基于模型自适应控制的锅炉自动控制系统设计与实现.docx

《基于模型自适应控制的锅炉自动控制系统设计与实现.docx》由会员分享，可在线阅读，更多相关《基于模型自适应控制的锅炉自动控制系统设计与实现.docx（6页珍藏版）》请在得力文库 - 分享文档赚钱的网站上搜索。

1、B Design and Realization of Automatic Control System for Boiler based on Model Free Adaptive Control Jing Wang*, Chao JiLiulin Cao, Qibing Jin Institute of Automation, College of Information Science and Technology Beijing University of Chemical Technology, Beijing, China Abstract: To achieve good pe

2、rformance of the utility boiler, dynamic variables such as drum pressure, steam temperature and water level of drum, etc. must be controlled. According to the operating rules and the characteristics analysis of industrial boiler, a complete control scheme for boiler is designed. Specifically, due to

3、 the typical nonlinear and coupled characteristics of outlet pressure of boiler, a model free adaptive control (MFAC) method is used to maintain it at the respected values. The automatic control system is realized by the DCS of SIMATIC PCS 7 Box. Comparisons between MFAC and traditional PID control

4、in detailed prove the practicability and robustness of the MFAC control scheme proposed here. Key words: boiler system; MFAC; PCS 7 BOX I. INTRODUCTION OILER unit, that produces superheated steam, is widely used in power, chemical plant and daily life. It always works under high pressure and tempera

5、ture, so its security is very important. On the other hand, all the quantity indexes of boiler such as the steam pressure, temperature and water level of drum will affect the performance of plant operation, which must be maintained at their respected values. However, the physical constraints exerted

6、 on the actuators must be satisfied by the control signals. These constraints can be the magnitude and saturation rate for the control valves of the fuel, steam and feed-water flow1 1, 2. An industrial boiler unit is a nonlinear complex system. *Corresponding author: Box81, Beijing University of Che

7、mical Technology, 15 Beisanhuan East Road, Chaoyang District, Beijing 100029, Peoples Republic of China. Email: (J Wang) Several dynamic models of the boiler system have been developed 35, and various control methods have been applied to boiler controller design, e.g., adaptive control 6, Linear Qu

8、adratic Gaussian (LQG) 7, predictive control 8, Robust control 9 and intelligent control 10,11. But in these works, accurate modeling is an inevitable. In fact, the boiler system is nonlinear, coupled, and uncertain, and its accurate model is hardly available in practice. Hence, a simple and robust

9、control method- model free adaptive control (MFAC) that is insensitive to modeling uncertainties and suitable for nonlinear dynamics is considered here. Recently, Hou 12 developed a new dynamical linearization method by introducing a concept of pseudo partial derivative (PPD), then MFAC scheme based

10、 on PPD is proposed for general nonlinear systems. As a data-driven method in nature, it has received increasing attention from the control community 13, 14. Since the boiler system is a complex process with multiple variables, strong coupling, strong interference and large-lag. Due to the influence

11、s of above factors, the traditional PID control effect is not ideal 15. According to the operating rules and the characteristics analysis of industrial boiler, a complete control scheme for boiler is designed, which includes combustion control system, drum control system, stem temperature and other

12、control systems. Especially, to achieve a good performance of boiler unit, model free adaptive control strategy (MFAC) is proposed to control the superheated pressure steam, in which the fuel flow is the manipulated variable. Comparisons between MFAC and traditional PID control in detailed prove tha

13、t the MFAC control scheme has better 978-1-4244-8736-3/11/$26.00 c 2011 IEEE 1881 practicability and robustness. II. TECHNOLOGY OF INDUSTRIAL BOILER The process of natural circulation boiler is shown in Fig.1. The feed water flow of boiler is F1101, and bypass valve is HV1101. The boiler feed water

14、is divided into two channels, one flows into the desuperheater E1101 for preheated. After preheated, it mixes with the other channel and flows into the economizer E1102. Two feed water pipes have respectively control valve, V1102 and V1103. The drum pressure is P1103 and there is a water level detec

15、tion L1102 in the middle of drum. In the drum, saturated stem with the temperature T1102 is obtained by gas-liquid separation. T1103 is superheated steam temperature, which have been passed the furnace. The role of desuperheater is fine-adjustment of superheated steam, and the pressure, temperature

16、and flow of end-point superheated steam is P1104, T1104 and F1105, respectively. V1105 is a control valve in the outlet pipe. Fuel feeds into the burner of furnace F1101 through fuel pump P1102 with flow rate F1103 and pressure P1101.The control valve V1104 locates in fuel pipeline. Air feeds into t

