Điều khiển vị trí quả bóng bàn trong ống

0 Position Control of a Ping-Pong Ball Position Control of a Ping-Pong Ball Within a Glass Tube Using a P.I.D Control System James Goddings St No.3131147 11/22/2015 Project Supervisor: Shyamal Mondal Dynamics and System Modelling (ENG_6_451) Table of Contents Aim Apparatus Hardware Software PID Controller Theory Proportional Controller Differential Controller Integral Controller Experimental Procedure Open Loop System Open Loop Results Closed Loop System Proportional Control Proportional Gain Results 10 Integral Control 11 Integral Gain Results 13 Differential Control 14 Integral Gain Results 15 Changing Demand Position 15 Changing Demand Position Results 17 Calibration of Ultrasonic Range Finder Scale 18 Conclusion 20 References Position Control of a Ping-Pong Ball James Goddings 3131147 Aim The aim of this investigation was to achieve stable position control of a Ping-Pong ball using A PID controlled DC fan adjusting airflow through a glass tube This investigation will achieve the objective of familiarising the operator with the following:  PID control systems  Matlab Simulink Models and Real Time Windows Target  Use of Feedback and Feed-Forward Control  Use of the Sensoray 626 Interface Card  Use of the FR04 Ultrasonic Range Finder Apparatus Hardware Figure 1.1 Ping-Pong ball in a glass tube apparatus Position Control of a Ping-Pong Ball James Goddings 3131147 Figure 1.2 Close-up of base of glass tube showing fan impeller assembly, vent and motor controller The hardware consists of: A Ping-Pong ball constrained vertically within a close fitting glass tube, with a fan at the base and a position sensor at the top; as shown in Figure 1.1 The fan is a Mini 9V DC model with an impeller mounted on a common shaft above it, drawing air in a side vent and propelling it upward through the glass tube The position sensor is an FR04 ultrasonic range finder that consists of elements, a transmitter and a receiver Activated by a 50 ms period pulse from one of the counters on a Sensoray 626 interface card, the transmitter emits a 20μs ultrasonic pulse which travels at the speed of sound down the tube, rebounds off the ball, and returns at the speed of sound to the receiver Simultaneous to the ultrasonic pulse, an electrical signal in the form of a 40 kHz series of pulses is output back to the Sensoray 626 interface card When the ultrasonic echo is received by the receiver the electrical output pulses are stopped, therefore the time it takes for the ultrasonic pulse to return (time of flight) can be measured by counting the number of electrical pulses received by the Sensoray 626 interface card (See Figure 1.3) Position Control of a Ping-Pong Ball James Goddings 3131147 Figure 1.3 Graphical representation of the ultrasonic range finder signal input and output traces “The Sensoray model 626 is a Multifunction Input/Output Interface card Some of the features include: • 48 digital I/O channels • 20 of the digital I/O channels have edge detection and interrupt capability • of the digital Outputs can be used as counter overflow outputs • Digital I/O connectors are industry standard pinouts • Watchdog timer with several selectable reset periods that can reset the PCI bus • Six 24 bit up/down counters arranged in pairs with: • Inputs that can be driven in various modes (1x, 2x, 4x) from incremental encoders inputs, the digital inputs, the paired counter’s overflow, the system clock or software driven • Can generate an interrupt on counter overflow or encoder/digital input index • Can be preloaded/cleared on an overflow • Output of second counter can be captured on the overflow of the first • Can be used as a programmable periodic interrupt generator • Counter circuitry can be battery backed up to prevent count loss during a power failure • Charge control of the backup Ni-Cad battery Position Control of a Ping-Pong Ball James Goddings 3131147 • 16 differential analog inputs (14 bit resolution) • analog outputs (13 bit resolution) with remote sense inputs to compensate for any external output resistance.”