1. Problems in Process Control Research
1.1 Insufficient integration of drying technology and control technology
The drying process is a typical multivariable, large inertia, highly nonlinear and complex system, making it difficult to establish an ideal mathematical model that conforms to the actual drying process; And building models requires a lot of effort, sometimes even impossible. Usually, for the convenience of research, the modeling conditions need to be simplified. The simplified equations cannot accurately reflect the drying process, and simplification often leads to errors. Some models, such as heat and mass transfer models, drying process optimization control models, fuzzy control and intelligent control models, have shortcomings. At the same time, the research on drying theory is limited to the circle of diffusion theory, and the characteristic function of the material itself has not been found, which also brings difficulties to the establishment of the model. Even if some drying processes can establish mathematical models, their structures are often complex and difficult to design and implement effective control. At present, research is mainly based on one-dimensional mathematical models for control, often only controlling a specific parameter, resulting in unsatisfactory control effects and inability to achieve multi-objective intelligent control. Without a good mathematical model, other indirect methods have to be sought when implementing control, which to some extent affects the accuracy and effectiveness of control. The combination of drying technology research and control technology research is not good enough, resulting in incomplete reflection of the role of dryer control in achieving the ideal efficiency of the dryer and improving product quality.
1.2 There is relatively little research on the control methods and control effects of the drying process
1.2.1 Less control variables in process control
The drying process control system is based on conventional univariate technology, and the control objective is mainly limited to the smooth operation of one or several variables to ensure smooth production and minimize accidents. With the increasing trend of large-scale, integrated, continuous, and complex grain drying industry, higher requirements have been put forward for the quality of process control. A good control system should not only protect the stability of the system and the safety of the entire production, meet certain constraints, but also bring certain economic and social benefits. In grain drying, once the temperature and humidity of the hot air in a certain drying section change, it not only directly affects the temperature and moisture content of the grain in that drying section, but also indirectly affects the temperature and moisture content of the grain in the next section and even the outlet of the drying tower. If the speed of the grain discharge motor slows down or accelerates, not only will the moisture content of the grain at the outlet of the drying tower change, but the temperature and moisture content of the grain in each drying section will also change accordingly. In this series of complex changes, there will inevitably be time delay, coupling, time-varying, and a series of nonlinear processes. If only the deviation and deviation change rate of the controlled variable are used as inputs to the control system, it is difficult to ensure its control effect when internal or external disturbances increase. Classic fuzzy control systems often simplify research problems into single input single output single variable fuzzy controllers, which have significant limitations in application. The input of the controller only includes the deviation and deviation changes of the controlled variable, essentially equivalent to a variable parameter single input PD regulator. Therefore, the complexity of the drying process determines that there is more than one control quantity and the controlled quantity, and there is a complex relationship between them. The ideal values of each controlled quantity will also have mutual constraints, making it difficult to find an ideal control scheme.
1.2.2 Advanced control applications are few and methods are centralized and single
Although the application of intelligent control in drying processes has been explored for decades, there has been little research on the design methods of advanced control systems for grain drying, and there has been a lot of research focused on certain methods. During the "15th Five Year Plan" period, the National Grain Administration spent a large amount of funds on solving the online testing and automatic control of moisture in the grain drying process, and conducted research and development work on some projects in conjunction with some grain depots. However, most design units adopted fuzzy control methods. By browsing domestic theses, it can also be seen that the majority of them use neural networks to establish mathematical models of drying towers, use fuzzy thinking to comprehensively evaluate the performance of dryers, and optimize the design of dryers; There is no report on the application of model predictive control. Although advanced control methods have many advantages, a single method also has various shortcomings. Fuzzy control is based on proficient operation and experience, and requires self-learning of the system to continuously adjust parameters in order to gradually approach the target value. However, there are many factors that affect the moisture content of grain during drying, making it difficult to find experienced parameters for skilled operators. Without using mathematical models that accurately reflect the control variables of the dryer for automatic control design, it is difficult to ensure the quality of grain after drying. Although adaptive control can solve uncertain problems to a certain extent, the algorithm is complex, computationally intensive, and has poor adaptability to unmodeled dynamics and disturbances in the process. The robustness of the system still needs to be further addressed, and its application is limited. Developing a system based on a user-friendly graphical interface is one of the development directions for drying process control. However, due to the long search time required for problem solving, the ability of the system to be used for online control is relatively poor. In the form of neural network modeling, networks based on BP algorithm have the disadvantages of long training time and frequent non convergence; Although using radial basis functions to approximate the drying process can greatly improve convergence speed and enable the network to converge globally, it is difficult to determine its center coordinates. Most existing nonlinear model predictive control methods can only be used for slow process control, which is disadvantageous for drying process control with high real-time requirements. Therefore, applying a single control strategy will inevitably fail to better leverage the advantages of process control.
