Abstract: Through analysis and research on the steam drum fluid position control scheme of alkali recovery boiler, an expert-coordinated intelligent control scheme is proposed, which improves upon the traditional three-impulse control method. This scheme can intelligently select the control strategy according to the actual operating conditions of the system, and practical operation results confirm its feasibility. Keywords: combustion workshop section; expert control; steam drum fluid In the alkali recovery process of the pulp and paper industry, the black liquor combustion system is the most complex and important section. The main equipment in the combustion section is the alkali recovery boiler. The effectiveness of the steam drum fluid level control of the alkali recovery boiler directly affects the boiler's efficient and safe operation. The boiler fluid level remains constant only when the boiler feedwater and evaporation rate are stable. If the boiler feedwater or evaporation rate changes, the liquid level will change. Therefore, it is necessary to observe the changes in liquid level to adjust the feedwater flow so that it changes accordingly with the size of the steam load, so as to keep the liquid level within the specified range. In engineering, an automatic liquid level control system is generally used to control the steam drum liquid level. The control effect depends on the structure of the control system and the control strategy. The following is an introduction to the control situation of the combustion section of a paper mill in Shaanxi. 1 Introduction to Boiler Steam Drum Liquid Level Control The steam drum liquid level is the main indicator of boiler operation and a very important controlled variable. Maintaining the liquid level within a certain range is the primary condition for ensuring the safe operation of the boiler. This is because: (1) If the liquid level is too high, it will affect the steam-water separation in the steam drum. Excessive water in the saturated steam will cause scale buildup in the superheated tubes, leading to damage. At the same time, the temperature of the superheated steam will drop sharply. (2) If the liquid level is too low, the water volume in the steam drum will be small. When the load is large, the vaporization rate of water will accelerate. Therefore, the water volume in the steam drum will change rapidly. If it is not adjusted in time, all the water in the steam drum will vaporize, causing the water-cooled wall to burn out or even cause an explosion. Many factors influence the change in boiler drum level. In actual operation, changes in boiler load inevitably cause changes in boiler drum steam pressure, disrupting the thermal equilibrium. Considering the nonlinearity and time-delay of the drum level system, various parameters change with operating conditions. Therefore, using conventional control methods to adapt to the multifaceted factors of the combustion process often fails to achieve the desired control effect. To address this, based on field operating experience, this paper constructs an intelligent expert coordinated control scheme combining conventional PID control and expert control to control the drum level of an alkali recovery boiler. By combining the advantages of traditional control and expert control system technologies, the scheme achieves complementarity in static and dynamic control performance, enabling the boiler level to remain self-controlled even under large disturbances, achieving good control results. 2 Intelligent Expert Coordinated Control Scheme for Boiler Drum Level 2.1 Characteristics of the Upper Drum Level Process in an Alkali Recovery Boiler The requirement for level control is to maintain the drum level within a certain required range, mainly by adjusting the feedwater flow to stabilize the level. From a mass balance perspective, if static balance can be maintained throughout the regulation process, with feedwater flow equal to evaporation flow, the liquid level can be consistently maintained at the set level. However, actual systems are dynamic; feedwater pressure and main steam pressure are constantly changing during regulation, and temperature and pressure compensation for flow is usually only empirical and approximate. Because the system is too complex to obtain real-time accurate feedwater and steam mass flow rates, we indirectly reflect the mass balance of the water system through the boiler drum liquid level. A primary condition for controlling the boiler drum liquid level to the required value is obtaining the actual measured value. However, there is a significant "false water level" phenomenon in boiler drum liquid level control, which requires in-depth analysis to design a satisfactory controller. For boiler drums, when the load increases, the steam flow rate increases, causing a sudden drop in the steam pressure inside the drum. The boiling point of water also decreases accordingly, intensifying boiling and increasing the volume of bubbles below the water surface, thus causing a sudden rise in the liquid level. On the other hand, when the feedwater flow rate suddenly increases, the lower-temperature feedwater absorbs some heat from the existing saturated steam-water mixture, reducing the volume of bubbles below the water surface and causing the liquid level to drop. This phenomenon, in the initial stage of changes in steam flow or feedwater flow, where the drum liquid level changes in the opposite direction, is called a "false water level." From a control perspective, a "false water level" is actually a non-minimum phase system. We will analyze the effect of an increase in steam flow on the drum as an example. With the feedwater constant, an increase in load leads to an increase in steam flow. If other factors are not considered, the drum liquid level will drop linearly because the steam flow is greater than the feedwater flow. 2.2 Common Adjustment Methods for Boiler Drum Liquid Level The task of automatic control of boiler drum liquid level is to ensure that the boiler feedwater volume tracks the boiler evaporation volume and maintains the drum liquid level within the allowable range of the process. Boiler drum liquid level adjustment systems typically adopt the following three