(Sensoray, 2004) The Sensoray 626 interface card acts as a conduit between the hardware system and the Matlab Simulink software, facilitating the control of the hardware system through models created in Simulink by the operator A single digital input on the interface card is used to count the 40 kHz electrical pulse output from the ultrasonic sensor, this is converted to a voltage output by one of the interface card’s digital to analogue converters and relayed to the Simulink model as a voltage The Simulink model interprets the voltage from the interface card as a position using predetermined range information and returns an output voltage to one of the interface cards analogue inputs, this is digitised through an analogue to digital converter and converted back through a digital to analogue converter to a 0-10V DC output to the motor Software Matlab is a long running mathematical software package originally designed with the intention of facilitating the processing of matrix calculations Over its 30year history it has been developed into a very powerful software package capable performing complex mathematical calculations and of modelling and simulating complex systems Simulink is a powerful plug-in tool developed as part of Matlab to enable graphical modelling of electronic hardware and process controllers within the Matlab environment, with the capability of simulating these models and components and outputting signals to an interface in real time using the Real Time Windows Target PID Controller Theory The PID controller is a tuning controller that is very useful as it does not rely on the operator having prior knowledge of the underlying mathematical model of the system being controlled In effect it combines separate controllers that each implements different mathematical operators on the system in the form of variable gains Proportional Controller The proportional controller’s purpose is to increase the speed of a system’s response and reduce the steady state error of the system response The output of the system is measured and fed back to the input of the system where the error between the output and the demanded input value is multiplied by a gain proportional to the error Position Control of a Ping-Pong Ball χd(t) e(t) + James Goddings 3131147 f(t) Kp - Plant System χ(t) Figure 1.4 Block diagram of a proportional controller applied to a plant system (Satar, 2013) 𝐸𝑞𝑛 1.1 𝑓 (𝑡) = 𝐾𝑝 (𝑥𝑑 (𝑡) − 𝑥(𝑡)) = 𝐾𝑝 𝑒(𝑡) There are downsides to proportional control; it reduces the damping’s effect of the overall system, resulting in the system becoming less stable At a point called the critical gain, the proportional gain overcomes the damping of the system, moving the system poles to the right hand side of the imaginary axis of the s-plane, causing the system to become unstable Proportional gain can also cause overshoot of the demand input, as a rule of thumb, an overshoot of up to 25% is tolerated in return for the increased system response rate In the case of this investigation the proportional gain will act as a proportional multiplier to the error in the Ping-Pong ball’s displacement from the demanded position Differential Controller The differential controller’s purpose is to increase system response stability; this is achieved by adding another feedback loop affecting the derivative (rate of change) of the primary control variable This increases damping, moving the system poles away to the left from the imaginary axis of the s-plane, thereby increasing stability, and is often used in conjunction with proportional control to alleviate the instability introduced by proportional gain Kd χd(t) e(t) + - f(t) Kp + - dχ(t) dt Plant System χ(t) Figure 1.5 Block diagram of a differential controller in conjunction with a proportional controller applied to a plant system (Satar, 2013) 𝐸𝑞𝑛 1.2 𝑓 (𝑡) = 𝐾𝑝 (𝑥𝑑 (𝑡) − 𝑥 (𝑡)) − 𝐾𝑑 𝑥̇ (𝑡) Position Control of a Ping-Pong Ball James Goddings 3131147 In the case of this investigation the differential gain will act as a multiplier to the error in the Ping-Pong ball’s velocity from the demanded velocity (zero for a fixed position), again too much gain will overwhelm the system’s response to the demand input becoming the overriding influence, causing instability Integral Controller The integral controller’s purpose is to decrease the steady state error of the system response to zero This is achieved by integrating the error in the primary control variable between the output and demand and feeding this forward as a ramp up of the control signal This is often used in conjunction with proportional control and differential control to reduce any steady state error to zero Integral control increases the type number of the closed loop system by one, making the model more complex Kd χd(t) e(t) + - f(t) Kp + + dχ(t) dt Plant System χ(t) Ki ∫dt Figure 1.6 Block diagram of an integral controller in conjunction with a differential controller and a proportional controller applied to a plant system (Satar, 2013) 𝐸𝑞𝑛 1.3 𝑓 (𝑡) = 𝐾𝑝 (𝑥𝑑 (𝑡) − 𝑥(𝑡)) + 𝐾𝑖 ∫(𝑥𝑑 (𝑡) − 𝑥 (𝑡)) 𝑑𝑡 − 𝐾𝑑 𝑥̇ (𝑡) In the case of this investigation the integral gain will act as a multiplier to the error in the Ping-Pong ball’s displacement from the demanded position Integral gain is very effective at reducing steady state