Abstract: Addressing the nonlinearity, large time lag, and difficulty in establishing accurate mathematical models of the core process in cement production—the calcination system—a programmable logic controller (PLC) combined with a fuzzy algorithm is employed to control the calcination process of a cement rotary kiln. Results show that the control system based on this method exhibits strong anti-interference capabilities, simple implementation, and good regulation quality.
Keywords: fuzzy algorithm; firing system; rotary kiln; programmable logic controller
Abstract: Aiming at the non-linear, time delay and the difficulty to build an accurate arithmetic model of the firing system ,which is the hard core in cement production, fuzzy algorithm is applied in PLC to control the cement firing system. It is confirmed that the system based on the theory can realize easily and resist interference effectively along with a better control quality.
Key words: Fuzzy Algorithm;Firing System; Rotary Kiln; PLC
1. Cement production process and control requirements
The main process flow of a cement production line can be divided into: raw material batching station, raw material mill and waste gas treatment, homogenization silo, calcination system, pulverized coal preparation and clinker silo, as shown in Figure 1. The raw materials from the raw material batching station are crushed by the raw material mill and then sent to the homogenization silo for homogenization. The homogenized raw materials are directly sent to the rotary kiln for calcination. The quality of cement mainly depends on the calcination of the raw materials in the kiln; good cement requires thorough calcination in the kiln, so the calcination system is the core link in cement production. After calcination, the cement is cooled and sent to the clinker silo, awaiting shipment.
Figure 1. Cement production process flow chart
1.1 Control requirements for cement firing system
In the cement clinker production process, the temperature of the calcination zone and the temperature of the kiln tail exhaust gas are the most important factors affecting the quality of cement clinker. Maintaining stability of these two temperatures not only ensures good calcination quality but also plays a crucial role in stabilizing the thermal conditions of the rotary kiln and the stable operation of the main equipment.
Meanwhile, the oxygen content of the kiln tail gas is an important parameter reflecting the combustion status inside the kiln. Too low an oxygen content indicates an oxygen-deficient combustion state, leading to incomplete combustion, wasting fuel, and producing black smoke and causing pollution. Conversely, too high an oxygen content results in excess air carrying away a large amount of heat, causing waste, and also produces harmful components such as nitrogen oxides and sulfur dioxide due to excessive combustion. Therefore, monitoring the oxygen content of the tail gas and maintaining it at 1.8%–2.5% ensures combustion operates under reasonable air-fuel ratio conditions, which is of profound significance for energy conservation, fuel saving, and environmental improvement.
Therefore, the firing zone temperature (BZ), kiln tail exhaust gas temperature (BE), and kiln tail exhaust gas oxygen content percentage (OX) are used as control variables, while the coal feed rate (CA), which has a significant impact on the control variables, and the primary air volume (FS), which affects the flame shape of the firing zone and the temperature distribution inside the kiln, are used as manipulated variables. The speed of the kiln tail Roots blower and the speed of the kiln head twin-pipe screw feeder motor are adjusted by changing the frequency converter.
Each manipulated variable affects different stages, such as preheating, calcination, slag formation, and cooling, and has a different time lag, ranging from a few minutes to several hours. For such complex systems with multiple variables and large time lags, traditional PID single-loop regulation often fails to achieve the desired results.
1.2 Implementing firing system control using fuzzy control algorithm
In recent years, with the development of intelligent control technology, many new control methods have emerged, one of which is fuzzy control. Fuzzy control does not require a precise mathematical model of the controlled object; instead, it determines the magnitude of the control variable based on control rules. This control method exhibits good control performance for systems with large time lags or random disturbances.
For the firing system, due to the large number of input and output variables that influence each other and form coupling, a multi-input multi-output fuzzy controller is considered. The control scheme is shown in Figure 2.
Figure 2. Control block diagram of rotary kiln
The inputs of the fuzzy controller are the temperature deviation of the burning zone BZ, the temperature deviation of the exhaust gas BE, and the oxygen percentage deviation of the exhaust gas OX. The two outputs of the fuzzy controller, CA and FS, control the frequency converters of the coal feed motor and the blower to adjust the coal feed rate and the air supply rate, so that the combustion operates under reasonable air-fuel ratio conditions. This control method has a good control effect on systems with lag or random disturbances, improves the control accuracy and reliability of the system, and thus meets the process requirements [1] [2].
