A cascade control method for tobacco redrying equipment

Abstract

The application discloses a cascade control method for tobacco redrying equipment, and is based on physical mechanism, and establishes temperature and humidity models respectively taking the redrying machine as an object. In terms of temperature, according to the law of conservation of energy, the environment in the cylinder is divided into internal air and tobacco material double control bodies, and a linear double-state space model containing hot air, steam and wall temperature input is constructed. In terms of humidity, according to the law of conservation of mass, the influence of humidification water spraying and steam condensate water is comprehensively considered, and a nonlinear evolution model introducing tobacco flow disturbance is constructed. The method solves the problem that the traditional black box model in the industry lacks physical meaning and is too dependent on experience, and reduces the degree of dependence on artificial experience. Meanwhile, the application establishes a temperature dominant and humidity collaborative cascade control mechanism, solves the multi-variable coupling conflict, realizes the synchronous and rapid reaching of the temperature and humidity double indexes, and greatly shortens the adjustment time.

Description

Technical Field

[0001] This invention relates to the field of automatic control, and in particular to a cascade control method for tobacco leaf re-drying and moistening equipment. Background Technology

[0002] Tobacco re-drying is a crucial step in tobacco production. Its main purpose is to adjust the moisture and temperature of the tobacco leaves, improve their processing resistance and filling properties, and remove insect eggs and mold, thus providing qualified raw materials for subsequent aging and tobacco processing. In the re-drying process, the leaf-moistening step is the core step in controlling the moisture and temperature at the exit of the tobacco pretreatment section. This step is usually carried out in a drum-type hot air leaf-moistening machine. By injecting steam, applying humidifying water, and introducing hot air into the drum, the loose tobacco leaves undergo sufficient heat and mass exchange with the humid and hot environment during tumbling, thereby achieving the predetermined process standards.

[0003] The leaf-moistening process is primarily completed in a drum-type hot air leaf-moistening machine, a large-scale horizontal rotating drum device. Its working principle involves multiple heat and mass exchange processes between the gas and solid phases. During production, loose tobacco leaves enter the drum through the feed inlet. As the drum rotates, the leaves are continuously thrown up and scattered by the lifting plates, moving towards the discharge outlet. During this process, the system injects energy and mass into the drum through various actuators: First, the steam system plays a dominant role. Steam is divided into multiple streams: one stream heats circulating air through a heat exchanger to create hot air, which is then delivered into the drum through ducts to maintain the ambient temperature; another stream is directly injected into the drum, providing both heat and some moisture; and a third stream heats the walls of the discharge chamber to prevent condensation dripping and assist in insulation. Second, the humidification water system atomizes water through high-pressure nozzles, spraying it directly onto the tobacco leaf surface or atomizing it in the air, primarily responsible for regulating the moisture content of the tobacco leaves. Finally, the dehumidification system is responsible for expelling excess hot and humid waste gas to maintain pressure balance within the drum.

[0004] Currently, most domestic re-drying enterprises still rely on PID (Proportional-Integral-Derivative) technology based on error feedback for adjustment in the leaf-curing process. The on-site system typically only focuses on the fuzzy correspondence between parameters and indicators, treating the leaf-curing machine as a black box of input and output. While this method has become the standard process formula due to its simple structure and ease of implementation in long-term production practice, it essentially avoids a mathematical description of the internal physical processes of the leaf-curing machine. Operators can only set parameters based on experience, relying on post-production deviations to maintain relative stability in the production process.

[0005] The humidity dynamics within a leaf humidifier are influenced by various factors. Due to the high ambient humidity and the enclosed structure of the humidifier, the evaporation effect caused by temperature is negligible. Therefore, the temperature and humidity system can be approximated as a decoupled system. Cascade control is based on this physical reality and employs a step-by-step control strategy: first, the steam flow rate is stabilized through the temperature control subsystem to maintain a constant flow during the humidity control phase; then, humidity control is achieved by adjusting only the humidifying water flow rate. This decoupling method simplifies control design and avoids the complexity caused by multi-variable coupling.

