Application of Fuzzy PID Control in DMF Recovery Control System

In the industrial wet synthetic leather production, dimethylformamide (DMF) plays an important role as a washing and curing agent. DMF is highly polluting. If it is lost into the atmosphere, water or soil, it will cause serious pollution to the environment. Therefore, the DMF lost on the production line today needs to be recycled.
The recovery of DMF adopts a multi-tower distillation process, which is a typical chemical distillation process, including primary and secondary concentration towers, distillation towers, evaporation tanks, etc. The process includes raw material preheating, negative pressure concentration, distillation, deacidification and other processes. In actual operation, the liquid level of the distillation tower fluctuates very quickly, and the influencing factors are very complex. Affected by the tower operating pressure, tower kettle heat, tower top reflux and feed and discharge, the process parameters are highly correlated and nonlinear, making it difficult to establish an accurate mathematical model of the controlled object, and conventional PID control is difficult to achieve real-time and effective control. In view of these characteristics, most companies can only use manual control output to solve the problem of unstable control.
In recent years, fuzzy control technology has developed rapidly and has been increasingly used in the field of industrial control. Since fuzzy control technology does not rely on the precise mathematical model of the object, it has strong robustness. Even if the control object does not have an accurate mathematical model, it can be stably controlled according to experience. In the actual distillation tower liquid level control of this system, there are overshoot and oscillation phenomena to varying degrees, and there is a problem of long adjustment time. Considering the characteristics of the controlled object such as nonlinearity, high order, large lag, and difficulty in determining the mathematical model, fuzzy control is suitable. However, fuzzy control itself also has certain limitations, such as poor steady-state performance. In order to solve these problems, the fuzzy controller needs to have the ability of self-learning and self-adjustment.
This paper designs a fuzzy-PID composite controller, which combines fuzzy control with PID algorithm using STEP7 to improve the control ability of nonlinear time-delay systems.
1 System composition
The hardware of the DMF recovery intelligent control system is mainly composed of SIMENSE PLC-300, industrial computer, liquid level sensor and other components. The software is written in STEP7, and the upper computer software uses Kingview software.
As shown in Figure 1, the system structure is divided into several parts: central processing unit (CPU), function module (FM), network communication module (CP), and signal module (SM). Among them, CP343 is a network communication module, responsible for communicating with the host computer; SM334 is an analog input/output module, responsible for collecting signals on site or giving output signals; SM321 is a digital input module, responsible for collecting digital signals on site; FM355C is an intelligent control module, which has the function of executing traditional PID algorithm and can realize PID control of the controlled object.

The liquid level of the distillation tower is collected by the liquid level sensor and converted into a 4 mA ~ 20 mA current signal and sent to the analog module SM334 of the PLC. The deviation liquid level E and the deviation change rate EC are calculated in the PLC and transmitted to the data block. The migration amount of each parameter of the PID controller is calculated by the fuzzy controller, and new PID parameters are formed in the PLC data block. Finally, the traditional PID algorithm is used to calculate and output a 4 mA ~ 20 mA signal to control the pneumatic valve, thereby achieving the effect of adjusting the liquid level.
2 Design of fuzzy controller
2.1 Fuzzification
The fuzzy PID parameter self-adjusting fuzzy controller is composed of three sub-fuzzy controllers. The input variables of each sub-fuzzy controller are the liquid level error E and the error change EC, and the output variables are △Kp, △Ki,
and △Kd respectively. Based on the analysis of field data and the control experience of liquid level, the domain of E and EC is designed to be [-6, +6], and the domain of the output variables △Kp, △Ki, and △Kd are [-10, +10], [-10, +10], and [-2, +2] respectively. The quantization level of fuzzy input and output is 7 levels, and the fuzzy set is defined as [NB, NM, NS, O, PS, PM, PB], which means negative large, negative medium, negative small, zero, positive small, positive medium, positive large [1-3]. The triangular function is used as the membership function to determine the membership of the fuzzy language variable, and the membership assignment table of each fuzzy variable can be obtained respectively. In this way, the precise quantity E and EC can be calculated and the corresponding fuzzy language variable input can be obtained.
2.2 Establishing fuzzy control rules
Based on data analysis and the experience of manual parameter adjustment by on-site operators, the control rules of each link are as follows.
(1) The role of the proportional link is to timely and proportionally reflect the deviation signal of the control system and immediately produce a control effect to reduce the deviation. Therefore, when the deviation is large, in order to improve the response speed, Kp should take a larger value; when the deviation is small and close to the steady state, in order to prevent overshoot and oscillation, Kp should take a smaller value. The fuzzy rule base of △Kp is shown in Table 1.

(2) The function of the integral link is to eliminate the static error. By integrating the error, there is a certain hysteresis effect on the system control. Therefore, when the deviation is large, Ki should take a smaller value to avoid excessive overshoot or system oscillation. When the error is small and close to the steady state, Ki should be increased appropriately to eliminate the steady-state error of the system and improve the control accuracy. The fuzzy rule base of △Ki is shown in Table 2.

