The influence of control methods in technological process to save ...

SISY 2013 • IEEE 11th International Symposium on Intelligent Systems and Informatics • September 26-28, 2013, Subotica, Serbia

The Influence of Control Methods in Technological Process to Save Electrical Energy Štefan Koprda, Zoltán Balogh, Milan Turčáni Department of Informatics, Faculty of Natural Sciences, Constantine the Philosopher University in Nitra Nitra, Slovakia [email protected], [email protected], [email protected] propose rule-based solutions for designing fuzzy FPID controllers. It is applied especially in the areas, where it is impossible to make a mathematic description of the control system, or it is very complex and inapplicable for the purpose of control. Due to the fact that fuzzy logic utilizes the simple qualitative expression of heuristic knowledge of the human experts, fuzzy controllers have been proposed for a variety of systems [6] including energy systems. Problems with the modelling of continuous processes are obvious when solving multi-level control systems, where besides the extensive process level of controlling there exists also another level of controlling responsible for the continuous optimization of the process. While fuzzy logic has been used as an effective tool in the development of the F-PID control algorithm, the performance of these controllers is limited to the extent of possible combinations of fuzzy systems characteristics such as, fuzzy rules, membership functions, and input and output scaling factors examined heuristically based on empirical knowledge and experimental trial and errors by the human designer [7]. It is natural that it is necessary to improve quality of any production technology in a complex way, by replacing the technology itself, as well as by continuous optimization of operation, where it is inevitable to exploit not only exact theoretical means, but also practical experience, or heuristic procedures from the level of dispatcher control. It is just here, where new opportunities for exploitation of fuzzy controller are opening, as a compensation for a man in control and optimizing processes [8]. The aim of the connected and discreet synthesis of control circuit is a design of controller structure and controller coefficients which allow the regulated quantity to follow very precisely and rapidly the changes of this quantity. The effect of faulty quantities is held back to a large extent. When designing the structure and the controller coefficient calculations, we need to know the attributes of the regulated process described by I/O measurements realized in an offline way (running input and output measurements, measurements of responses to normed input) or on the bases of online measurements realized in a closed loop with a known-structured controller. The result of the measurements realized on the process is a mathematical model which represents the dynamic attributes of the regulated process or those characteristic quantities representing the essential dynamic

Abstract— The development of computer technology has led to new modern management methods in the work experience. Evidence of this can be found in developments such as AC (adaptive control), robust control and, most clearly, AI (artificial intelligence) and expert systems. It is important to prefer regulator adaptation with minimal overshooting and speed and stability control for meeting criteria of technological process. This article describes one of the possibilities for finding PSD controller coefficients to reach the best quality control in comparison with standard procedures. It is important to prefer regulator adaptation with minimal overshooting, speed and stability control for meeting criteria of technological process. Here is one of the possibilities to effectively find the PSD controller coefficients to reach the best quality control in comparison with standard procedures.

I. INTRODUCTION Automated Control Systems deprives human monotonous and tiring work. During the whole operation, they are able to guarantee stable and high quality of activity what is, in classical systems, an unquestionable asset in the comparison of rolling performances of a human. From the classical point of view, everything relates to the dominant position of controller in the control circuit. The optimal coefficient settings of the controller are still an actual problem of the industrial controller operation. A number of methods/techniques used in practice are about fifty years old, but there is a line of new ones which enriches the contemporary regulation theory with the elements of robustness, adaptation and selfsetting [1] Proportional-Integral-Derivative (PID) control is a traditional linear control method used in many applications [2]. The PID controller algorithm is widely used for industrial automation tasks and thermal comfort heating and cooling applications where error, derivative of error, and integral of error are used in the calculation of control law [3][4]. Along with the development of computer technique we are currently more and more encountered with the infiltration of modern methods of controlling into the practice. Adaptive control, robust control and last but not least also the artificial intelligence and expert systems can serve as examples. An important part of artificial intelligence in the practice represents and will, from now on, represent fuzzy logic and its applications. Tzafestas and Papanikolopoulos [5] have indicated that exploitation of fuzzy logic allows for using the human approach to

978-1-4799-0305-4/13/$31.00 ©2013 IEEE

77

Š. Koprda et al. • The Influence of Control Methods in Technological Process to Save Electrical Energy

attributes of the regulated process (rise time, delay time, critical frequency, critical amplification, time constants, traffic delays, etc.). In most cases, these quantities are enough for primal determination of controller coefficients [9].

     

II. METHODS OF RESEARCH While finding the transitional characteristic, the procedure was as follows: the thermo-dynamic system was heated by a bulb of 55 W. The heating period lasted about 24 hours. The sensors of temperature were placed on the upper and the lower part of the system. Four types of sensors were used with the name Dallas DS18B20. Sensors placed in the system were connected to the micro processing device DN 20 and by the serial interface RS 232 it was consequently connected to the computer on which the MATLAB program was running with a created control program for the given measurement. The transitional characteristics of the system are shown in Figure 1.