[0006] Prior art related to the present invention includes: Patent: Modeling method and apparatus for a tobacco re-drying production line; Application publication number: CN119599533A.

[0007] References: [1] He Guangyu. Research on active disturbance rejection control algorithm of hot air leaf humidifier for leaf re-baking [D]. Changsha: Hunan University, 2019. [2] Zhang Liping, Li Yingna, Yang Yi, et al. Optimization of control parameters for leaf-moistening machine based on improved gray wolf algorithm [J]. Information and Control, 2024, 53(6): 793-803. [3] Li Bingbing, Zhuang Weichao, Liu Haoji, et al. Research on Ecological Driving Control Strategy of Connected Electric Vehicles Based on Self-Learning MPC [J]. Journal of Mechanical Engineering, 2024, 60(10): 453-462. [4] Quality requirements for threshed tobacco leaves: YC / T 147—2023[S]. 2023. [5] Xu Yizhou. Research and design of intelligent control system for tobacco leaf moisture [D]. Shanghai: Donghua University, 2008. [6] Hong Juchen. Research and application of complex control strategies in hot air leaf humidification system [D]. Shanghai: Donghua University, 2022. The main technical problems of existing technologies are: The lack of accurate physical mechanism models and theoretical support leads to a mismatch between energy and mass supply.

[0008] As the tobacco industry continues to demand more refined processing and stricter requirements for product quality homogenization, the limitations of existing technologies lacking clear physical mechanism models are becoming increasingly apparent. Because the industry has not yet established a mathematical model that can accurately describe the energy and mass transfer processes within the leaf-conditioning machine, it is impossible to fundamentally grasp the dynamic evolution of the system when faced with the inherent large hysteresis, strong coupling, and nonlinear physical characteristics of the leaf-conditioning machine, severely impacting the stability of tobacco leaf quality.

[0009] Secondary technical issues: The strong physical coupling mechanism of temperature and humidity was ignored, and there was interference between multi-loop control systems.

[0010] Physically, steam injection has a dual effect: it provides both heat and moisture. Existing control strategies typically only perform simple feedforward compensation. When the temperature loop increases steam to raise the temperature, it inevitably leads to a passive increase in humidity. At this point, the humidity loop detects excessive moisture and reduces the amount of water added, which in turn causes the temperature to rise further. This control method, lacking physical-level coordination, results in the two controllers not working together during the adjustment process. This often leads to mutual cancellation and interference of control actions, making it difficult for the system to stabilize quickly and causing significant energy waste. Summary of the Invention

[0011] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0012] This invention discloses a cascade control method for tobacco leaf re-drying and moistening equipment, comprising the following steps: Step S1: Leaf wetting process management and physical mechanism analysis to determine the modeling object of the leaf wetting machine; Step S2: System boundary determination and variable definition; Step S3: Temperature state-space modeling based on energy conservation; Step S4: Nonlinear modeling of humidity based on mass conservation; Step S5: Design of cascade MPC controller. Considering the physical relationship between the thermodynamic response and humidity response of the leaf humidifier, a temperature-priority and humidity-coordinated control strategy is adopted. Design a temperature controller based on linear model predictive control, a cascade information transmission control, and a humidity controller based on nonlinear model predictive control. Step S6: Online rolling optimization and feedback execution. The cascade MPC control strategy is deployed in the execution phase of the leaf humidifier's field control system. The system runs online periodically based on the sampling period. Through a closed-loop design of real-time sampling, status observation, rolling optimization, and control issuance, precise control of the temperature and humidity of the leaf humidifier is achieved.

[0013] Preferably, step S1 includes: clarifying the working principle, physical structural boundaries, and energy and mass transmission paths of the leaf humidifier; the leaf humidifier includes a feeding chamber, a rotating drum, a discharging chamber, a frame, a hot air system, a pipeline system, and a dehumidification system, and the modeling object is the internal space of the drum of the leaf humidifier and its associated actuators.