error when a system has a stable or slow changing system response, however with a dynamic rapidly changing system response the integrator can “wind up” causing instability Position Control of a Ping-Pong Ball James Goddings 3131147 Experimental Procedure Open Loop System The first part of the investigation is a simple test to familiarise the operator with the procedures of building and operating Matlab Simulink models and compiling them for export to the Real Time Windows Target environment for use with the Sensoray 626 Interface Card, the FR04 Ultrasonic Range Finder and the Ping-Pong ball in a glass tube/fan arrangement An existing open loop Simulink model is opened in Matlab which enables a voltage gain value to be applied to the fan controller and the resulting Ping-Pong ball position to be displayed in the form of a value with an arbitrary scale This value is an output from the ultrasonic range finder that had been linearised by the Sensoray 626 interface card to have a range between 0-10V to correspond with the 0-10V scale of the output voltage to the fan (The Ping-Pong balls resting position at the bottom of the tube corresponds to a value of 1.3V) Open Loop Results This resulted in Table 1.1 providing knowledge of voltages required to initiate the Ping-Pong ball climbing the glass tube, hovering stationary and falling in a controlled manner Fan Voltage [V] 6.1 6.85 7.57 7.65 7.5 7.55 P-P Ball Observation Started to vibrate Started to spin Started to lift off Upward movement Controlled downward movement Hovers stationary Table 1.1 Observed Voltages required for initiating movement of the Ping-Pong ball Closed Loop System The second part of the investigation involves designing and implementing a Matlab Simulink PID control model to control the position of the Ping-Pong ball in the glass tube This is done in a methodical manner introducing proportional control, then integral control, then differential control to establish and optimise control of the ball’s position When sufficient control is established, the arbitrary scale resulting from the output signal from the ultrasonic range finder is calibrated against the measuring tape to the right of the glass tube Figure 1.7 Shows the Simulink model used to apply PID control to the Ping-Pong ball in a glass tube system through the interface card Position Control of a Ping-Pong Ball James Goddings 3131147 Figure 1.7 Closed Loop Matlab Simulink model with PID block Proportional Control For the PID investigation, an initial demand position of on the arbitrary range finder scale is selected The proportional controller gain is then increased in steps to determine its effect on the control of the Ping-Pong balls position and the ability of the control system to track the demand position For all of the charts following: Yellow Line = Demand Position Cyan Line = Ping –Pong Ball’s Position Magenta Line = Fan Controller Voltage James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Position/Output Voltage Chart 1.1 Demand Position = 8, Proportional Gain = 1.0, Steady State Position = 1.5, Steady State Output Voltage = 6.7V Time (s) Chart 1.2 Demand Position = 8, Proportional Gain = 2.0, Steady State Position = 4.1(Oscillating 1.2), Steady State Output Voltage = 7.7V (Oscillating 2.2V) James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Chart 1.3 Demand Position = 8, Proportional Gain = 4.0, Steady State Position = 5.8(Oscillating 2.4), Steady State Output Voltage = 9.8V (Oscillating 6V) Proportional Gain Results Charts 1.1-1.3 show that increasing the proportional gain increased the speed of the system response and decreased the error in the steady state response, however it also decreased stability of the system and introduced oscillations at values above 1.5 A proportional gain of 1.2 was chosen to continue the investigation as it maintained stability whilst offering an increase in the speed of the system response and reduction the error of the steady state response 10 Position Control of a Ping-Pong Ball James Goddings 3131147 Position/Output Voltage Integral Control Time (s) Position/Output Voltage Chart 2.1 Demand Position = 8, Proportional Gain = 1.2 Integral Gain = 0, Steady State Position = 1.9, Steady State Output Voltage = 7.7V Time (s) Chart 2.2 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.1, Steady State Position = 7.7, Steady State Output Voltage = 7.7V, Rise Time = 27s 11 James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Position/Output Voltage Chart 2.3 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.2, Steady State Position = 7.9, Steady State Output Voltage = 7.8V, Rise Time = 16s Time (s) Chart 2.4 