Thermodynamic system of the first order, Temperature sensors, Humidity sensor, Heat source, Power supplies, Facility serving for temperature and humidity measurement named DN 20, Facilities for the measurement of consumption,

Control programme, MATLAB.

created

in

the programme

Power source

Computer with control program to create in MATLAB

DN 20

Thermodynam ic system of first stream

Block of control bulbs and fan

Power source Figure 2. Block diagram of measuring device

B. The design concept of the termodynamic system of the first order The thermodynamic system was created by means of a wooden frame with the dimensions x = 1,0 m, y = 0,5 m and z = 0,5 m. Thereafter, this frame was veneered from inside by an insulating plasterboard with the same dimensions as the size of the frame. The thickness of the insulating plasterboard is 1 cm. This frame was veneered from the outer side by polystyrene with the same dimensions as the frame. The thickness of the polystyrene is 5 cm. The volume of the system is 0,25 m3. Figure 2 describes the real model of the thermodynamic system. The control of the 55W bulb was permitted by the means of the control circuit connected to the device DN 20 shown in Figure 3.

Figure 1. Measurement transmission performance of heat system model

The equation of this thermal system will be: G( p) 

KP Tp  1

(1)

where T will be defined as (1-1/e) fold of the final value. KP is stable value p is Laplace operator e is Euler´s number To calculate the T, the median characteristic has decided for T = 14290 s KP = 1 Then the consequential transmitting function of the system is: G p  

1 14290p  1

(2)

A. The arrangement of measuring string for temperature regulation by a two-state regulation

Figure 3. The control circuit connected to the device DN 20

Measuring chain consists of:

78

SISY 2013 • IEEE 11th International Symposium on Intelligent Systems and Informatics • September 26-28, 2013, Subotica, Serbia

catalogue heat conductivity of one kind of plasterboard, which was 0,113 Wm-1K-1, which is a lower value. This could have been caused by thermal losses, or different kind of plasterboard.

C. Heat conductivity of insulating plasterboard In this section we determined heat conductivity of the insulation plasterboard by means of the following relationship: S (3) Q   t1  t 2  l The supplied heat is marked as Q, λ is heat conductivity to be determined, S is the board surface, l is the board thickness, t1 - t2 is the difference of temperatures at both sides and τ is the time during which heat Q is supplied. It is expected that the temperature difference is stabilized and that no heat is being lost. The relationship can be adjusted as follows: Q l (4)   S t1  t 2 

To realize a two-state regulation, a regulatory scheme was created in the MATLAB program which ensures this regulation. The control scheme is in Figure 4.

The fraction Q/ τ represents the supplied heat during the time period. If its source is electric heater elements, it is possible to substitute the output UI directly. Thus the relationship is: UIl (5)  S t1  t 2  The shape of the insulating plasterboard was cuboid with a square base with the side dimension 201 mm and thickness 13 mm. It was heated from one side by a foil with a resistance wire inside, powered up by a source with an alterable voltage and current. The board was cooled from the other side by water flowing through a metal cooler. The insulation plasterboard was separated from the foil and the cooler by a silicone heat foil. It was also equipped with incisions for thermoelements, by means of which temperature was determined on both sides. The whole system was separated from the environment by polystyrene boards. Prior to the measurement itself it was necessary to calibrate thermoelements (to determine the dependence of voltage on the thermoelement on the temperature difference). This was done so that one end of the thermoelement was immersed in hot water, while the other was put aside with the surrounding temperature. Voltage of the thermoelement is directly proportional to the temperature difference on its ends. 1 (6) U  T k Temperature of the environment was T1 = 26,4 ºC, the water temperature was T2 = 82 ºC. Voltage on the thermoelement was U = 2,34 mV. We thus obtain k = (T2T1)/U which equals 23,76 KmV-1. Having determined properties of the thermoelement, the measurement itself was executed. The results are presented in the table. TABLE I.

Figure 4. Control scheme for two – state controller

The number of used sensors and their deployment in the thermodynamic system was verified by a series of experiments. Homogenous groups or statistically significant differences in the temperature measured by the used sensors were identified using an analysis of experimental data. Individual experiments differed by the number of sensors and their deployment. The process of the experimental data was inspired by [10][11][12] [13]. TABLE II. SENSOR

T , K

15

0.74 0.92

0.06

22.69

, Wm-1K-1 0.157

18

0.88 1.28

0.10

31.3

0.163

22

1.08 1.60

0.23

43.31

0.177

Mean

1

2

3

4

temperature2 27.61241 **** temperature1 27.87892 **** temperature5 27.88321 temperature3 27.97136 temperature4 28.16232

**** **** ****

From the multiple comparison one homogenous group (temperature1, temperature5) was identified. In case of the other sensors statistically significant differences in the obtained temperatures were proved. Exploitation of four sensors in our system turned up satisfactory.

