IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO.

1, FEBRUARY 2002

43

Development and Control of a Prototype Pneumatic Active Suspension System

Winfred K. N. Anakwa, Member, IEEE, Dion R. Thomas, Scott C. Jones, Jon Bush, Dale Green, George W. Anglin, Ron Rio, Jixiang Sheng, Scott Garrett, and Li Chen

AbstractReal physical plants for control experimentation are valuable tools in a control laboratory. This paper describes a prototype pneumatic active suspension system, which was designed and built over a number of years as a sequence of student projects. The physical plant, which models a quarter-car suspension, consists of a wheel, coil springs, a pneumatic actuator for active damping, position, and velocity sensors, and an alternating current (ac) motor for simulating road disturbance input signal. An electronic subsystem is used to process the sensor signals which are sent to a Motorola 68HC16 microcontroller-based evaluation board. The microcontroller controls a four-bit automatic binary regulator that controls airflow to the pneumatic actuator for damping. A mathematical model of the suspension system was derived analytically and validated experimentally. MATLAB and Simulink (MathWorks, Inc., Natick, MA 01760 USA) were used to analyze and design a digital state feedback plus integral controller for the system. The digital controller was implemented on a Motorola 68HC16 microcontroller. The controller was able to reject a physically generated 0.01143-m negative step road disturbance input. The details of the design construction, modeling, analysis, computer simulation, controller implementation, and experimental results are presented. Index TermsActive suspension, microcontroller, physical plants, pneumatic actuator, sensor electronics.
Fig. 1. Diagram of suspension system.

I. INTRODUCTION N ORDER for students to gain experience in design and construction of a physical plant, as well as control experimentation, a decision was made in fall 1990 to build a prototype pneumatic active suspension system as a student project. The suspension system was selected, at that time, because the automotive industry was seriously considering implementation of active suspension systems on automobiles [1][8]. A quarter-car model was selected because it would be feasible to test in a laboratory environment. Pneumatic actuation was
Manuscript received November 2, 2000; revised October 1, 2001. This work was supported in part by the Department of Electrical and Computer Engineering of Bradley University. W. K. N. Anakwa is with the Department of Electrical and Computer Engineering, Bradley University, Peoria, IL 61625 USA. D. R. Thomas is with Raytheon Systems Company, El Segundo, CA 90245 USA. S. C. Jones is with Siemens Energy and Automation, Alpharetta, GA 30005 USA. J. Bush is with JB Electric and Communications, Morrison, IL 61270 USA. D. Green is with Midwest Generation Edison Mission Energy, L.L.C., Pekin, IL 61444 USA. G. W. Anglin is with Daramic, Inc., Owensboro, KY 42303 USA. R. Rio is with PWR, L.L.C., Syracuse, NY 13210 USA. J. Sheng is with Thomson Multimedia, Inc., Indianapolis, IN 46201 USA. S. Garrett is with Caterpillar, Inc., Peoria, IL 61656 USA. L. Chen is with Corvis Corporation, Columbia, MD 21046 USA. Publisher Item Identifier S 0018-9359(02)01472-3.

chosen instead of hydraulic actuation to avoid the possibility of oil spill in the laboratory. The design, the construction, and the physical structure are presented in Section II, and the associated sensors and actuator electronics are covered in Section III. The mathematical modeling, computer simulation, and experimental validation of the model are presented in Section IV. The design of the digital control algorithm, its implementation on a Motorola 68HC16 microcontroller-based evaluation board, and experimental results are given in Section V. Final conclusions are summarized in Section VI. II. DESIGN AND PHYSICAL STRUCTURE A typical quarter-car model of an active suspension can be modeled as a system consisting of masses, springs, a passive damper, and an active damper, as shown in Fig. 1, where mass of driver and seat; mass of wheel and tire; spring constant of coil of shock absorber; damping coefficient of hydraulic damper spring constant of tire; active control input force; road disturbance input displacement of ; displacement of .

