Rios [7] developed a virtual pendulum environment for teaching and research purposes. Casini [] considered the virtual laboratory technologies and remote laboratory technologies to development of control system students, however did not indicate development of remote MOSART environments Rodriguez [8] implemented as many as 33 dynamic system models. He reported that the models and systems in the environment covered a diverse set of dynamic characteristics such as nonlinearity, instability and having no minimum phase, 10

2 which were used to illustrate fundamental system and control concepts. In all, various authors reported several dynamic systems that included - Cart-on-Track system, Fixed-base Pendulum system, Cart-pendulum system, Rotary Pendulum system, Seesaw-cart-pendulum system, Robotic manipulator system, Fire-arm system, Helicopters with different DOF etc. Researchers used different technologies and section III illustrates the details of these technologies for building MoSART environments. III. SOFTWARE TECHNOLOGIES USED Rodriguez [8] developed the GUI for the MoSART environment using Microsoft Visual C++ based on the MFC framework [4], which was compatible with Windows XP/000/NT/ME/98. The communication between the software was achieved by the use of C-based S-functions compiled as DLLs. Microsoft ActiveX facilitated the link between Simulink and the environment. Animations of models were achieved using Microsoft Direct 3D. Gamino [3] used Visual C++ to develop the user interface of the system. C++ was used since OpenGL libraries could be used to handle animations of the system and MATLAB DLLs could be run from the interface. The communication link between MATLAB and C++ was established by using the MATLAB DLL in the C++ interface for MATLAB computations. Lim [6] used Visual C based on the MFC framework to create a MoSART environment. Communication across the platforms was achieved by using the MATLAB -5.0 engine. Casini [] used HTML pages and Java applets to provide a user interface. The Real Time Workshop (RTW) was used to transform Simulink models into C source code which can be used by Java applet. The applet communicated with the HTML page and server by using a TCP connection over the internet. In summary, the approaches presented so far are limited to Microsoft technologies especially windows operating system and Visual C++/MFC/.NET development tools. Further, all resources and the final application will be available only one stand-alone machine. The suggested approach in this paper overcomes these major drawbacks and provides more flexibility to both developers and users. IV. PLATFORM INDEPENDENT MOSART ENVIRONMENT For this study a MoSART environment for Windows7/XP was created with a user interface which was created in Visual C++ based upon the MFC framework. MFC framework was used due to the ease of interface creation and system compatibility. The animations in this system were created using VirtualDUB software and Microsoft Paint. Animations are controlled by use of the MFC Animation Control function in the GUI. MATLAB and Simulink 010 are used to process dynamic system models and generate system responses. The communication link between MATLAB and C++ GUI was achieved by reading and writing to a text file on the local machine, which can be passed to a remote MATLAB machine using simple Message Passing Interface (MPI) for further processing. Fig. 1 provides an overview of the entire application. Fig. shows the Data communication and flow in the environment. Start User selects the dynamic system to be analyzed. MATLAB reads textfile generated by GUI to accept user inputs. MATLAB then runs scripts which act on user inputs to execute corresponding Simulink models. MATLAB Program END Visual C++ GUI Application Visual C++ GUI User selects whether or not the system is under control User Inputs are written to hard disk via textfile. MATLAB then generates plot of system response as bitmap image on hard disk. User may then run the animation to observe the physical behavior of the system. Fig. 1 System Overview FromGUI.txt MATWRITE.txt Fig. Data Communication Flow User selects the input to the system. User selects magnitude of input to the system MATLAB then writes to hard disk based on Simulink system response to determine animation of physical system. GUI then reads data from hard disk written by MATLAB to display corresponding AVI Clip. MATLAB Using the above suggested approach, two dynamic models have been modeled and simulated in the MoSART which are explained in the next section. V. DYNAMIC SYSTEM MODELS IMPLEMENTED The models implemented in this system were all Single Input Single Output (SISO) systems. Mathematical models for the following systems were implemented within Simulink block diagrams: Inverted Pendulum on Cart system DC Motor Speed Control system Non-linear models for both systems are available. The systems included display the following characteristics: nonlinear, highly non-linear, and unstable. These models can be used to illustrate fundamental system concepts and the application of the controllers to the system. A. Inverted Pendulum on Cart System The inverted pendulum on cart (Fig.3) is inherently a Multiple Input Multiple Output (MIMO) system however in 11

