Experimental Modular Platform for the Fuzzy Control of the Temperature Inside an Enclosure

Dan Mihai*, Costel Caramida**

*University of Craiova, Romania, ** ICMET Craiova E-mail dmihai@em.ucv.ro

Abstract - The paper deals with an experimental hardware platform designed in a modular way in order to implement and test different control algorithms for the temperature inside a thermal enclosure. For a fast and efficient design, most of the acquisition, control and peripheral boards are taken from a modular ready to use family. In order to exploit all the software and hardware facilities support, some extra boards were designed. The enclosure uses a thermoelectric element that is able to realize the heating or cooling by means of a special power supply, described also in the paper. The functionality aspects and the quality results are proved by the on-line recordings.

I. INTRODUCTION

The preoccupations to realize different kind of high quality thermal enclosures is nowadays often present in many fields. Ref. [5] makes a study on the design and construction of a small such enclosure (about 1 meter cube) suitable to perform measurements at temperature extremes. In the opposite dimensional side, in ref. [3] is describes the development, design and fabrication of a large thermal enclosure system, which is used to perform thermal testing on a variety of spacecraft systems and subsystems. The application area is much larger. Ref. [7] deals with the temperature control inside a building where the multilevel structure, the distributed network and the random perturbations makes quite difficult the control tasks. Ref. [6] concerns the temperature control for the crops drying. In reference [9] is described an application of the temperature control for a coke oven in metallurgy.

After the power part is well established or build, the next main aspect concerns the control algorithm to be implemented. Reference [10] gives an interpretation for the intelligent control, not necessarily associated with artificial intelligence but focused on robust control, advanced control algorithms integrated into a hierarchical pyramidal and distributed structure.

More and more researches are oriented to the use of intelligent algorithms like fuzzy and genetic algorithms. The fuzzy control became lately an obvious option, in simple but efficient forms, as in [9] or as a hybrid control. Reference [1] describes a new methodology for the systematic design of fuzzy PID controllers based on theoretical fuzzy analysis and genetic-based optimization. Fuzzy PID controller produces superior control performance than the conventional PID controllers, particularly in handling nonlinearities due to time

delay and saturation. The paper [4] makes an analyze of the possibilities for an intelligent control of the temperature inside a building with the integration of a fuzzy control into a multilevel control / monitoring structure.

Often cited as an important advantage (but not the only one – [8]), the robust property of a fuzzy controller makes this choice successful in many applications where the non-linearities, strong perturbations and the lack of an analytical model or tool does not allow the use of the classical algorithms.

A subsequent and useful gain of the work is the pedagogical tool obtained having a versatile system for controlling the temperature. It brings a strong and convincing support in teaching embedded systems, for the design and optimization of the control algorithm of any kind. Even in an era where the virtual teaching and experiment becomes very popular, the integration of physical experiment remains a valuable option – [2], [11], [12], [13]. For the training of the students in electrical engineering, the temperature control will be always a valuable support in the understanding, designing and experiencing both hardware and software knowledge and tools. So, the main objective of the work is related to the necessity for having a modular experimental platform for the study of different classic or unconventional control algorithm for temperature inside an enclosure.

II. THE HARDWARE SUPPORT

The hardware design is in connection with a systemic and software stage (not included here) and implies the next steps: - the choice of a modular system and of a control processor. The author preferred the very popular (and successful) PIC family, the PIC16F877 microcontroller (MCU) – [17] and the E-blocks system containing all typical boards - [16]. - the configuration of the system – fig.1, with a main processor(s) board, data acquisition boards, peripheral boards, power supply units; - the choice of the appropriate drivers; - the identification of all physical connections between modules / circuits, accordingly to the MCU ports and to the existing or new designed interfacing / processing boards; - the transfer of the program into the MCU flash memory; The main parts of the platform are: - the PIC micro multiprogrammer board, with the application microcontroller unit and a second unit for programming through an USB port connected to a notebook;

978-1-4244-7300-7/10/$26.00 ©2010 IEEE 367

- the Sensor board, making the connection of the analog signals (from the temperature sensor board and several potentiometers) to the input A port of the microcontroller; - the Control board, for the access of all operator controls (binary and analog) to the input port of the MCU; - the Digital – analog conversion board, that makes a D/A special conversion, compatible with the power supply unit; - the Power supply unit for heating / cooling - specially designed for this application (with both output polarities); - the Regulated power unit for the control boards; - the Graphic display board, able to display 4 colored diagrams for the main variables; - the LCD display board (2x16 char.), programmed to display many alphanumerical data (temperatures, controls, text); - the LED Test board (9 bits), indicating many binary states – including software markers for the main real-time tasks.