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.4, Steady State Position = 8.0, Steady State Output Voltage = 7.8V, Rise Time = 7s 12 James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Chart 2.5 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.8, Steady State Position = 8.0(Oscillating 0.6), Steady State Output Voltage = 7.8V, Rise Time = 2.5s, Overshoot =20% Integral Gain Results It is evident from Charts 2.1-2.5 that increasing the integral gain reduces the lead-lag of the system, thereby increasing the speed of the system response and decreasing the error in the steady state response by ramping up the gain in the system more the longer an error between the demand and system response existed; however it also decreased stability of the system and introduced overshoot and oscillations at values above 0.8 An integral gain of 0.8 in conjunction with a proportional gain of 1.2 was chosen to continue the investigation as these values offered a rapid system response and reduction in the steady state error The integral gain did introduce a 20% overshoot, however this is commonly accepted prior to introducing differential control, as differential control can reduce or eliminate overshoot 13 Position Control of a Ping-Pong Ball James Goddings 3131147 Position/Output Voltage Differential Control Time (s) Position/Output Voltage Chart 3.1 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.8, Differential Gain =0.02, Steady State Position = 8.0 (Oscillating 1.2), Steady State Output Voltage = 7.6V, Rise Time = 2.5s, Overshoot =15% Time (s) Chart 3.2 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.8, Differential Gain =0.1, Steady State Position = 8.0 (Oscillating 0.6), Steady State Output Voltage = 7.6V, Rise Time = 3s, Overshoot =12% 14 James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Chart 3.3 Demand Position = 8, Proportional Gain = 1.2, Integral Gain -= 0.02, Differential Gain =0.5, Steady State Position = 8.0 (Oscillating 0.4), Steady State Output Voltage = 7.7V, Rise Time = 4s, Overshoot =2% Integral Gain Results Charts 3.1-3.2 show that increasing the differential gain was reducing the overshoot of the system response, however instability and oscillations were being introduced, increasing the error in the steady state response Upon consultation with the laboratory supervisor, Mehdi Zahir he suggested a review of the integral control procedure and the integral gain final value This review resulted in an optimised system shown by Chart 3.3, where the integral gain had been reduced considerably, and the differential gain increased, producing a system response that was fast, stable and with minimal steady state error Changing Demand Position Having arrived at the PID system in Chart 3.3 it was realised that this system is optimised for a demand position of 8.0, a large difference from the rest position of 1.3, but what if the demand position was changed? A further investigation is embarked upon to ascertain if the systems response would be suitable for all demand positions 15 James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Position/Output Voltage Chart 4.1 Demand Position = 6, Proportional Gain = 1.2, Integral Gain -= 0.02, Differential Gain =0.5, Steady State Position = 6.0, Steady State Output Voltage = 7.8V, Rise Time = 4s, Overshoot =2% Time (s) Chart 4.2 Demand Position = 4, Proportional Gain = 1.2, Integral Gain -= 0.02, Differential Gain =0.5, Steady State Position = 4.0, Steady State Output Voltage = 7.7V, Rise Time = 7s 16 James Goddings 3131147 Position/Output Voltage Position Control of a Ping-Pong Ball Time (s) Position/Output Voltage Chart 4.3 Demand Position = 2, Proportional Gain = 1.2, Integral Gain -= 0.02, Differential Gain =0.5, Steady State Position = 2.0, Steady State Output Voltage = 7.7V, Rise Time = 25s Time (s) Chart 4.4 Demand Position = 2, Proportional Gain = 1.2, Integral Gain -= 0.02, Differential Gain =0.55, Steady State Position = 2.0, Steady State Output Voltage = 7.7V, Rise Time = 21s Changing Demand Position Results It is evident from Charts 4.1-4.3 that decreasing the demand position, and thereby the difference from the rest position of 1.3 to the demand position, resulted in decreasing 17 Position Control of a Ping-Pong Ball James Goddings 3131147 response times for the PID system implemented Adjusting the differential gain in Chart 4.4 improved this slightly, however further investigation would have been required to obtain proportional, integral and differential gains more suitable for a variety of demand positions, and varying demand positions, such as a time dependant cyclic demand instead of a step demand Unfortunately due to the limited time of the lab session this