17、he burner with flow rate F1104 through the frequency conversion fan K1101. The frequency of converter is S1101. (it is normalized from 0 to 100%). The flue gas flow and temperature in the outlet of economizer is F1107 and T1105. There are on-line analysis instrumentation to measure the oxygen conten

18、t of flue gases. DO1101 is the damper in the economizer (flue gas) outlet. The furnace temperature T1101 is extremely high, which is measured by infrared non-contact sensor so that it can only provide a general reference. And furnace draft is P1102 which is a important parameter 16. The nominal oper

19、ating conditions are specified as follows 1. Superheated steam pressure P1104:3.8Mpa; 2. Superheated steam temperature T1104:450 3. Superheated steam flow F1105:11Kg/s 4. Drum water level L1102:45-55% 5. Furnace draft P1102:10-300mm; 6. Fuel gas oxygen content A1101:1-3%. Fig.1.The technology of ind

20、ustrial boiler III. CONTROL SYSTEM DESIGN AND IMPLEMENTATION A. General design of basic process control system According to the operating rules and the characteristics analysis of industrial boiler, a complete control scheme for boiler is designed, which is divided into six parts as follows: (1) Dru

21、m water level cascade-three-element control system (2) Combustion system and flue gas oxygen content closed loop control system (3) Superheated steam temperature split control system (4) Furnace draft control system (5) Superheated steam flow single-loop control system The summary of the basic proce

22、ss control design is shown in Table 1. B. Implementation of control system The basic process control system is implemented in SIMATIC PCS 7 BOX. Here a brief explanation of control implementation is given: flue gas oxygen content control system. It is the core part of boiler control system. 1882 201

23、1 Chinese Control and Decision Conference (CCDC) Table1 Summary of basic process control fossr boiler No. Control loop Controlled variable Manipulated variable Controller Control action Control law 1 Drum water level cascade-three- element control loop Drum water level Feed water flow postive PI Fee

24、d water flow Feed water flow postive PI 2 Fuel flow-air quantity Ratio control loop Fuel quantity loop Fuel quantity Fuel quantity postive PI Air quantity loop Air quantity Air quantity postive PI Superheated steam pressure control loop Superheated steam pressure Fuel quantity & Air quantity postive

25、 MFAC Flue gas oxygen content control loop Flue gas oxygen content Air quantity postive PI 3 Superheated steam temperature split control loop Superheated steam temperature Attemperator cold water flow negative PI 4 Furnace draft control loop Furnace draft Flue gas flow negative PI 5 Superheated stea

26、m flow control single-loop Steam flow Steam flow postive PI The main loop of combustion system is superheated steam pressure controller, the vice loop is fuel flow-air quantity controller. Using the structure control language (SCL) provided by PCS 7 V6.1, an MFAC module is created based on the princ

27、iples described in Section 4. The whole control scheme realized by continuous function chart (CFC), in which lots of basic modules are used, such as analog input module, analog output module, MFAC module, PID control module, adder module, multiplier module and analog operation module. According to t

28、he control scheme, CFC of combustion and flue gas oxygen content control system can be created, as shown in Fig. 2. (a) superheated steam pressure controller CFC (b) flue gas oxygen controller and fuel flow-air quantity controller Fig.2. Combustion system and flue gas oxygen content closed loop cont

29、rol system CFC After finishing preparation work, each control system is switched to automatically mode based on the no-disturbance switching rule. Then all of the parameters of this boiler control system is adjusted. When the boiler is operated in normal condition, some disturbances are introduced i

30、nto the boiler system, such as load changing and so on. Because the control goal and difficulty of boiler are keeping the superheated steam pressure and temperature within normal limits in the whole operation, the dynamic responses of them are mainly discussed here. 2011 Chinese Control and Decision

31、 Conference (CCDC) 1883 u(k 1) 2 k k IV. SUPERHEATED STEAM PRESSURE CONTROL BASED ON tracking error, while the second part u(k) u(k 1)2 can MFAC Since the superheated steam needed by downstream usually goes through the convection and radiation section of the boiler, the superheated steam system cont

32、rolled by restrict the great change of control input. Then the one-step-ahead control sequence can be obtained by using the gradient descent optimizing technique for the cost function Eq.(6). fuel flow is a big lag system obviouslyand the traditional PID control effect is not ideal. Here MFAC contro

33、ller 17 is selected to control the superheated steam, which composed u(k ) u(k 1) (k ) (k) y* (k 1) y(k) (7) of pan-model and control function. It is noted that the Where (0 1) is optimal step. pan-model does not depend on concrete mathematical model of controlled object. The nature of MFAC is refin