The thermal unit uses a thermoelectric cooler (Peltier element) – [20], between 2 heat sinks with air forced cooling. The temperature probe – [19], allows a larger temperature field [-40°C, +135°C] than the designed temperature interval

for the enclosure: [18°C, 50°C]. The power supply unit can deliver 5.3 A in the range [-12, +12] V DC. More details are revealed by the fig. 2 and the real electric circuits for the power part (enclosure, power supply unit) are given by fig. 3. The enclosure has inside a low power incandescent lamp for an intuitive image of the instantaneous electric power, some red / blue LEDs for signalling the heating / cooling regimes and protection switching elements for each main circuit. The construction was designed with the possibility of a thermal perturbation activation by opening gradually a window. The power supply unit was designed for an independent operation too.

III. THE ON-LINE COMPUTATIONS FOR THE CONTROL ALGORITHM

The control algorithm must deliver the control variable as function of the temperature error (in CPU units):

kAτUkτ*Ukτε −= (1)

Fig. 1 The structure of the experimental platform.

368

Because the reverse dependence of the temperature sensor,

a new variable – Uτ A k was added so that the control algorithm could be convergent. This variable is computed combining the dependences from fig. 4a and b.

) mτ

Ukτ

(UMm

ΔUMmΔ

Mk −⋅−=τ

τττ (2)

Fig. 2 The configuration of the main Input / Output subsystem.

Fig. 3 The power circuits.

369

) mτk(τMmΔτA MmτΔU

mAτUkAτU −⋅+= (3)

Hence, finally, a simple relation:

kτUmτUMτUkAτU −+= (4)

The MCU EEPROM memory is used for delivering the

control values (pre-computed and stored) and to record in real-time the main variable of the systems – the temperature error ε τ.

The control value is taken from a Look-Up-Table (LUT) filled with the results of a fuzzy control algorithm as function of the error ετ k:

ck = EEPROM (Addr. k) (5)

MτεkτεkAddr ++= 1 (6)

ADCrefV

MmADCΔU)ADCn

(Mτε ⋅−= 12 (7)

For a symmetrical fuzzy control surface, only a half LUT memory is necessary and:

)kτεEEPROM()kτ(εkc ⋅= sgn (8)

For the on-line savings in EEPROM of the temperature error, the space available is divided as follows: - [0,… , A FLUT] for the storage of the off-line computed fuzzy control; - [A FLUT+1, …, Amax] for the storage of the ε τ k.

At each Is ⋅T moment (Is means a saving index), a saving is performed to an address incremented each time. The maximum duration for such recording is:

Δtmax = (Amax - AFLUT) ⋅ Is ⋅T [s] (9)

IV. THE EXPERIMENTAL RESULTS

An overall image of the experimental platform is presented by the fig. 5. The LCD board displays – fig. 6, mainly the set-up temperature and the real temperature inside the enclosure. The numbers from the right side concern the frame index related to the sampling period; the image says that this is the first frame and the controls (computed and outputted) will be displayed by the 6th. The first image offered by the graphical board displays a cycle with an initial temperature of 31.5 oC, followed by a cooling to 26.1 oC and a heating to 33 oC. The next photo certifies a good entry to the steady state regime. A valuable certification of a right real-time operation was obtained recording the main MCU tasks – fig. 7. The channels notation: SAMPL – the sampling period; ACQUI – acquisition; T/CAL – temperature computation; T/LCD – displaying time for the temperatures; CONTR – the control computation; FUZZ – the fuzzy control delivering; C/LCD - displaying time for the controls; PLOTS – time for updating the graphical display. The most consuming time are the tasks for the LCD unit; that is why the program was designed for displaying the controls (computed and outputted) only from time to time (1/10 sampling periods) – see fig. 7 a and 7b, the last one requiring about 50 % of the control cycle. The fuzzy task, based on a LUT (EEPROM) algorithm is very fast (62 μs – fig. 7c). Because the robustness of the fuzzy control related to the sampling period, the system is able to offer a good control for a large range of the sampling period, from hundreds of milliseconds – fig. 7e to several seconds (and more) – fig. 7d. Such recordings are useful both for having a real confirmation of the right timing as designed by the program and for a precise quantitative evaluation of the time intervals.

V. CONCLUSIONS

The paper reveals the results of a design chain for the systemic, software and hardware parts. The author considers a powerful graphical language for the control program (Flowcode), with many facilities in exploiting the hardware and software abilities of the target microcontroller integrated into an easy to use modular hardware. The use of this kind of hardware / software support leads to a fast, efficient and not expensive way to implement a fuzzy control algorithm. The experimental platform validates entirely the design requirements and its behavior is very closed to the virtual operation under the graphical programming language. Several useful computation formulas were proposed and they have been confirmed by the real-time results. The platform is intended to be an open system for new functions, operation modes and applications.