2. Fuzzy Control Algorithm Design
2.1 Input Fuzzification
Fuzzy control consists of three parts: input fuzzification, fuzzy inference, and defuzzification. Both the temperature deviation of the firing zone and the temperature deviation of the exhaust gas are precise input values. To use fuzzy control technology, they need to be converted into membership functions of fuzzy sets. Currently, the three most widely used fuzzers are single-valued fuzzers, Gaussian fuzzers, and triangular fuzzers. Studies have shown [1] [2] [3] that Gaussian fuzzers or triangular fuzzers can overcome the noise contained in the input variables, while single-valued fuzzers cannot. Considering the system, the Gaussian fuzzer, which combines descriptiveness and simplification, is selected here.
The fuzzy subsets of the control variables and the manipulated variables are selected as follows:
The fuzzy subset of the firing zone temperature BZ is defined as: {low, normal, high}, which can be represented symbolically as: {NB, ZR, PB};
The fuzzy subset of oxygen proportion OX in exhaust gas is defined as: {low, moderate, high}, and symbolically represented as: {NB, ZR, PB};
The fuzzy subset of kiln tail exhaust gas temperature BE is defined as: {low, normal, high}, and symbolically represented as: {NB, ZR, PB};
The fuzzy subset of coal supply speed CA is defined as: {decreasing slightly lower, remaining unchanged, slightly higher, increasing}, and symbolically represented as: {NB NS ZR PS PB}. The kiln tail blower speed...
The FS fuzzy subset is defined as: {lower, slightly lower, unchanged, slightly higher, increased}, and is represented by the symbols: {NB NS ZR PS PB}.
The universe of discourse for state variables and control variables is divided as follows:
The input state variable, combustion zone temperature BZ, is quantized into 11 levels, with a universe of discourse of [-100, -80, -60, -40, -20, 0, 20, 40, 60, 80, 100];
The proportion of oxygen (OX) in the exhaust gas is quantified into 11 levels, with a domain of discourse of [-2.5, -2, -1.5, -1, -0.5, 0, 0.5, 1.5, 2.0, 2.5];
The kiln tail exhaust gas temperature (BE) is quantified into 11 levels, with a domain of discourse of [-50, -40, -30, -20, -10, 0, 10, 20, 30, 40, 50];
The output control variable, coal supply rate (CA), is quantified into 9 levels with a universe of discourse of [-2.1, -1.5, -0.9, -0.3, 0, 0.3, 0.9, 1.5, 2.1].
The kiln tail blower speed FS is quantified into 9 levels, with the domain of discourse being [-70, -50, -30, -10, 0, 10, 30, 50, 70].
The membership function graphs of the input and output quantities are shown in Figure 3.
Figure 3. Membership functions of input and output quantities
2.2 Fuzzy Decision Making and Fuzzy Control Rules
Since BZ, BE and OX are defined as three fuzzy subsets, there are a total of 3×3×3=27 rules. This paper combines the operation manual of cement kiln control and summarizes a lot of practical operation experience of skilled operators to derive the fuzzy rule base. Table 1 is taken from a cement kiln operation manual. This part describes why the operator must adjust the fuel supply rate and the air flow in the kiln according to different situations, using the temperature of the combustion zone (BZ), the oxygen content of the exhaust gas (OX) and the temperature of the cement kiln tail end (BE). It can be seen that the environmental factors and control behaviors are described in qualitative terms as "high", "moderate", "low", "slightly increased" etc. [3][4].
Table 1 Cement Kiln Operation Manual
The language rules in Table 1 can be expressed using the fuzzy language "IF-THEN" statement:
IF BZ=ZR, OX =NB, AND BE=NB, THEN CA=PB, FS=PB
IF BZ=ZR, OX =NB, AND BE=ZR, THEN CA=NS, FS=ZR
IF BZ=ZR, OX =NB, AND BE=PB, THEN CA=NB, FS=NB
IF BZ=PB, OX =ZR, AND BE=ZR, THEN CA=PB, FS=PB
IF BZ=ZR, OX =ZR, AND BE=ZR, THEN CA=ZR, FS=ZR
IF BZ=ZR, OX =ZR, AND BE=PB, THEN CA=NB, FS=NB
IF BZ=ZR, OX =PB, AND BE=NB, THEN CA=PB, FS=PB
IF BZ=ZR, OX =NB, AND BE=PB, THEN CA=ZR, FS=NS
The general principle for rule selection is: when the error is large, the control quantity should be selected to eliminate the error. When the error is small, the control quantity should be selected to prevent overshoot and prioritize system stability [4]. In practical applications, there are a total of 27 fuzzy IF-THEN rules, as shown in Table 2.