00189359/02$17.00 2002 IEEE

44

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 1, FEBRUARY 2002

Some standard values taken from [9] and [10] are kg, kg, 1 N/m, N/(m/s), N/m. and The objective was to design and construct a laboratory scale quarter-car pneumatic active suspension system with the folkg; kg; lowing specifications: N/m; N/(m/s); N/m. The physical system was an assembly of a sprung mass, two linear springs, pneumatic cylinder and valving, and position and velocity sensors. A three-phase inverter-controlled alternating current (ac) gear motor driving a cam was included for generation of a sinusoidal road surface disturbance input. The signals from the sensors were processed by a microcomputer which controlled the pneumatic damper. The engineering analysis and calculations leading to the selection of system components are discussed next. The mechanism for road disturbance input was developed with the goal of driving the suspension system in the vertical direction. It was found that a circular cam with an offset bore would produce a very good approximation of a sinusoidal waveform when driving a load placed upon its surface. The cam had 0.0254-m stroke and it generated a waveform modeled as (1) was a function of the speed of rotation of the where cam. Next, a motor was selected to drive the cam. The motor rating was determined using the following assumprad/s; maximum load tions: maximum value of kg; generated waveform ; mass m/s; maximum maximum velocity m/s ; maximum force acceleration N; power equation is power W; and horse power hp. A 0.5-hp Dayton three-phase ac gear motor (Dayton Electric Manufacturing Company, Niles, IL 60712 USA), which has rated full-load torque of 11.30 N-m and no-load speed of 288 rpm, was selected. A motor speed controller was needed to provide a mechanism for varying the frequency of the road disturbance input signal. The Toshiba ESP-130 transistor inverter VF 10S-20 15BO (Toshiba International Corporation, Houston, TX 77041 USA), equipped with a frequency meter, an adjustable dial, and a RUN-STOP switch was chosen for motor speed control. Fig. 2 shows the wiring diagram of the inverter connected to the ac motor. The inverter made it possible to attain a low-frequency road m. The analysis disturbance input signal leading to the selection of an air cylinder for active damping is discussed next. The sizing of the air cylinder was determined, assuming that the total load consisted of any frictional loads in the system, the force exerted by maximum sprung mass, the force needed to accelerate the sprung mass, and the force required to exhaust the air from the other end of the cylinder. The kg m/s loads were: force due to sprung mass N; acceleration force kg m/s 280 N; fricN; and total load N. tional and exhaust forces

Fig. 2.

Gear motor and inverter wiring schematic.

After calculating the total load, a working pressure of 2.76 10 Pa was chosen to size the cylinder. Using the equations force and (3) a minimum piston area of 0.002 62 m was obtained. This yielded a minimum cylinder bore diameter of approximately 0.058 m. Appropriate cushions were selected for the air cylinder to avoid damage due to shock. Because the cam produced a total stroke of 0.0254 m, Parker Fluidpower air cylinder model 3.00 with 0.0762 m stroke was selected 2.50CTCMA14C (Parker Hannifin Corporation, Richland, MI 49083 USA). For control of airflow to the air cylinder, Parker Fluidpower four-bit automatic binary regulator (ABR) model W21 540 179P, and Parker Fluidpower three unit air line regulator, filter, lubricator were selected. Additional components selected before assembling the active suspension system were: two chrome vanadium springs with spring constant 5812.425 N/m; two stainless steel guide rods and bearing; two Bourns linear potentiometer position sensors (Bourns, Inc., Riverside, CA 92507 USA); two magnetic linear velocity sensors with 1.772 V/m/s; one cam follower with spring constant 11 207.56 N/m; Parker Fluidpower air line hoses; and Parker Fluidpower air line couplings. The cam was designed by a student, and the design was sent to Kress Corporation (Brimfield, IL 61517 USA) for fabrication [11]. The physical structure of the active suspension system resulting from assembling the components is shown in Fig. 3. A photograph of the active suspension system is shown in Fig. 4. The dimensions of the active suspension system are 0.953-m high, 0.61-m wide, and 0.61-m deep. An air compressor supplying air at 0.552 MPa and two air storage tanks were used to maintain the air pressure at the inlet of the four-bit regulator at approximately 0.276 MPa during operation. III. SENSOR AND ACTUATOR ELECTRONICS The sprung mass position and wheel mass position, shown in Fig. 1, were measured by identical linear Bourns potentiometers. Two identical amplifier and filter combination circuits were used to process the measured position signals. Fig. 5 shows the amplifier and filter circuit. The circuit in Fig. 5 was calibrated to produce 81.10 V/m. The velocities of the sprung mass and wheel mass were measured pressure area (2)