34、ing the pan-model when it is controlled. This method can improve the controller performance. A general discrete time nonlinear system is given as follows: It is obviously that MFAC controller is composed of pan-modeling identification algorithm (4) and basic control algorithm (7). The block diagram

35、of MFAC method is shown in Fig.3. y(k 1) f ( y(k), y(k m),u(k),u(k n) (1) Where y(k), u(k) are the system input and output respectively; m, n are the system dimension. According to the system dynamic input, its pan-model is defined as: y(k) y(k 1) (k 1)u(k 1) u(k 2) (2) Where (k 1) could be estimate

36、d by every estimation algorithm of time-varying parameter. Here proposing the criterion function of parameter estimation as follows: J (k) y* (k) y(k 1) (k)u(k 1)2 (k) (k 1)2 (3) Where y* (k ) is an output reference value; is a weight factor. In order to obtain the Pseudo- partial- Fig.3. Block diag

37、ram of MFAC system derivative (k ) , the gradient descent optimizing V. RESULTS ANALYSIS technique is used for the cost function Eq.(2), then (k) (k 1) u(k 1) y(k) (k 1)u(k 1) (4) According to the above operation, control experiments include: boiler load increasing and decreasing. Where is optimal s

38、tep. And set (k ) (1) if (k ) (0)or u(k 1) (5) An expanded integrated square error (EISE) index is proposed by considering set-point tracking, noises restraining and constraint of control signal in practical applications, as shown in the following: J (u(k) y* (k 1) y(k 1)2 u(k) u(k 1)2 (6) Where is

39、a weight factor. In this criterion function, the first part y* (k 1) y(k 1)2 can reduce the steady (a)MFAC 1884 2011 Chinese Control and Decision Conference (CCDC) If (k ) ( 0) or u(k 1) Let be (k ) (1) (k ) (k 1) k u (k 1) y(k ) (k 1)u(k 1) u(k 1) 2 u(k) u(k 1) k(k ) y* (k 1) y(k) (k ) 2 2 (b) PID

40、Fig.4. Comparison of MFAC and PID controller for boiler increasing from 11Kg/s to 16Kg/s Table 2 Comparison of MFAC and PID for load increasing Control system MFAC PID P1104 Overshoot% 7.76 8.29 Rise times 900 1150 Response times 1800 2100 F1105 Initial valueKg/s 11.059 11.060 Overshoot% 4.88 5.475

41、Steady-state value Kg/s 16.030 16.045 T1104 Overshoot% 4.86 5.01 P1102 Response process No effect Minimal oscillation L1102 Response process No effect No effect A1101 Response process No effect No effect Under the normal condition, boiler load (superheated steam flow F1105) is increased from 11Kg/s

42、to 16Kg/s (equivalent to from 40t/h to 60t/h), control results are shown in Fig. 4 and Table 2. After this control system is stable, the boiler load begins to fall from 16kg/s to 13kg/s (equivalent to from 60t/h to 48t/h), results are shown in Fig.5 and Table 3. It is shown that when boiler load is

43、changing (increasing or decreasing), the superheated steam pressure P1104 and temperature T1104 will be affected, but restored to the set-point values 3.8 Mpa and 450 finally. The control process and results in load decreasing is not a strictly inversion of that in load increasing, which means that

44、the boiler is a typical nonlinear system. As shown in Fig.5, the response time of load changing (including increasing and decreasing) using MFAC and PID controller are 4700s and 5000s respectively. The MFAC control strategy has better features than PID controller, such as fast response, strong anti

45、interference performance. VI. CONCLUSION This paper shows practical applications of a MFAC in an industrial boiler system. According to the operating requirements and the characteristics analysis, a complete control scheme for boiler is designed, which includes five parts: combustion and flue gas ox

46、ygen content control system, drum water level control system, superheated steam temperature control system, furnace draft control system and superheated steam flow control system, in which MFAC is used to control the superheated steam pressure. The basic process control system is implemented in SIMA

47、TIC PCS 7 BOX. Experiment results show that MFAC controller has been achieved the industrial control requirement well. Compared with PID controller, MFAC controller has several advantages, such as faster response speed, strong anti-interference ability and robustness. (a) MFAC (b) PID 2011 Chinese C

48、ontrol and Decision Conference (CCDC) 1885 Fig.5. Comparison of MFAC and PID controller for boiler load decreasing from 16Kg/s to 13Kg/s Table 3 Results of MFAC and PID for load falling ACKNOWLEDGMENT This work is supported by National Science Foundation of China under Grant No. 60974031. REFERENCES 1 Chen PC, Shamma JS. Gain-scheduled l1-optimal control for boilerturbine dynamics with actuator saturation. Int J Process C