[0014] Preferably, step S2 includes: analyzing the energy flow and mass flow transmission mechanism inside the leaf-humidifying machine system, thereby dividing the thermodynamic boundary and control volume of the system, and clearly defining the state variables, control input variables and disturbance variables.

[0015] Preferably, step S3 includes: analyzing the thermodynamic mechanism inside the leaf-humidifying machine, constructing a set of differential equations describing the dynamic characteristics of the system, establishing energy conservation equations for the internal air and tobacco material respectively, and converting them into a standard state-space form.

[0016] Preferably, step S4 includes: establishing a dynamic model for the moisture content of tobacco leaves at the outlet of the leaf conditioning machine; the humidity change process is affected by the definition of wet basis moisture content and material flow rate, exhibiting significant nonlinear characteristics, thereby analyzing the dynamic process of moisture addition and mixing based on the law of conservation of mass.

[0017] Preferably, in step S5, designing a temperature LMPC controller based on minimizing temperature deviation and energy consumption includes: Using the discrete state-space equation obtained in step S3 as the prediction model, and setting the prediction time domain as P and the control time domain as N, an optimization objective function of the form of quadratic programming QP is constructed to minimize the temperature tracking error and the rate of change of the control quantity. Under the premise of satisfying the equipment constraints, the optimal hot air temperature and steam flow rate are calculated to maintain the stability of the outlet tobacco temperature.

[0018] Preferably, in step S5, a cascade calculation path based on the steam flow parameter transmission is designed: a) Within each control cycle, the temperature LMPC controller will be operated first; b) Parameter locking and transmission: After solving the above QP problem, the temperature LMPC controller obtains the optimal control sequence for future time moments, from which the calculated optimal steam flow sequence is extracted; c) Feedforward compensation: The optimal steam flow sequence is sent as an instruction to the actuator and simultaneously transmitted in real time to the humidity NMPC controller as a known look-ahead quantity. This allows the humidity controller to predict the amount of moisture increase brought by the steam when calculating the water addition, thereby achieving precise coordinated control.

[0019] Preferably, in step S5, the design of a humidity NMPC controller based on minimizing the deviation between humidity and water addition includes: Using the nonlinear mass conservation model constructed in step S4, a humidity prediction controller is designed. The nonlinear difference equation obtained in step S4 is used as the prediction kernel. The objective function is defined, and the nonlinear prediction model in step S4 is expanded using Taylor series. The nonlinear prediction model near the operating point is approximated with a first-order linearization. The Jacobian matrix is ​​calculated, and the nonlinear model is transformed into a time-varying linear approximation model. The original nonlinear objective function is approximated as a standard quadratic form, and a quadratic programming problem is constructed for solution.

[0020] Preferably, step S6 includes: (1) Real-time sampling and data preprocessing At the beginning of each control cycle, real-time operating data of the leaf-wetting machine is collected through field sensors. The collected raw signals are then processed by moving average filtering to remove high-frequency noise and obtain the observation values ​​used for calculation. (2) State observation and feedback correction The internal state of the model is corrected by a feedback correction loop to ensure the accuracy of the prediction. For the temperature linear model, a discrete Kalman filter is designed. The deviation between the measured outlet temperature and the model prediction is used, combined with the Kalman gain matrix, to make the optimal estimate of the dual state vector of the system. For the nonlinear humidity model, the deviation between the measured moisture content and the model prediction is calculated. This deviation is then applied as an additive perturbation to the subsequent prediction trajectory to shift and correct the future moisture prediction values, thereby eliminating steady-state errors. (3) Cascade rolling optimization solution Based on the corrected state and current operating parameters, the optimization problem is solved in cascade order to calculate the optimal control quantity at the current moment.