RESULTS OF HEAT CONDUCTIVITY OF PLASTERBOARD

U, V I, A U1, mV U2, mV

MULTIPLE COMPARISON (TUKEY HSD TEST)

D. The arrangement of measuring string for temperature regulation by PSD regulation In the PSD regulation we used a control scheme created in the MATLAB program. The temperature in the thermo dynamic system was set at 28° C and this temperature is reached by a bulb. After reaching it, a fan is starting to

Source: own research

Heat conductivity of plasterboard λ = 0,166 Wm-1K-1 (arithmetic average). For comparison we found in the

79

Š. Koprda et al. • The Influence of Control Methods in Technological Process to Save Electrical Energy

work. The duration of the regulation was 5, 75 hours. The time of sampling was Tvz = 30 s. The value of the bulb´s and the fan´s power output fluently changed according to the temperature of our system. Kuhn´s method was used to set the PSD coefficients. It can be used because in the small period of sampling the PSD control is comparable to PID. Kuhn´s method is used to calculate the parameters of PSD regulator because this method is appropriate for PSD regulator settings with regulated first rate system. The setting procedure of PSD regulator coefficients is as follows [14][15]: The general equation of regulated system is:

Gs   K P

T T

Figure 6. Block diagram for two – condition controller (1) average temperature, (2) regulation 1, (3) regulation 2, (4) output of fans

s  1T2 N s  1.....TmN s  1 T s e (7) s  1T2 N s  1.....TmN s  1 1N 1N

D

In the block diagram (Figure 6), we can set the temperature value on which we want to regulate the system. The reached results are shown in the following graph.

The overall time constant is defined as: T  T1  T2  ... Tn  T1N  T2 N  ... TmN  TD (8) It is necessary to determine the stable value of the transitionally characteristic K in Figure 5 to calculate the PSD regulator coefficients.

On-Off regulácia (17_1_2007),Tvz=30s,Tpoz=28°C 29

27

teplota[°C]

T, ºC (2)

28

26

25

24

23

22 0

2000

4000

6000

8000

10000

12000

14000

16000

18000

20000

čas[s]

time, s (1) (1)

Figure 7. Continuance of two – condition regulation for temperature 28ºC, (1) time, (2) temperature

Figure 5. Transmission characteristic

The time of sampling was Tvz = 30 s. The span of temperature in steady state is 0,65°C. The block diagram allows us a separate realization of PSD regulator. The block diagram is in Figure 8.

We calculate each PSD regulator coefficient as: k = 1/K Ti = 0,66 Td = 0,167 T = T is related to regulated first rate system and K = 1 is related to stable value of transitionally characteristic. Then the PSD regulator parameters calculated by Kuhn´s method for the system are as follows: k=1 Ti = 9432 s Td = 2387 s III.

RESULTS AND DISCUSSION Figure 8. Block’s block diagram for PSD controller (1) average temperature, (2) regulation 1, (3) regulation 2, (4) output of fans

The temperature in the thermo dynamic system was set at 28° C and this temperature is reached by a bulb. After reaching it, a fan is starting to work. The duration of the regulation was 5,75 hours. The time of sampling was T vz = 30 s. The bulb´s and fan´s power output value was set at 100% in the two-state regulation. Block diagram is in Figure 6.

The most important part in this block diagram is the Matlab Function where is the equation of PSD regulator.

80

 T a0  k 1   d    Tvz

   

 T a1  k 1   vz    Ti

 Td   2  T   vz

(9)

   

(10)

SISY 2013 • IEEE 11th International Symposium on Intelligent Systems and Informatics • September 26-28, 2013, Subotica, Serbia

T  a2  k  d   Tvz  U  a0en  a1en1 a 2 en 2

(12)

U n  U n1  U

(13)

IV.

(11)

Before measuring, an assumption was determined that the value of the temperature span in two-state or PSD regulation can be at most 1°C. In two-state regulation we recorded that the span of temperature was 0,65°C that represents 65% of the maximum. In the PSD regulation the temperature span was 0,15°C which is 15% of the maximum. We can note that in the given system PSD controller can regulate better by 50% than the two-state controller. We reach far more precise setting in the system by the use of PSD controller and thereby optimal conditions in technological process. The temperature can be regulated in the range from 33°C to 15°C.

where en is the n-th value of regulatory anomaly en1 is n-1 value of regulatory anomaly en2 is n-2 value of regulatory anomaly Un is n value of regulating variable Un1 is n-1 value of regulating variable U is increase of regulating variable a0, a1, a2 are coefficients of PSD controller The experimental coefficients of PSD regulator to acquire the best criterions of regulation quality were set as follows:

ACKNOWLEDGMENT This publication is supported by the Fund for supporting the Centres of Research and Development with internationally comparable quality of operations, Faculty of Natural Sciences, CPU Nitra, Slovakia.

k = 12 Ti = 10 600, s Td = 1, s

REFERENCES

There are the reached results in Figure 9.