ANAKWA et al.: PROTOTYPE PNEUMATIC ACTIVE SUSPENSION SYSTEM

45

Fig. 6.

Velocity sensor amplifier and filter circuit.

Fig. 3.

Physical structure of active suspension system.

Fig. 7. Control circuit for automatic binary regulator.

IV. MATHEMATICAL MODELING Suppressing the argument , the equations of motion for the active suspension system of Fig. 1 can be written as (4) and
Fig. 4. Photograph of active suspension system.

(5) When 0, the transfer function from to reduces to (6) where . When the active suspension system was excited with a sinusoidal input disturbance , it was observed that the amplitudes of and were equal. This indicated that the tire was not comwas too large. Consequently, the limiting pressed because approached infinity, was calculated. As form of (6), as approached infinity, (6) reduced to (7) Using numerical values of N/m led to and Kg, N/m/s,

Fig. 5. Position sensor amplifier and filter circuit.

by identical magnetic linear velocity sensors. Two identical amplifier and filter combination circuits were used to process the measured velocity signals. Fig. 6 shows the amplifier and filter circuit. The circuit in Fig. 6 was calibrated to produce 7.5 V/m/s. In Section V, the microcontroller that used the position and velocity signals to control the pneumatic damper will be discussed. In the meantime, the circuit diagram of one of the four identical electronic circuits which control the four-bit binary regulator valves is shown in Fig. 7. The optoisolator in Fig. 7 was used to isolate the microcontroller from the high voltage circuit. The inductor in the circuit of Fig. 7 represents one of the solenoids of the four-bit binary regulator.

(8) To validate the mathematical model given by (8), the response of the physical active suspension system to a negative disturbance step input of amplitude 0.011 43 m was recorded. Equation

46

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 1, FEBRUARY 2002

Fig. 8. Experimental and MATLAB simulation step responses.

(8) was simulated in Simulink (MathWorks, Inc., Natick, MA 01760 USA) with a scale factor of 81.10 V/m to convert position signal to volts [12]. The physical system response and computer simulation response curves are shown in Fig. 8. In Fig. 8, the noisy curve is the experimental step response, and the smooth curve is the computer simulation step response. The agreement between the two curves showed that (8) modeled the behavior of the suspension system, without the actuator force , fairly accurately. A separate experiment was performed to obtain a model for the combination of the four-bit automatic binary regulator and the air cylinder, which generated the control force . In the experiment, the air pressure at the inlet of the automatic binary regulator was set at 0.2484 MPa. A computer program running on the Motorola 68HC16 opened the automatic binary regulator valve in increments of one bit from zero to 15, while the corwere recorded. From the experiresponding waveforms of mental results, the transfer function from the binary regulator can be expressed as valve input to control force (9) has units of Newtons. The control force Fig. 9 shows a simulation diagram of the suspension system with road disturbance input , binary regulator input , and output . MATLAB (MathWorks, Inc., Natick, MA 01760 USA) simulation and experimental step responses of the suspension system 0, are from ABR input to , when the disturbance input shown in Fig. 10. The upper curve is MATLAB simulation, and the lower curve is the experimental result. Because the experimental results agreed fairly well with computer simulation results, the mathematical model shown in the simulation diagram of Fig. 9 was used to design a controller for the pneumatic suspension system. V. CONTROLLER DESIGN AND IMPLEMENTATION There are several control algorithms that can be implemented to control the active suspension system. Some examples are gain-scheduled adaptive schemes and robust -infinity disturbance rejection controllers [16][18]. However, a preliminary

Fig. 9.