[0021] Preferably, the cascade rolling optimization solution process includes: (1) Real-time sampling and data preprocessing At the beginning of each control cycle, real-time operating data of the leaf-wetting machine is collected through field sensors. The collected raw signals are then processed by moving average filtering to remove high-frequency noise and obtain the observation values ​​used for calculation. (2) State observation and feedback correction The internal state of the model is corrected by a feedback correction loop to ensure the accuracy of the prediction. For the temperature linear model, a discrete Kalman filter is designed. The deviation between the measured outlet temperature and the model prediction is used, combined with the Kalman gain matrix, to make the optimal estimate of the dual state vector of the system. For the nonlinear humidity model, the deviation between the measured moisture content and the model prediction is calculated. This deviation is then applied as an additive perturbation to the subsequent prediction trajectory to shift and correct the future moisture prediction values, thereby eliminating steady-state errors. (3) Cascade rolling optimization solution Based on the corrected state and current operating parameters, the optimization problem is solved in cascade order to calculate the optimal control quantity at the current moment.

[0022] Preferably, the cascade rolling optimization solution process includes: 1) Solving for temperature using LMPC Substitute the corrected temperature state into the temperature LMPC controller, call the online quadratic programming solver, and under the premise of satisfying the valve opening and incremental constraints, solve the control increment sequence in the future control time domain, extract the first element of the optimal sequence, and obtain the optimal hot air temperature, steam flow rate, and discharge chamber wall temperature required at the current moment. 2) Parameter transfer and model update The calculated steam flow rate is locked and input into the humidity model along with the real-time collected tobacco flow rate. Using the Taylor expansion method, the Jacobian matrix of the humidity model is updated according to the current steam flow rate and tobacco flow rate to complete the local linearization of the nonlinear model. 3) Solve for humidity NMPC Based on the updated linearized humidity model, a quadratic programming subproblem is constructed, and the QP solver is called to calculate the optimal mechanical water supply flow rate required to achieve the target humidity after deducting the contribution of steam condensate. 4) Control command output and execution Based on the rolling time domain control principle, only the calculated control quantity at the current moment is used as the final command output. The calculated physical quantity is converted into a standard signal that can be recognized by the field actuator. Before the signal is issued, an engineering limit check is performed to prevent the actuator from overtravel or malfunction due to calculation abnormalities. 5) Control command output and execution After completing this control action, the system enters the next sampling time k+1 and repeats the steps 1) to 4) above. Through online rolling optimization, the control system can sense the fluctuations in tobacco leaf flow and changes in steam pressure in real time, and adjust the control strategy in an instant, thereby maintaining a high degree of stability in the temperature and humidity at the outlet of the leaf humidifier throughout the entire production process.

[0023] The present invention has at least the following beneficial effects: 1. This invention, based on the laws of energy and mass conservation, constructs a set of differential equations with clear physical meaning. This method overcomes the lack of physical model support in existing technologies by constructing a leaf-humidifying mechanism model based on energy and mass conservation, transforming the leaf-humidifying machine from a black box into a white box system with a visible internal state. The established dual-control volume temperature state-space model, by calculating the internal air temperature state variable, can quantify the energy accumulation within the system in real time. The model can accurately calculate the internal energy accumulation rate using differential equations based on the current steam and hot air input. The nonlinear humidity model established based on mass conservation clarifies the mechanism by which key parameters affect tobacco leaf moisture changes, reducing the industry's reliance on manual experience.

[0024] 2. This invention establishes a temperature-driven, humidity-coordinated cascade control mechanism, resolving multivariate coupling conflicts. This method utilizes a mechanistic model to clarify the dual role of steam flow rate in the energy and mass equations and designs a cascade MPC control architecture. The system prioritizes calculating the steam flow rate that meets the temperature requirements and transmits this as known look-ahead information to the humidity controller in real time. This cascade control strategy based on physical mechanisms eliminates conflicts between control loops, achieving simultaneous and rapid attainment of both temperature and humidity targets, significantly shortening the adjustment time.

[0025] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the control method principle framework of the present invention. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0028] It should be understood that terms such as “having,” “comprising,” and “including” as used herein do not imply the presence or addition of one or more other elements or combinations thereof.