3 this environment; the cart s horizontal position is ignored as an output. Hence the inverted pendulum on cart system becomes a SISO system. The equations governing the dynamics of the system are as follows: (M + m) x +bx + miθ cosθ - miθ sinθ = F ( I + mi ) Θ + mgi sin Θ = -mixcos Θ (1) () of observing the operation of the system and also to control the same. The user interface for the inverted pendulum on cart is shown Fig. 5. Fig. 4 Free body diagram of rotor and Kirchhoff's equivalent circuit Fig.3 Inverted Pendulum on Cart Where x denotes the position of the cart (m), θ the angle which the pendulum makes with the vertical (rad), F the force applied to the cart (N), I moment of inertia (gm/m ), m the mass of the pendulum (gm), M the mass of the Cart (gm), b friction of cart (N/m/sec), and g acceleration due to gravity (m/sec ). B. DC Motor Speed Control System The DC Motor speed control is a simple system encountered in an introduction to Automatic Control. This system is inherently a SISO system. The free body rotor diagram and Kirchhoff s circuit for the DC Motor Speed Control system (Fig. 4) was used to develop the dynamic equations shown below: di dt d Θ 1 dθ = ( K i b ) t dt J dt 1 dθ = ( Ri + V Ke ) L dt (3) (4) Where, θ the position of shaft in degrees, V is EMF applied to motor, R and L are resistance and inductance of armature, T is torque, b is damping ratio and J is moment of inertia. Mathematical models of both the systems have been developed in MATLAB/Simulink to exploit its built-in features and to gain accuracy. The following section provides illustration of user interface screens developed in this work. VI. USER INTERFACE A simple one screen user interface has been developed as part of this work using Visual C++. This interface basically provides user to input the required parameters for the purpose Fig.5 Inverted Pendulum on Cart User Interface The Simulink model of inverted pendulum on cart is shown in Fig. 6. It can be seen that both dynamic models are provided through one single screen and user can select any model to be displayed for the purpose of study and investigation. Fig. 7 shows the DC motor speed control system user input and animation in one screen. Depending on the user inputs, the animation will appear and speed is indicated as shown. The output and response is shown in Fig. 8. VII. GENERATION OF ANIMATION When the GUI data is written to MATLAB, the corresponding Simulink model is executed and the system response is exported as a Bitmap which may be displayed in the GUI. The bitmap is executed using the MATLAB command print. The bitmap image is stored in the current directory of MATLAB which will be the GUI source folder. Based on the system responses to the applied inputs of Sine, Impulse, Step and Ramp signals, AVI s are created which portray the behavior of the physical systems based on the user inputs. For the Inverted Pendulum on Cart system, the system response displays the displacement angle of the pendulum with respect to time when the input is force acting on the cart. 1

4 Fig.6 Inverted Pendulum Simulink Model which AVI will be loaded in order to display the system s physical behavior. The GUI will then load the corresponding AVI based on the MATLAB output data. It can be seen that the two dynamic models have been successfully simulated and presented in the proposed MoSART environment. Though this may seem to be same as the previously published approaches, it is different in the sense, it does not use any special or proprietary communication technologies such as MATLAB engine or DDE or COM server. Table 1 provides all the tools and technologies that have been used in this work. This approach simply uses text files which are portable over local networks or even over internet so that the MoSART environment works on remote machines. However the design is implemented on Microsoft technologies to demonstrate this approach works well just as previous approaches. Obviously, this method has several advantages such as fewer MATLAB machines which can take up the model-parameter processing and many client/user machines which can display the input/output animations. Most importantly, version changes of operating systems, development environments will not effect this approach. Amplitude Fig. 7 DC Motor Speed Control User Interface Settling Time (sec): 0.66 Rise Time (sec): Step Response Final Value: Time (sec) Fig. 8 Closed loop step response of DC motor speed control system VIII. ADDITION OF A NEW SYSTEM This approach is very generic and hence supports any dynamic system once properly modeled. This section describes how a new dynamic system can be incorporated into this environment. Fig. 9 illustrates the step-by-step process of the incorporation of a new model. A complete analysis of the system should be undertaken separately using these equations under different inputs, disturbances and with and without controllers. Such analysis provides an idea to the developers on the range of all parameters and system behavior so that necessary interfaces can be built. Standard feedback controllers can be used to illustrate the system operation with and without control. The outputs corresponding to different inputs can be plotted and then exported to separate JPEG and AVI files so that they can used by VC++ GUI. This completes the model simulation on Matlab side. A new menu item is created (either with a radio button or as an item in a drop-down menu) in the MoSART environment. The necessary source code should be added to the menu item resource for processing the inputs and also to displaying the outputs using case statements. For the DC Motor speed control system shows the variation of the speed of the motor as the input is changed. Animation is created using a diagram of the physical system which comprised of elements to display the output. For the Inverted pendulum on cart, the displacement of the pendulum can be clearly visualized while the rotor of the DC motor shows its speed. The AVI s of all system responses have been created frame by frame using Virtual Dub and MS Paint software to create and compile images. When the Simulink model is executed, MATLAB writes to the hard disk to control 13