Simulation diagram of suspension system.

Fig. 10.

MATLAB simulation and experimental step responses.

state variable feedback plus integral controller was designed to reject road disturbance input.

ANAKWA et al.: PROTOTYPE PNEUMATIC ACTIVE SUSPENSION SYSTEM

47

The state variable equation of the open-loop system shown in Fig. 9 was written as

(10) where

Fig. 11. loop.

Response of system to a negative step input with no controller in the

[15]. The corresponding gains for the discrete time controller became

(13) with the corresponding closed loop system poles in the at and where is the measured sprung mass displacement; is the road disturbance input; is the input to the automatic kg; N/m/s; and binary regulator; N/m. The objective was to design a state variable feedback plus integral controller to reject road disturbance input , maintain at a desired reference value, and keep at zero after a transient period of 1 s [13]. After extensive MATLAB-Simulink computer simulations and some laboratory experimentation, the pole placement design method was selected. The following desired closed loop poles in the plane were chosen: plane

If is the sampling instant and the discrete time controller equation can be written as

, then

(14) An assembly language program was written to implement and were (14). First, desired reference values of selected and stored on the microprocessor. An initial command signal from the microcontroller to the automatic binary reguand . After that, the road disturlator set and from the bance input was applied and the signals position and velocity sensors were converted to digital form by 10-bit analog-to-digital converters. The microcontroller solved (14) to reject the road disturbance input . In order to prevent oscillation around the desired reference , a dead band error of 0.0018 m was included in the value of and was within control program. If the error between the dead band, no control action was taken by the microcontroller. (The assembly language program, which is five pages long, can be obtained from Dr. Anakwa at wa@bradley.edu.) One preliminary experimental result is now presented. The bottom of the suspension system was raised 0.011 43 m and dropped. This represented a negative road disturbance step input . Fig. 11 shows the response of the system with no control. and the lower In Fig. 11, the upper curve is displacement curve is velocity . Fig. 12 shows the response of the system with the microcontroller in the loop.

Using the poles for pole placement design resulted in a control law

(11) where

(12) is the desired reference value of , and and where is the desired reference value of . s was selected to convert A sampling interval the continuous time controller to a discrete time controller for implementation on the Motorola 68HC16 microcontroller [14],

48

IEEE TRANSACTIONS ON EDUCATION, VOL. 45, NO. 1, FEBRUARY 2002

Fig. 12. Response of system to a negative step input with microcontroller in the loop.

In Fig. 12, the upper curve is displacement , and the lower curve is velocity . The response curves in Fig. 12 show that the system was able to recover from the road disturbance input in approximately 750 ms. The system response can be made faster by tuning the controller gains. Additional experimental work is in progress to tune the controller gains and apply sinusoidal and square wave disturbance inputs to the system. Future projects will include implementation of gain scheduling, nonlinear and robust control algorithms [16][18]. VI. CONCLUSION This paper has described a prototype pneumatic active suspension system, which was designed and built over a number of years as a sequence of student design projects. During the design phase, the students gained experience in selection and integration of system components using knowledge of electromechanics, electronics, pneumatics, sensors, actuators, and microcontrollers. The prototype active suspension system is a useful tool for analytical modeling and experimental modeling from inputoutput data. After a student has developed a model and validated it experimentally, different control algorithms can be designed and implemented on the Motorola 68HC16 microcontroller. In addition to control experimentation, students can use the system to learn assembly language programming and hardware interfacing using the eight analog to digital converters and the port of the Motorola 68HC16 microcontroller. The prototype pneumatic active suspension system is a fairly complex physical plant that makes it possible for students to perform industrial-level control experimentation. ACKNOWLEDGMENT The authors would like to thank Kress Corporation for providing free engineering services; E. Anglin for the use of his compressor; D. Miller, C. Mattus and S. Gutschlag for their technical assistance; and K. Polen for preparing the manuscript. REFERENCES
[1] D. Cho and J. K. Hedrick, Pneumatic actuators for vehicle active suspension applications, ASME J. Dyn. Syst. Measurement Control, vol. 107, pp. 6772, Mar. 1985.