further work lay outside of the scope of the investigation Calibration of Ultrasonic Range Finder Scale When sufficient control was established, the arbitrary scale resulting from the output signal from the ultrasonic range finder was calibrated against the measuring tape to the right of the glass tube This involved using the Simulink model in figure 1.7 with the PID values shown in Chart 4.3 and setting a range of range finder demand values, then reading off the corresponding steady state position of the Ping-Pong ball on the measuring tape to the right of the glass tube This resulted in Table 1.2 Position Demand [V] 1.4 1.5 4.5 7.5 10 Range Finder Output [V] 1.4 1.5 4.5 7.5 10 Measured Position [mm] 1310 1300 1110 890 680 470 260 110 Table 1.2 System response to demand values, and the corresponding steady state position of the Ping-Pong ball on the measuring tape to the right of the glass tube This table was charted, Chart 5.1, and the mathematical relationship evaluated relating the arbitrary scale resulting from the output signal from the ultrasonic range finder to the measuring tape to the right of the glass tube A new Simulink model with a block facilitating the input of a demand value for the position of the Ping-Pong ball in mm was developed, Figure 1.8, and tested This allowed for a demand to be tracked effectively by the Ping-Pong ball upon a user input related to the measuring tape 18 Position Control of a Ping-Pong Ball James Goddings 3131147 Range Finder Output Position [V] 12 10 y = -0.0072x + 10.8465 R² = 0.9997 0 200 400 600 800 1000 1200 1400 Position on Measuring Tape [mm] Chart 5.1 Relationship between the arbitrary scale resulting from the output signal from the ultrasonic range finder and the corresponding steady state position of the Ping-Pong ball on the measuring tape to the right of the glass tube Figure 1.8 Simulink model facilitating the input of a demand value for the position of the Ping-Pong ball in mm corresponding to a position on the measuring tape of the glass tube 19 Position Control of a Ping-Pong Ball James Goddings 3131147 Conclusion The aim of this investigation was to achieve stable position control of a Ping-Pong ball using A PID controlled DC fan adjusting airflow through a glass tube This was achieved and the PID system was tuned to control the system adequately to allow sufficient control of the Ping-Pong ball’s position within the tube to accurately calibrate it to a physical scale The optimum PID control algorithm derived from his investigation was most dependent on proportional and derivative gains, and was most effective at tracking larger changes in positional demand to the system There is still room for improvement of the PID system developed in responding to small demand changes; further testing would consist of investigating this and the systems response to other demand types, i.e cyclic demands The system was very dynamic with little damping (underdamped) due to the main dynamic force acting on the system in the form of turbulent airflow being variable and transient in its nature This results in the system having some poles close to the imaginary axis of the splane, which is evident from the system becoming unstable with the addition of larger proportional and integral gains The other dynamic force, gravity, remains constant and has a predictable effect which enables the system to be controlled as it results in the system having some poles close to, or situated on, the real axis The investigation was successful in achieving the objectives of familiarising the operator with Matlab Simulink Models and Real Time Windows Target, Use of Feedback and FeedForward Control, Use of the Sensoray 626 Interface Card and Use of the FR04 Ultrasonic Range Finder It was especially informative in the use and tuning of PID controllers on dynamic systems, which the author feels he has a much better understanding of as a result of carrying out this work 20 Position Control of a Ping-Pong Ball James Goddings 3131147 References Ashwini Miryala, Kyle Scarlett, Zachary Zell, Brandon Kountz, (2006), The University of Michigan Chemical Engineering Process Dynamics and Controls Open Textbook - PID Downslides & Solutions, available at https://controls.engin.umich.edu/wiki/index.php/PIDDownsides accessed on 14/11/2015 LSBU (2015), London South Bank University School of Engineering –Control Systems Lab Room 405 – PID Position of a Ping Pong Ball in a glass tube Satar Dr T (2013), Principles of Control Lecture Notes - PID Control Sensoray (2004), Sensoray Model 626 PCI Multifunction I/O Board Revision F available at http://www.sensoray.com/downloads/man_626_1.0.5.pdf accessed on 14/11/2015 A-1