[0029] See Figure 1 This invention discloses a modeling method for cascade control of tobacco leaf re-drying and humidifying equipment. Based on physical mechanisms, this method establishes temperature and humidity models for the humidifying machine: Regarding temperature, according to the law of conservation of energy, the internal environment is divided into two control volumes: internal air and tobacco leaf material, constructing a linear dual-state space model including hot air, steam, and wall temperature inputs; regarding humidity, according to the law of conservation of mass, considering the effects of humidifying water spraying and steam condensate, a nonlinear evolution model incorporating tobacco leaf flow disturbances is constructed. This method solves the problem of traditional black-box models lacking physical meaning and relying too heavily on experience, reducing the dependence on human experience.

[0030] The steps to achieve the invention include: Step S1: Leaf-wetting process management and physical mechanism analysis This step forms the basis for establishing a physical mechanism model. Through in-depth interpretation of the equipment manual for the WF3524 drum-type hot air leaf humidifier (hereinafter referred to as the leaf humidifier), combined with on-site investigation of the production process, the working principle, physical structural boundaries, and energy and mass transmission paths of the system are clarified.

[0031] On-site investigation and equipment manual analysis: According to the leaf humidifier's instruction manual, the modeling object is clearly defined as the internal space of the leaf humidifier's drum and its associated actuators. The leaf humidifier mainly consists of a feeding chamber, a rotating drum, a discharging chamber, a frame, a hot air system, a piping system, and a dehumidification system.

[0032] The tobacco leaves enter through the feed chamber and, under the action of the lifting plates and pushing rods on the inner wall of the drum, are continuously thrown up and scattered as the drum rotates, forming a loose curtain of material that moves towards the discharge end. This motion characteristic determines that the tobacco leaves have a large contact area with the environment inside the drum, providing the physical conditions for heat and mass exchange. The hot air system consists of an air heater, a circulating fan, and air ducts. The circulating fan sends hot air, heated by the heater, into the drum from the feed end, flowing in the same direction as the material. This constitutes the most important convective heat exchange environment of the system, and the hot air temperature is the key variable for controlling the temperature inside the drum. Regarding the discharge chamber, the instruction manual mentions that heating coils are installed around the discharge chamber, which plays an important role in maintaining the discharge outlet temperature and preventing condensate dripping; these should be considered as an auxiliary heating source for the tobacco leaves during modeling.

[0033] The system's input sources are divided into steam and water paths.

[0034] After the total steam is branched, there are four main routes that affect the process: Air heating path: Steam enters the heat exchanger through a regulating valve, heating the circulating air to the set temperature. This path only provides sensible heat and does not increase the moisture content inside the cylinder.

[0035] Direct injection path inside the drum: After being depressurized, steam is directly injected into the inside of the drum through a nozzle. In this path, the steam comes into direct contact with the tobacco leaves and air, which both releases enthalpy and increases temperature, and increases moisture content due to condensation.

[0036] Steam-water mixing atomization path: Steam serves as the power source, atomizing the humidifying water in a dual-media nozzle. In this path, the steam primarily functions as an atomizer, but it also enters the cylinder to participate in the heat and mass balance.

[0037] Heat tracing and insulation path: Steam is introduced into the feed chamber housing and the discharge chamber coils. This path mainly heats the walls through heat conduction and radiation, thereby heating the tobacco leaves flowing through the discharge port.

[0038] Water supply pipeline: After humidification, the water is filtered and metered, and then sprayed into the front and rear chambers of the cylinder through a dual-medium air-water nozzle.

[0039] Step S2: System boundary determination and variable definition Based on the preliminary investigation of equipment structure and process flow in step S1, this step will analyze the energy flow and mass flow transmission mechanism inside the system in depth, and scientifically divide the thermodynamic boundary and control volume of the system accordingly, and clearly define the state variables, control input variables and disturbance variables.