[2] C. Yue, T. Butsuen, and J. K. Hedrick, Alternative control laws for automotive active suspensions, ASME J. Dyn. Syst. Measurement Control, vol. 111, pp. 286291, June 1989. [3] K. C. Cheok, N.-K. Loh, H. D. Mcgee, and T. F. Petit, Optimal modelfollowing suspension with microcomputerized damping, IEEE Trans. Ind. Electron., vol. IE-32, pp. 364371, Nov. 1985. [4] K. C. Cheok and M. Sunwoo, An application of explicit self-tuning controller to vehicle active suspension systems, in Proc. 29th Conf. Decision Control, Dec. 1990, pp. 22512257. [5] D. Hrovat, Optimal active suspension structures for quarter-car vehicle models, Automatica, vol. 26, no. 5, pp. 845860, 1990. [6] T. Aburaya, M. Kawanishi, H. Kondo, and T. Hamada, Development of an electronic control system for active suspension, in Proc. 29th Conf. Decision Control, Dec. 1990, pp. 22202225. [7] T. Tseng and D. Hrovat, Some characteristics of optimal vehicle suspensions based on quarter-car models, in Proc. 29th Conf. Decision Control, Dec. 1990, pp. 22322237. [8] M. Satoh, N. Fukushima, Y. Akatsu, I. Fujimura, and K. Fukuyama, An active suspension employing an electrohydraulic pressure control system, in Proc. 29th Conf. Decision Control, Dec. 1990, pp. 22262231. [9] A. Alleyne and J. K. Hedrick, Nonlinear control of a quarter car active suspension, in Proc. 1992 Amer. Control Conf., pp. 2125. [10] A. Alleyne, P. D. Meuhaus, and J. K. Hedrick, Application of nonlinear control theory to electronically controlled suspensions, Veh. Syst. Dyn., vol. 22, pp. 309320, 1993. [11] Kress Corporation, Brimfield, IL. [12] MATLAB Version 5.2, Mathworks, Inc., Natick, MA, 1999. [13] G. Franklin, J. D. Powell, and M. Workman, Digital Control of Dynamic Systems. Reading, MA: Addison-Wesley, 1998. [14] CPU16 Reference Manual, Motorola, Inc, Phoenix, AZ, 1991. [15] M68HC16Z1EVB Users Manual. Phoenix, AZ: Motorola, Inc, 1991. [16] J.-S. Lin and I. Kanellakopoulos, Nonlinear design of active suspensions, IEEE Control Syst. Mag., pp. 4559, June 1997. [17] J. E. Bobrow and F. Jabbari, Adaptive pneumatic force actuation and position control, ASME J. Dyn. Syst.Measurement Control, vol. 113, pp. 267272, June 1999. [18] M. Yamashita, K. Fujimori, C. Uhlik, R. Kawatani, and H. Kimura, H control of an automotive active suspension, in Proc. 29th Conf. Decision Control, 1990, pp. 22442250.