1.3 More detection than control, low accuracy and stability of moisture sensors
The detection and control instruments for grain drying parameters are directly related to the quality and economic benefits of drying. There are not many applications of automatic control in domestic grain dryers. Some dryers are equipped with digital display of air temperature, over temperature alarm, and grain discharge speed display devices, but they cannot be automatically controlled. The domestic grain moisture detection instrument only measures and displays the moisture content of grains, without forming a real-time and online control system that is compatible with grain drying equipment, and cannot achieve automatic control of the grain drying process. Grain moisture testing is difficult to achieve online and rapid measurement. Currently, the drying equipment used in China cannot achieve automatic control of the grain drying process due to the lack of a standardized dynamic process moisture detection method. The accuracy and stability issues of online moisture testing sensors have not been well resolved, and have not truly matured to the stage of reliable detection, which has affected the accuracy of the process method.
2. Development directions
2.1 Improvement of Drying Process Model
Continue to conduct in-depth research on the internal heat and mass transfer laws of materials during the drying process; Establishing a mathematical model that can reflect the state of the drying process can help improve the automatic control of the drying process. At the same time, an intelligent model of the drying process can be established to replace mathematical models, and the intelligent control system can approach the real system and effectively control it. If artificial neural network technology is used to establish mathematical models, it can map multiple independent variables to multiple dependent variables, making it particularly suitable for complex grain drying processes.
2.2 Integration and penetration of multiple control methods
It is difficult to fully leverage the advantages of a single advanced control technology. An inevitable trend is for various control strategies to permeate each other, complement each other's strengths and weaknesses, and combine to form composite control strategies. The composite control strategy combining multiple control strategies overcomes the shortcomings of individual strategies and has better characteristics, which can better meet the requirements of different applications and is the future development direction. Research has shown that using neural networks instead of fuzzy mathematical reasoning methods will greatly improve the online control capability of the system; The neural network system that combines artificial neural networks with systems is a beneficial attempt for problem solving; The combination of neural networks and traditional control theory endows the control system with a considerable degree of intelligence. Therefore, the composite control strategy will promote the application of neural network control research, which is limited to mathematical simulation and laboratory research, to practical system control. Fuzzy PID explicit control, fuzzy variable structure control, adaptive fuzzy control, fuzzy predictive control, fuzzy neural network control, fuzzy control and other composite controls are emerging, and it is believed that they will have greater development and wide applications.
2.3 In depth study of control strategies
The design of the drying process system can no longer rely solely on traditional control theories and techniques based on quantitative mathematical models. It is necessary to further develop advanced process control systems, study advanced process control laws, and transplant and transform existing control theories and methods into the field of process control. These aspects are also receiving increasing attention from the control community. Further strengthen the research on control theory, such as focusing on the three major mechanisms of predictive control: predictive models, feedback correction methods, and optimization strategies, and comprehensively studying and breaking through them; There is an urgent need to develop real-time model predictive control methods for drying process control, which can reduce online calculation time while ensuring drying quality; Pay attention to interdisciplinary research, draw on other effective control methods, solve existing problems in process control, and continuously improve, develop, and innovate existing drying process control algorithms; Further improve the reliability of the automatic control system for drying quality and establish control algorithms with adaptive capabilities.