adjustment methods: (1) single-impulse feedwater adjustment system; (2) double-impulse feedwater adjustment system; (3) triple-impulse feedwater adjustment system. Drum liquid level control strategies have been theoretically studied for many years, and various mature control methods, including single-impulse, double-impulse, and triple-impulse, have been formed. However, in specific industrial settings, the stability of liquid level control cannot be guaranteed. Disturbances often cause "false water levels" in the steam drum to varying degrees, and under different loads, they also exhibit different degrees of time delay and non-minimum phase characteristics. This dynamic characteristic of the steam drum water level changing constantly at different stages makes it impossible to achieve satisfactory control results using a single-structure control method. Considering various adaptive or parameter self-tuning algorithms would complicate the system mechanism, make tuning difficult, and fail to achieve ideal results under changing field conditions. Due to the complexity and modeling difficulties of alkali recovery boiler systems, adaptive and self-tuning methods based on mathematical models remain difficult to implement in actual control, and successful control still relies on engineers with operational experience. Based on field operational experience, this paper adopts an expert coordinated control scheme, constructing an intelligent expert coordinated control scheme that combines an expert system with conventional PID control to control the upper steam drum water level of the alkali recovery boiler. This allows the boiler water level to remain automatically maintained even under large disturbance amplitudes, achieving good control results. 3 Intelligent Expert Coordinated Control Scheme 3.1 Expert Controller Principle An expert control system can be defined as a system with the function of simulating or extending/expanding expert intelligence, designed using a combination of artificial intelligence expert system technology and control theory. The essence of this system is to base its construction and operation on expert knowledge of the controlled object and control laws, and to utilize this knowledge intelligently to optimize and practicalize the controlled system as much as possible. The basic structure of a typical expert control system is shown in Figure 1. An expert control system consists of a knowledge base, a control algorithm library, a real-time inference engine, information acquisition and processing, and a dynamic database. [align=center] Figure 1: Structure of an Expert Control System[/align] 3.2 Basis for Steam Drum Liquid Level Regulation Strategy The system's expert control scheme is an indirect expert control system. The knowledge base mainly consists of empirical data and a rule base. Information acquisition involves sending standard signals from field instruments to the PLC (Programmable Logic Controller), which is then sampled and analyzed by software. All other aspects are implemented using programming software. We adopted a three-impulse liquid level control strategy for the boiler drum in the alkali recovery combustion section of a paper mill in Shaanxi Province. The system design used the main feedwater valve as the control valve for the boiler drum liquid level. The design concept was as follows: the boiler drum liquid level measurement signal was used as the main control signal to form the main regulating loop; the steam flow signal was used as the feedforward signal to form the feedforward regulating loop; and the main feedwater flow was used as the cascade signal to form the secondary regulating loop. The main regulating loop, feedforward regulating loop, and secondary regulating loop together constituted the boiler drum liquid level cascade three-impulse automatic control system. Its control system block diagram is shown in Figure 2. [align=center] Figure 2 Block diagram of boiler drum liquid level cascade three-impulse control system[/align] Because the system's requirements for static signal coordination are not very strict, the main controller can automatically correct errors caused by inaccurate signal coordination, achieving error-free regulation. However, when the boiler load changes significantly, it is difficult to achieve satisfactory control results. Therefore, the common practice is to use automatic control when the load is stable, and switch to expert-coordinated control when the load fluctuates significantly. The core of expert-coordinated control lies in the establishment of a rule base. Based on the characteristics of the controlled object, the most suitable control law for the current state is selected to effectively control the process. 3.3 Specific Implementation of the Rule Base The rule base is the core of the boiler combustion expert-coordinated intelligent control system. It stores expert knowledge about the combustion process. Specifically, it includes five smaller control rule bases: 1. A boiler drum level expert system rule base established based on the relationship between feedwater flow, drum pressure, load, and drum liquid level; 2. A boiler drum feedwater regulation expert system rule base established using feedwater valve characteristic curves and field operating experience: feedwater flow, feedwater pressure, and valve opening; 3. A boiler drum feedwater and exhaust protection expert system rule base established using the characteristics of feedwater electric valves, safety valves, and emergency drain valves, and relevant field operating experience; 4. A controlled object regulation control expert system rule base established based on the characteristics of the controlled object on-site, related operation, regulation, and control rules, and the operator's operating experience; 5. A controlled object anomaly and accident control expert system rule base established based on the specific on-site controlled object status and necessary relevant operating experience. As shown in Figure 3. [align=center] Figure 3 Control rule base representation of the coordinated control system[/align] Due to space limitations, we will only explain the control rule bases of the steam drum level expert system rule base and the protection expert