[0040] (1) Analysis of energy and mass transport paths Based on a clear understanding of the physical structure, this invention further isolates the energy flow of the thermodynamic process and the mass flow of the mass transfer process during the leaf-wetting process, which serve as the core physical basis for constructing the state-space equations.

[0041] System thermodynamic boundary setting Internal air zone: includes free-flowing hot air, flash steam, and suspended water mist within the drum; Tobacco material zone: includes all tobacco leaves retained within the drum and the moisture adhering to their surfaces.

[0042] Since the outer wall of the drum is covered with an insulation layer of not less than 30mm thick and the discharge chamber has a heat tracing coil, the heat radiation loss from the outer wall of the drum to the external environment is ignored when modeling. An insulation assumption is made, and it is believed that the heat exchange between the system and the outside world is mainly carried out through the inlet and outlet material flow and the exhaust air flow.

[0043] Energy transfer pathway (temperature model mechanism) Heating Path 1: Gas-to-Solid Phase Strong Convection Heat Transfer. Hot air generated by the heat exchanger enters the drum, raising its temperature. Since the air temperature is higher than the tobacco leaf temperature, heat is transferred to the tobacco leaves. This process has significant thermal inertia and is affected by drum rotation speed and filling rate.

[0044] Heating Path 2: Release of Latent Heat of Vapor. High-enthalpy saturated steam, introduced by direct injection steam and atomized steam, is directly injected into the cylinder. Part of the steam mixes with air to raise the air temperature; the other part directly contacts the cooler tobacco leaf surface, undergoing a phase change (condensation) and releasing a huge amount of latent heat of vaporization.

[0045] Heating Path 3: Wall Conduction and Radiation. The metal walls of the discharge chamber are heated by tracing steam. As the tobacco leaves flow through the discharge end, they gain heat through conduction and radiation from the heated walls. This heat primarily corrects the temperature of the tobacco leaves at the outlet, having a relatively small impact on the internal air temperature.

[0046] Quality transport path (humidity model mechanism) The changes in wet substrate moisture content within the leaf-conditioning machine are primarily influenced by water addition and phase change processes. Based on field surveys, the leaf-conditioning machine operates in a high-humidity environment. Therefore, this model, based on the physical assumption that high humidity inhibits evaporation, ignores the reduction in moisture evaporation from the tobacco leaf surface and focuses on the pathways of moisture increase: Humidification Path 1: Mechanical Humidification. Humidifying water added through the water supply pipeline is atomized into droplets. Some of these droplets settle directly onto the surface of the tobacco leaves and are absorbed; the rest remain suspended in the air and are gradually captured as the tobacco leaves tumble.

[0047] Humidification Path 2: Phase Change Humidification. Directly injected steam and atomized steam condense upon entering the cylinder, producing condensate for humidification.

[0048] (2) System control body Based on the above path analysis, in order to make the model both reflect the physical essence and facilitate calculation, it is necessary to abstractly define the physical system. To accurately describe the energy and mass transfer paths, this invention divides the internal space of the leaf-wetting machine drum into different interacting control volumes: Vapor phase control volume: includes free air, suspended vapor and unsettled water mist inside the drum.

[0049] Solid phase control volume: includes all tobacco leaf material retained inside the drum. Liquid phase control volume: contains the liquid inside the drum.

[0050] It is assumed that the energy exchange between the gas phase control volume and the solid phase control volume is calculated using the gas-solid heat transfer coefficient kex.

[0051] (3) Variable definition Output variable Y: y1 represents the outlet tobacco leaf temperature (°C); y2 represents the moisture content (%) of the exported tobacco leaves, expressed on a wet basis, which is the ratio of water mass to the total mass of wet tobacco leaves.

[0052] Control input variable U: u1: Hot air temperature (°C), which drives convective heat transfer by changing the enthalpy of the air entering the control body; u2: Steam flow rate (kg / h), corresponding to direct steam injection and atomization sections; u3: Discharge chamber temperature increase / wall temperature (°C), which corrects the tobacco leaf temperature at the discharge end through contact heat conduction. This variable is localized and is mainly used for fine-tuning the outlet temperature, with little impact on the overall internal environment.

