1 Introduction
Ammonia stripping is an important part of titanium yellow production waste liquid treatment and by-product iron oxide production. The quality of ammonia stripping system control directly affects the operation of the entire production system, and the temperature control in the ammonia stripping tank is the key to this process. However, the ammonia stripping system has strong nonlinearity, large time delay and uncertainty, which makes it difficult to describe with a precise mathematical model. General control methods are difficult to achieve ideal control effects [1].
This article, based on the actual situation of Chongqing Lurun Titanium Yellow Powder Co., Ltd., introduces an adaptive control system based on a Smith predictor. This system combines a Smith predictor with adaptive control technology, employing a closed-loop control method to adjust the inlet steam flow rate to achieve automatic temperature control within the tank, achieving good results.
2. Research on Temperature Control Algorithm for Ammonia Stripping Tank
2.1 System Modeling
For the ammonia separation process, the transfer function of the ammonia tank temperature control object can be approximately expressed as a first-order inertial element with hysteresis through corresponding calculations and simplifications based on thermodynamic principles [2]:
2.2 Adaptive Smith Predictive Control Algorithm
Considering that the temperature control system of the ammonia stripping tank is a large time-delay system, and the Smith predictor has been considered a theoretically effective method for solving time-delay systems since its inception. However, conventional Smith predictive control can only completely eliminate the influence of time delay when the predictor model is completely consistent with the temperature control system of the ammonia stripping tank [3]. However, the environment in the ammonia stripping tank is very complex and many uncertain factors will occur. Therefore, there will definitely be errors between the model and the actual system. In addition, if there is a disturbance in the control process, the system parameters will change, which will also lead to inconsistency between the predictor model and the system. In view of this, we consider combining model reference adaptation and Smith predictive control. The controlled object is identified online by the generalized error between the reference model and the output of the controlled object. By adjusting the reference model, the dynamic performance of the reference model is made as consistent as possible with the dynamic performance of the controlled object. The reference model here is the Smith predictive controller to be used. Then, the Smith predictive controller is adjusted by the adaptive mechanism, so that a better control effect can be obtained.
The block diagram of the adaptive Smith control system is shown in Figure 1.
Figure 1. Structure diagram of the adaptive Smith control system
Based on the actual system discussed in this paper, it can be seen that the error in equation (2) is within the allowable error range of the control system. This transforms the time-delay system from a nonlinear system into an approximately linear system. The transfer function is:
The model discussed in this paper is a first-order inertial element plus a pure time-delay element. The time-delay element has already been approximated as linearized; therefore, the model discussed in this paper approximates a second-order system. The mathematical model is as follows:
3. Algorithm Simulation
To verify the effectiveness of the adaptive Smith control algorithm, this paper compares it with the conventional Smith control algorithm.
Figure 3 Simulation comparison during model matching
As shown in Figure 3, the adaptive Smith controller has a smaller system overshoot compared to the conventional Smith predictor controller. However, because the adaptive Smith controller requires more computation, its response speed is correspondingly slower. Then, keeping the controller parameters constant, the static gain K of the object model is doubled to 2.4. The response curves of the conventional Smith controller and the adaptive Smith controller are shown in Figure 4.
Figure 4 Simulation comparison under model mismatch
As can be seen from Figure 4, the conventional Smith controller exhibits severe oscillations and excessively large peak values; while the adaptive Smith controller, although also exhibiting some overshoot, still manages to meet the control requirements well.
4. Conclusion
For controlling the temperature of the ammonia stripping tank, the conventional Smith predictive control algorithm exhibits good control performance when the model is matched. However, under model mismatch conditions, the adaptive Smith control algorithm better meets the control requirements and demonstrates better stability and robustness.
About the author: Wang Suiping (1956–), male, from Jiaozuo, Henan Province, holds a doctoral degree and is a professor. His research areas include artificial intelligence, deep-sea robotics, fieldbus, and computer control systems.
Contact number: 13467514859 Email: [email protected]
Mailing Address: Room 317, Democracy Building, Central South University Main Campus, Changsha, Hunan Province, China. Postcode: 410083
References
[1] Zeng Linsheng, Wang Xianbin. Development and application of steam-free ammonia stripping control system in Jinan Iron and Steel Group [J]. Shandong Metallurgy, 2007, 29: 108-110
[2]Beler-BaykalB. , BayramS. ,AkkaymakE. , et al. Removalofammoniumfromhumanurinethroughionexchangewithclinoptiloliteanditsrecoveryforfurtherreuse[J]. WaterScienceandTechnology, 2004, 50(6): 149-156
[3] Zhu Xiaodong, Wang Jun, Wan Hong. Control of pure time delay systems based on Smith prediction [J]. Journal of Zhengzhou University, 2004, 25[1]: 77-81
[4]PandaRC,HungSB,YuCC.AnintegratedmodifiedSmithpredictorwithPIDcontrollerforintegratorplusdeadtimeprocess[J].IndustrialandEngineeringChemistryResearch,2006,45(4):1397-1407
[5] Qi Chunzi, Wu Hongxin, Lü Zhenduo. Research on stability of multivariable full-coefficient adaptive control system [J]. Control Theory and Applications, 2000, 17(4): 489-494