system rule base here. 3.3.1 Steam Drum Level Expert System Rule Base The expert controller selects appropriate parameter values ​​according to the rules based on the changes in level deviation e (Δe), load change or rate of change D (ΔD), and pressure change (ΔP) collected by the system. For unsteady-state mode, the steam drum level exhibits a false water level phenomenon due to sudden changes in load. This is manifested in the characteristic variables as sudden changes in steam pressure and steam drum level. Therefore, the changes in steam pressure and level deviation are used for identification. When their changes are greater than the sum of the set values, the level system can be considered to be in unsteady-state mode. If the load increases significantly, the steam pressure decreases, and the saturated water in the steam drum boils due to supersaturation, causing the water level to rise rapidly. At this time, because it is a false water level, the valve cannot be closed to suppress the rise in water level using the usual method. Once the pressure inside the steam drum is balanced, the liquid level will drop due to increased load and unchanged feedwater flow. Therefore, the correct operation is to appropriately open the valve to prevent excessive drop in liquid level. The control rule is to adjust the control based on changes in deviation. When the load decreases, the control quantity is appropriately reduced. The threshold and proportional coefficient are determined according to control requirements and actual conditions. Except in the above situations, the steam drum liquid level is stable. If the current steam drum liquid level error exceeds the ideal value, we will use the on/off mode, fully opening or closing the feedwater valve to quickly eliminate the error. When the error trend increases, the control quantity should be increased to correct the deviation as quickly as possible; in this case, the proportional mode is used. A relatively large proportional gain can be achieved. This is then multiplied by the gain amplification factor to increase the control quantity. Once the steam drum liquid level error reaches its extreme value (maximum and minimum points), it is increased by a small value based on the original holding value (holding mode 2), and this output is maintained until the liquid level error e reverses sign (holding mode 1). 3.3.2 Steam Drum Level Protection Expert System Rule Base When the process system is normal, meaning the steam drum level is within the control requirement range (within ±3cm of the set value), normal control rule base adjustments and control outputs are performed. When the process system is abnormal, meaning the steam drum level is between the normal and fault control requirement values, the system switches to the abnormal system control rule base for abnormal adjustments and control. When the process system malfunctions, meaning the steam drum level is below 20% of the lower limit or above 70% of the upper limit, the system switches to the fault system control rule base for fault adjustments and control. Therefore, the fault protection rule base for the boiler steam drum mainly includes the following: High steam drum pressure protection: Open the main steam valve, triggering the main fuel (black liquor) trip (MFT) protection; Low steam drum level: Trip the main fuel (black liquor) trip (MFT) protection; High steam drum level: Open the emergency drain valve; MFT protection: Close the feedwater electric valve. Based on the expert coordinated control library established above, we conducted simulation studies using MATLAB on the three-impulse control method of the feedwater system of the alkali recovery combustion boiler, employing both conventional PID control and intelligent expert coordinated control. The simulation results are shown in Figure 4. The simulation results show that the overshoot of the expert coordinated control is significantly smaller than that of the conventional control, and the entire process is more stable. When disturbed by steam flow fluctuations, the expert coordinated control scheme provides more timely, reasonable, and effective control, overcoming the influence of false water levels. [align=center] Figure 4 Simulation Curve[/align] In summary, for the entire operating condition, regardless of whether the disturbance originates from within or outside the system functional group, the functional group regulation system must ensure that the process system operates within a safe, stable, and economical range. Using an expert controller for the regulation and control of the boiler drum system yields excellent results because the expert controller's decision table is derived from actual on-site operating conditions, and the control rules are composed of extensive expert knowledge and operational experience. This improves the dynamic characteristics (overshoot, transient time) of process regulation and control. Furthermore, the expert controller does not require knowledge of the mathematical model of the controlled object, and the expert controller system possesses strong anti-interference capabilities, robustness, and adaptability. 4 Summary Based on on-site operation experience, this paper adopts an expert coordinated control scheme, which can effectively overcome the shortcomings of three-impulse control. This method is used to control the boiler drum liquid level in the combustion section of the alkali recovery of a certain paper mill in Chengwu. Even under large disturbance amplitude, the boiler liquid level can still maintain self-control and achieve good control effect in practical application. References: [1] Zheng Wenjie, Zhang Xirui. Design of boiler drum liquid level control based on fuzzy PID parameter self-adjustment [J]. Water Conservancy and Electric Machinery, 2007, 29, (1): 16-19 [2] Ma Wentong, Yu Nanhua. A fast dynamic simplified simulation model for drum liquid level [J]. Boiler Technology, 2007, 38 (3): 4-6 [3] Sun Jun. Simulation of boiler drum liquid level and its expert PID control system in ship boilers [J]. Computer Simulation, 2007, 24 (4): 162-164 [4] Yu Haibin, Chu Jian. Expert PID control of CFB boiler drum liquid level [J]. Mechanical and Electrical Engineering, 2000, 17(3): 103-106