REFERENCES

[1] B. Hu, G.K.I. Mann, R. G. Gosine, “New methodology for analytical and optimal design of fuzzy PID controllers, IEEE Trans. on Fuzzy Systems, VOL. 7, NO. 5, OCTOBER 1999, pp. 521-539.

Fig. 4 The sensor characteristic and the needed dependence for thealgorithm convergence.

a. b.

370

[2] D. Mihai, “Virtual vs. experiment, programmable vs. wired logic, hardware vs. software in teaching digital control for electromechanical engineering”, Proc., IEEE EUROCON - Computer as a Tool, Ljubljana, Slovenia, 2003, Vol. II, pp 7-12.

[3] D.A. Shearer, “Development, design and construction of a large thermal enclosure”, Depart. E8-10, Test Services Eng., Lockheed Martin Missiles & Space, Sunnyvale, California, 1999.

[4] Iv. Polanecka, T. Saloky, “Intelligent control of heating processes”, Depart. of Autom. and Control, Faculty of Mech. Eng., Tech. Univ. Kosice, Slovakia, 2005.

[5] J. G. Dumoulin, P. Charron, E. Chouiery, S. Mishra, “A study of appropriate materials to construct a small enclosure to be used for passive intermodulation measurements at temperature extremes (Thermal PIM), 4th Int. Workshop on Multifactor, Corona and Passive Intermod. in Space RF Hardware, 8-11 Sept. 2003, ESTEC, Noordwijk, The Netherlands.

[6] K.N. Magdi, E.R.M. Hamdi, A.I. Awad, A. A. El-Hamid, “Environmental control for plants using intelligent control systems, Fifth Int. Workshop on Artif. Intel. in Agric., IFAC, Cairo, Egypt, 2004, pp. 101-106.

[7] K. Passino, J. Finke, “Hierarchical and distributed building temperature control”, The Ohio State Univ., Depart. of Electr. Eng., 2002.

[8] K. Seungwoo, C. Youngwan, P. Mignon, “Multirule-base controller using the robust property of a fuzzy controller and its design”, IEEE Trans. on Fuzzy Systems, 1996, pp. 315-327.

[9] L. Gongfa et al., “Hybrid intelligent control of coke oven”, Grant No.20063002066, Sci. and Tech. Found. of Hubei Prov. China, College of Machinery and Automation, Wuhan Univ. of Sci. and Techn., Hubei, China, 2007.

[10] L. Smutny, “Intelligent measurement, diagnostic”, Acta Montanistica Slovaca, 8 / 2003, vol. 4, Czech Rep., pp. 157-159.

[11] R.van de Molengraft, M. Steinbuch, B. de Kraker, “Integrating Experimentation into Control Courses”, IEEE Contr. Sys. Mag., Febr. 2005

[12] S. Bex et al., “The people mover educational project, design, simulation, and implementation of a “real” control system application”, IEEE Contr. Sys. Mag., Oct., 2004.

Fig. 6 Some results given by the LCD display and by the graphical display.

Fig. 5 The experimental platform in operation: the overall image and the control boards.

371

[13] S. Nooshabadi, J. Garside, “Modernization of teaching in embedded systems design - an international col. project, IEEE Trans. Educ., vol. 49, no. 2, May 2006.

[14] W. Wolf,, Computers as Components: Principles of Embedded Computing Systems, San Francisco, CA, Morgan Kaufman, 2001.

[15] ***, Drivers and software, CDRom ELSAM, Matrix Multimedia, London, 2008.

[16] ***, E-blocks Solutions, Matrix Multimedia, London, 2006, http://www.matrixmultimedia.com/eblocks /index. php.

[17] ***, PIC16F87XA, Data Sheet DS39582B, Microchip Technology Inc., 2003.

[18] ***, RC12-8 Single-Stage Thermoelectric Cooler, Technical Data Sheet, Doc. #102-0279 Rev. B, Marlow Industries Inc, Dallas.

[19] ***, Stainless Steel Temperature Probe TMP-BTA, Vernier Software & Technology, Beaverton, OR.

[20] ***, Thermoelectric Cooling Systems Design Guide, Marlow Industries Inc, Dallas, 1998.

a.

b.

c.

d.

e.Fig. 7 The real-time tasks revealed by the logic analyzer recordings in different operating conditions.

372