(3) For controlled objects with large inertia and hysteresis, the differential link can predict the trend of error change. It can add an effective correction signal before the deviation signal value becomes too large, speed up the system response speed, and reduce the adjustment time. Since the differential link is more sensitive to interference signals, the value of Kd should comprehensively consider the system's response speed and anti-interference ability to improve the stability of the system. In this system, a larger Kd value should be taken when the error is large, and a smaller Kd value should be taken in the middle of control or close to steady state, thereby weakening the control effect of the process and increasing the ability to suppress disturbances. The fuzzy rule base of △Kd is shown in Table 3.

The control rules in this system adopt the Mamdani algorithm, which is a production rule based on IF-THEN. Its structure is simple and easy to modify.

2.3 Clarification
The clarification process is to convert fuzzy language variables into executable precise quantities, using the maximum membership method, that is, μ(u*)≥μ(u), u∈U, μ is the membership function of u, and u* is the value of the fuzzy control quantity corresponding to the maximum membership.
3 Software design implemented by STEP7
The programming software STEP7 of SIMENSE S7-300 PLC provides a wealth of extended function modules, which provides convenient conditions for implementing algorithms of various functions [8]. In the STEP7 project, the main program module OB1 implements the call and data transfer of the subroutine module and is constantly refreshed. OB35 is the interrupt service program module used to respond to system interrupts. The OB100 module is the system initialization module, which automatically runs when the system is powered on to initialize various parameters. In the STEP7 project, the function FB1 module is written as the main fuzzy controller, and the sub-functions FC1~FC4 are written to complete the entire fuzzy control function. The fuzzy control query table of each sub-fuzzy controller is manually entered into DB5-DB7. Among them, FC1 is responsible for calculating the liquid level deviation E and the deviation change rate EC, FC2 is responsible for fuzzifying E and EC and converting the precise numerical value into fuzzy language variables, FC3 is responsible for online query, by looking up the table in the DB block, according to the input of the fuzzy language variable, the corresponding fuzzy language variable output is obtained, and FC4 is responsible for clarifying the output of the fuzzy language variable and converting it into precise values ​​△Kp, △Ki, △Kd. When the system is in automatic control state, FB1 calls FC1, FC2, FC3, and FC4 respectively to complete the calculation of the fuzzy control function. After clarification, the three parameters are adaptively corrected according to the following method:
Kp=Kp+△Kp, Ki=Ki+△Ki, Kd=Kd+△Kd.
The PID parameters Kp, Ki, and Kd corrected by fuzzy and reasoning are stored in the PLC data block and applied to the ordinary PID algorithm through the FM355 module to form a fuzzy PID algorithm with adaptive function.
The flow chart of the fuzzy PID algorithm software design is shown in Figure 3.

4 Analysis of system operation effect
After applying the fuzzy PID control algorithm, the control effect of the DMF recovery control system on the liquid level of the distillation tower has been significantly improved compared with before.
From the control curve, it can be seen that the overshoot is small and the time required to reach steady state is short when the fuzzy PID self-adjusting control method is used. The fuzzy PID self-adjusting control method combines fuzzy control and PID control to adjust the parameters online. When the initial deviation is relatively large, the proportional constant Kp is automatically increased to improve the response time of the system; when the deviation is small, the integral constant Ki is appropriately increased to eliminate the static error, reduce overshoot, and shorten the steady-state time, thereby improving the control accuracy of the system and improving the dynamic performance. As shown in Figures 4 and 5.

Fuzzy reasoning is used to imitate the logical thinking of the human brain and to deal with control problems with unknown or inaccurate models. When using Siemens series PLC to implement fuzzy control, special programming equipment is required, which is expensive and complicated to use. The fuzzy controller implemented by software methods in this paper greatly saves enterprise costs and realizes full automation of production. The system that must be manually operated and output due to unstable control has been eliminated. In addition, the query table is generated by offline calculation using the fuzzy rule library, and it is directly queried online in the system, which optimizes the system's operating speed and has a high engineering application value. This control method has now been applied to the DMF recovery control system of a chemical company. The control has been significantly improved compared with the past, greatly reducing the system adjustment time and overshoot. The control is stable and accurate, bringing great benefits to the company.
References
[1] Wang Yaonan, Sun Wei. Intelligent control theory and application [M]. Beijing: Machinery Industry Press, 2008: 31-46.
[2] Han Junfeng, Li Yuhui. Fuzzy control technology [M]. Chongqing: Chongqing University Press, 2003: 8-56.
[3] Shan Dong. Fuzzy control principle and application [M]. Beijing: China Railway Publishing House, 1995: 54-128.
[4] Zhang Qing, Cheng Weiming, Shen Yaozong, et al. Application of fuzzy control in rubber vulcanization temperature control [J]. Optical Precision Engineering, 2001, 9(4): 385-387.
[5] Wang Shuqing, Dai Liankui, Zhu Heyun, et al. Industrial process control [M]. Beijing: Chemical Industry Press, 2005.
[6] Zhang Xiang, Xiao Jian. Research on TS fuzzy control of nonlinear network control system [J]. Journal of Hefei University of Technology: Natural Science Edition, 2010, 33(7): 1010-1014.
[7] Li Qingchun, Shen Deyao. A PID fuzzy controller (fuzzy PI+fuzzy ID type) [J]. Control and Decision, 2009, 24(7): 1038-1042.
[8] Tian Yuan, Liu Zhenjuan. Implementation of fuzzy control in SIMENSE PLC system [J]. China Instruments, 2005(5): 72-75.