O. Lukáč, B. Brachtýr, P. Bystriansky, “Modelovanie a simulácia výmenníka tepla” Sekel 2002: Zborník z medzinárodného odborného seminára Trenčianske Teplice 24. 26.09.2002, pp. 7781. [2] A. Prodic and D. Maksimovic, “Design of a digital PID regulator based on look-up tables for control of high-frequency dc–dc converters,” Proceedings of IEEE Workshop on Computers in Power Electronics, 2002, pp. 18–22. [3] MM. Ardehali, M. Saboori and M. Teshnelab, “Numerical simulation and analysis of fuzzy PID and PSD control methodologies as dynamic energy efficiency measures,” Energy Convers Manage, 2004, vol.45, Issue 13-14 pp. 1981-1992. [4] S. Koprda, Z. Balogh and M. Turčani, “Modeling and comparison of fuzzy PID controller with PSD regulation in the discrete systems” International Journal of Circuits, Systems and Signal Processing, 2011, vol. 5 (5), pp. 496-504. [5] S. Tzafestas and NP. Papanikolopoulos, “Incremental fuzzy expert PID control,” IEEE Trans Ind Electron 1990; vol. 37(5), pp. 365– 371. [6] AB. Shepherd and WJ. Batty, “Fuzzy control strategies to provide cost and energy efficient high quality indoor environments in buildings with high occupant densities,” Build Services Eng Res Technol 2003, vol. 24(1), pp 35–45. [7] DA. Linkens and HO. Nyongesa, “Genetic algorithms for fuzzy control part1: offline system development and application,” IEEE Proc Control Theor, 1995, vol. 142(3), pp161–185. [8] W. Frey, M. Sajidman and H. Kuntze, “A Neuro-fuzzy Supervisory Control System for Industrial Batch Processes,” The Ninth IEEE International Conference on Fuzzy Systems, San Antonio, Texas, 2000, pp. 116-121. [9] Strunk Jr W andWhite EB. “The elements of style” 3rd ed. New York: Macmillan; 1979. [10] M. Munk, J. Kapusta and P. Švec, “Data preprocessing dependency for web usage mining based on sequence rule analysis,” IADIS European Conference on Data Mining 2009, ECDM'09, Algarve, 2009, pp. 179-181. [11] M. Munk, J. Kapusta and P. Švec, “Data Preprocessing Evaluation for Web Log Mining: Reconstruction of Activities of a Web Visitor,” Procedia Computer Science, 2010, Vol. 1, No 1, pp. 2267-2274. [12] D. Klocoková and M. Munk, “Usage analysis in the web-based distance learning environment in a foreign language education: case study,” Procedia - Social and Behavioral Sciences, 2011, vol. 15, pp. 993-997.

Temperature, C°

[1]

Time, s Figure 9. Continuance of PSD regulation for temperature 28 ºC, (1) time (2) temperature

The span of temperature in the PSD regulation is 0,15°C. Consequently, the given PSD regulator in comparison to the two-state regulation, it is observable that by optimization of the PSD regulator parameters it is possible to reach bigger span by 0,5°C than with a twostate regulation. A. Comparison of electric energy consumption of a twostate and a PSD controller To compare the consumption of the two controllers, the same time of measurement had to be fulfilled which was 3,6 hours. The controller consumption comparison is shown in the table below. TABLE III.

CONCLUSION

COMPARISON OF CONSUMPTION: TWO – STATE CONTROLLER AND PSD CONTROLLER

Consumption

Two-state controller

PSD controller

Consumption in 3,6 hours [Wh]

201,726

132,462

Consumption in 1 hour [KWh]

0,056035

0,036795

81

Š. Koprda et al. • The Influence of Control Methods in Technological Process to Save Electrical Energy

[13] D. Klocoková, “Integration of heuristics elements in the webbased environment: Experimental evaluation and usage analysis,” Procedia - Social and Behavioral Sciences, 2011, vol. 15, pp. 1010-1014.

[14] M. Sazima, V. Kmoníček, J. Schneller et al. “Teplo,”. Praha: SNTL, 1989,pp. 592. [15] Brož, K. “Vytápění,” ČVUT Praha, 1995.

82