Winfred K. N. Anakwa (S65M67) received the B.Sc., M.Sc., and Ph.D. degrees in electrical engineering from Brown University, Providence, RI, in 1967, 1969, and 1972, respectively. He has held faculty appointments at Brown University, University of Science and Technology, Kumasi, Ghana, the Ghana Atomic Energy Commission, Pennsylvania State University, University Park, and Concordia University, Montreal, QC, Canada. He is currently a Professor of Electrical and Computer Engineering at Bradley University, Peoria, IL. His research interests include real-time system identification, self-tuning control using microcontrollers, robust control of multivariable systems, rapid prototyping of control algorithms using embedded system development tools, and development of prototype physical plant models for control experimentation. During his sabbatical leave in 1995, he worked on application of advanced multivariable control theory concepts to electronic management systems for diesel engines at Caterpillar, Inc. He teaches continuous-time and discrete-time control theory and design courses, the senior laboratory design course, and coordinates the senior capstone design projects. He has also served as Chairperson of the Illinois Valley Section of the IEEE. Dr. Anakwa is a member of the American Society for Engineering Education (ASEE), ISCA, and the Instrument Society of America (ISA).

Dion R. Thomas received the Bachelors degree in electrical engineering from Bradley University, Peoria, IL, and the Masters degree in electrical engineering from Purdue University, West Lafayette, IN. He is currently employed in the Advanced Programs Systems Engineering Laboratory at Raytheon Systems Company, Electronic System Group, El Segundo, CA, working on the development and deployment of space-based electrooptical surveillance and reconnaissance payloads. His area of specialty is in-line-of-sight pointing and stabilization. He has recently been involved with the development of the next-generation space-based infrared system (SBIRS) architecture.

ANAKWA et al.: PROTOTYPE PNEUMATIC ACTIVE SUSPENSION SYSTEM

49

Scott C. Jones received the B.S. degree (magna cum laude) in electrical engineering from Bradley University, Peoria, IL, in 1991. He is currently with Siemens Energy and Automation, Alpharetta, GA, as a Senior Account Manager, specializing in servicing manufacturers in the power generation industry.

Ron Rio received the B.S. degree in electrical engineering from Bradley University, Peoria, IL, in 1994. Until February 2000, he worked for a telephony services company as an electrical engineer, project manager, and engineering manager. In February 2000, he cofounded PWR, L.L.C., a telephony design and building firm focusing on supporting the communications industry, which was ranked eighth of 100 companies by Entrepreneur magazine in 2001. He oversees all operations including engineering, project management, installation, and testing.

Jon Bush received the B.S. degree in electrical engineering from Bradley University, Peoria, IL, in 1993. He is currently the President of JB Electric and Communications, Morrison, IL, a small electrical and business telephone contractor.

Jixiang Sheng received the B.S. and M.S. degrees in electrical engineering from Shanghai University of Technology, Shanghai, China, and Bradley University, Peoria, IL, in 1988 and 1997, respectively. From 1988 to 1995, he was an Instructor at Shanghai Building Materials Institution, Shanghai, China. He is currently with Thomson Multimedia, Inc., Indianapolis, IN, as an Electrical Engineer.

Dale Green received the B.S. degree in electrical engineering from Bradley University, Peoria, IL, in 1993. He is currently with Midwest Generation EME, L.L.C., Pekin, IL, as an Instrument and Controls Engineer.

Scott Garrett received the B.S. degree in mechanical engineering from Bradley University, Peoria, IL, in 1998. Before attending Bradley University, he was with Computer Age, Inc., in the network support group for six years. He is currently with Caterpillar, Inc., Peoria, IL, as a Development Engineer in the Hydraulic Research Group.

George W. Anglin received the B.S. degree in electrical engineering from Bradley University, Peoria, IL, in 1994. He completed six years in the U.S. Navy as an Electronics Technician prior to college. His first job after college was designing generator controls for various uses around the world. Currently, he is working for Daramic, Inc., Owensboro, KY, as a Project Engineer overseeing large capital projects in their plants worldwide.

Li Chen was born in Beijing, China in 1972. He received the B.S. degree in electrical engineering with a specialty in control systems from Beijing Institute of Light Industry, Beijing, China, in 1995 and the M.S. degree in electrical engineering with a specialty in computer engineering and control systems from Bradley University, Peoria, IL, in 1999. He is currently with Corvis Corporation, Columbia, MD, as an Embedded Software Engineer.