5 TABLE I SOFTWARE TECHNOLOGIES USED FOR DEVELOPMENT Software Use of Software in MoSART 1. MATLAB/Simulink MATLAB/Simulink is used to simulate the mathematical computations of the dynamic systems. MATLAB reads the input data from a designated text file, which is prepared by the Visual C++ GUI. Simulink output data is used to create animation and plots in the VC++ GUI.. Visual Studio 010 The GUI is created using MFC which handles both input and output. This GUI can be located on a separate computer, called client computer 3. MS Notepad Notepad is used as an editor for text-files in the system. It is also extensively used for testing the functionalities of both the GUI and MATLAB, especially during data transfer. 4. Google Sketch Up Google SketchUp is used to create the frames for the DC motor model animations. 5. Microsoft Visio 007 MS Visio is used to create the frames for the inverted pendulum on cart system animations. 6. VirtualDub VirtualDub is used to create the completed animations from the dynamic frames and then compiling them into an AVI file. Start Develop system dynamics equations Create animations for system responses for physical systems Add radio button for new system in Visual C++ GUI. Develop Simulink model for system based on dynamic equations Display system response Add method to new radio button for user input End Create models for different system inputs Implement controllers on system model Link GUI inputs to Simulink models Link animations to MATLAB data Fig. 9 Step-by-step process for incorporating a new model This completes the implementation of user interface and integration with Matlab/Simulink. IX. CONCLUSION A new approach for design and development of MoSART has been presented. The architecture, design features and advantages are illustrated. This environment is ideal for students pursuing a first course in Automatic Control. It introduces them to the behavior of dynamic systems. The animation features provide users a clear distinction between the system in open loop and the system in closed loop conditions as well as behavior of system when different inputs are applied. REFERENCES [1] Bishop R.H., Modern Control Systems Analysis and Design: Using MATLab and Simulink. California: Addison Wesley Longman Incorporated, 1997 [] Casini, M; D. Prattichizzo; and A. Vicino., The Automatic Control Telelab: A User Friendly Interface for Distance Learning. IEEE Transactions on Education 46(): 5-57, 003 [3] Gamino, M; J.C. Pedraza; J.M. Ramos; and E. Gorrostieta, Matlab C++ Interface for a Flexible Arm Manipulator Simulation Using Multi- Language Techniques, IEEEProc. Fifth Mexican International Conference on Artificial Intelligence Mexico City, Mexico, 006 [4] Horton, I Ivor Horton s Beginning Visual C++, Wiley Publishing, Indiana 010 [5] Johansson M.; M. Gafvert; and K. J. Astrom, Interactive Tools for Education in Automatic Control. IEEE Control Systems 18(3): 33-44, 1998 [6] Lim C.I An Interactive Modelling, Simulation Animation and Real Time Control (MoSART) Helicopter Environment: A Tool for Enhancing Education and Research. IEEE Proc. International conference Wescon, pp Anaheim, California, USA, 1998 [7] Rios, C; C. Lim; R.P. Metzger; and A.A. Rodriguez, Multivariable Analysis and Control of a Cart-Pendulum-Seesaw system using an Animation Tool. In Proc. 38 th IEEE Conference on Decision and Control Phoenix, Arizona, USA, [8] Rodriguez, A. A.; R.P. Metzger Jr.; O. Cifadaloz; and T Dhirasakdanon, Description of a Modelling, Simulation, Animation and Real Time Control (MoSART) Environment for a Class of Electromechanical Systems. IEEE Transactions on Education 48(3), pp , 005 [9] Rodriguez, A. A. and R.P. Metzger Jr, An Interactive Flexible Inverted Pendulum Modelling, Simulation, Animation, and Real Time control Environment for Enhancing Control Design. In Proc. American Control Conference Denver, Colorado, USA, 003 [10] Rodriguez, A. A.; O. Cifadaloz; M. Phielipp; J. Dickeson; P. Koziol, D. Miles; M. Garcia; R. McCullen; J. Willis; and J. Benavides Description of a Modelling, Simulation Animation and Real Time Control (MoSART) Environment for a Broad Class of Dynamical Systems. In Proc. IEEE Conference on Decision and Control San Diego, California, USA, 006 Musti K.S. Sastry (M 04 SM 07) is currently with the Department of Electrical and Computer Engineering, University of the West Indies, St Augustine Campus, Trinidad and Tobago. He obtained his doctoral degree from National Institute of Technology, Warangal, India in Electrical Engineering. He is a senior member, IEEE, a member of the Institution of Engineering and Technology (IET), UK and Institution of Engineers, India. His research interests include, but not limited to Electrical Power, Control Systems and Computer Applications in Engineering. Richard Balgobin is currently with Department of Electrical and Computer Engineering. His interests include Modeling of dynamic systems and Development of user interactive computer simulations. He is a student member, IEEE since