Table 2 Fuzzy Control Rule Table
2.3 Output Defuzzification Decision
The process of representing the fuzzy set obtained from fuzzy reasoning with a definite value is called defuzzification. There are various methods for defuzzification decisions, among which the centroid method is relatively reasonable and suitable for systems with high accuracy requirements. The formula for defuzzification using the centroid method is:
In practical applications, in order to reduce online calculations, fuzzy control tables are often generated through offline calculations. After the input values are fuzzified, the corresponding control values can be found in the table when the input is entered into the fuzzy controller, which enhances the real-time performance of the system.
3. PLC Implementation of Fuzzy Control Algorithm
The Mitsubishi FXon PLC was selected for the control system, with 12-bit FX-4AD and FX-2DA analog input/output modules respectively. The PLC-based fuzzy algorithm program consists of three main parts: input fuzzification, fuzzy control table lookup, and output defuzzification. The fuzzy algorithm is implemented in the PLC program via interrupts. The PLC-based fuzzy control program is essentially a lookup table program. In the actual control process, in each sampling cycle, the measured errors of the three input quantities are multiplied by their respective quantization factors to obtain the corresponding universe elements E1, E2, and E3 needed to look up the lookup table. By querying the offline-calculated fuzzy control table through the lookup program, the output control quantities ΔU1 and ΔU2 can be obtained. These are then multiplied by the scaling factors ku1 and ku2 and added to the output Uk-1 from the previous sampling time to obtain the actual voltage control quantity applied to the frequency converter, thereby changing the speed of the kiln tail fan and the kiln head double-pipe screw feeder motor. The specific flowchart is shown in Figure 4.
In program design, the following should be noted:
(1) Depending on the situation, add an offset to the data; unsign the signed numbers to simplify the calculation.
(2) The input quantity is collected into the data storage area of the PLC using the A/D module. After the input quantity is subjected to amplitude limiting quantization, it is determined whether e and ec exceed the limit. If they exceed the limit, they are set to the upper or lower limit value. Otherwise, the input quantity is multiplied by the quantization factor and quantized into the corresponding elements E and EC in the fuzzy universe of the input variable.
(3) Determine the quantization factor and put the quantization factor into the holding relay of the PLC.
(4) Based on the corresponding elements in the corresponding input fuzzy domain, look up the fuzzy control rule table to obtain the fuzzy output quantity, and then multiply it by the output quantization factor to obtain the actual output quantity, which is then controlled by the output of the D/A module.
4. Conclusion
Combining fuzzy control algorithms with a PLC to control a cement kiln firing system, and using a PLC for fuzzy control, retains the reliability, flexibility, and adaptability of PLC control systems while improving the intelligence level of the control system. This system demonstrates stable control performance in practical applications, making it an ideal solution. By selecting appropriate sampling periods and quantization factors, the system can achieve good performance indicators, thus meeting control performance requirements.
Figure 4 Flowchart of PLC implementing fuzzy control algorithm
References:
[1] Li Youshan, Li Jun. Fuzzy Control Theory and Its Application in Process Control. [M] Beijing: National Defense Industry Press. 1997: 46-77
[2] Holmblad .L .P, and JJOsterguard[1982].Control of a cement kiln by fuzzy logic” [J] In:Gupta,MM, and E.Sanchez,eds.,Fuzzy Information and Decision Process,Noth-Holl and,Amstrerdam,pp.398～409
[3] Wang Lixin. Fuzzy Systems and Fuzzy Control Tutorial [M]. Beijing: Tsinghua University Press, 2003, pp. 167-171.
[4] Liu Yinghui. Microcontroller-based modification of electric arc smelting furnace based on fuzzy theory [J]. Microcomputer Information, 2006, 8-2: 136-138