[0053] u4: Water flow rate (kg / h), corresponding to the sum of water added to the front and rear chambers. Mechanical humidification directly increases the water quality within the system. State variables: x1: Return air temperature (°C), which is the temperature at which the internal air returns to the circulating air duct after passing through the humidifier. It can be approximated as the average temperature of the internal air. x2: Average temperature of tobacco leaves inside the machine (°C); M: Average wet basis moisture content of tobacco leaves in the machine (%).

[0054] Step S3: Temperature state-space modeling based on energy conservation Based on the system boundaries and variable definitions determined in step S2, this step delves into the thermodynamic mechanisms within the leaf-humidifying machine, constructing a set of differential equations describing the system's dynamic characteristics. This invention establishes energy conservation equations for both the internal air and the tobacco material, transforming them into a standard state-space form.

[0055] (1) Construction of the differential equation for the conservation of internal air energy First, the gas-phase control volume inside the leaf conditioner drum is analyzed. The return air temperature x1 is selected as the state variable of this control volume. According to the first law of thermodynamics, the rate of change of internal energy of the gas-phase control volume is equal to the algebraic sum of the input heat power and the output heat power. The main energy terms acting on the gas-phase control volume are as follows: Hot air convection gain term: The hot air injected by the hot air system is the main energy source. The heat brought by the hot air temperature u1 is proportional to the temperature difference, and the gain coefficient is defined as α1. This coefficient is related to the air volume of the circulating fan and the specific heat capacity of the air.

[0056] Steam enthalpy injection term: Steam entering through direct injection and atomizing nozzles brings high-enthalpy saturated steam into the gas phase control volume. This energy input is directly proportional to the steam flow rate, and the gain coefficient is defined as α2. When constructing the steam energy input term, considering that saturated steam contains a huge latent heat of vaporization, which is much greater than the sensible enthalpy corresponding to the internal temperature of the leaf conditioner and the temperature of the tobacco leaves, and that the influence of internal temperature fluctuations on the net heat release of steam is a high-order minor quantity, this invention approximates that the energy injected by steam is mainly proportional to the steam mass flow rate.

[0057] Gas-solid convective heat dissipation term: Convective heat transfer occurs between the internal air and the swirling tobacco leaves. Heat is transferred from the high-temperature air to the low-temperature tobacco leaves, resulting in a loss of air energy. This loss is proportional to the gas-solid temperature difference, and the heat transfer coefficient is defined as g. ex .

[0058] Exhaust overflow dissipation: To maintain pressure balance inside the cylinder, an equal amount of humid and hot gas is inevitably discharged (moisture removal) along with the injection of steam. This overflow gas carries away the enthalpy based on the current temperature. Mathematically, this loss is reflected in the physical causes contained in the coefficient of the steam flow coefficient in the negative feedback effect of the state.

[0059] Based on the above analysis, the energy conservation differential equation of the internal air is constructed as shown in Equation (1).

[0060] (1) By arranging and merging like terms, separating state variables and control variables, the standardized differential equation is obtained as shown in formula (2): (2) (2) Construction of the differential equation for energy conservation of tobacco materials Secondly, the solid-phase control volume inside the leaf-conditioning machine drum is analyzed. The outlet tobacco leaf temperature x2 is selected as the state variable of this control volume. The main energy terms acting on the solid-phase control volume are as follows: Gas-solid convection endothermic term: The tobacco leaves absorb heat from the air, and its value is equal to but opposite in sign to the dissipation term in formula (1), i.e., g ex (x1-x2).

[0061] Steam condensation heat release term: A portion of the steam flow directly contacts the lower-temperature tobacco leaf surface, undergoing phase change condensation and releasing latent heat of vaporization. This energy is directly absorbed by the tobacco leaf and is proportional to the flow rate; the latent heat utilization coefficient is defined as β1.

[0062] Wall contact / radiative heat transfer: The tracing steam in the discharge chamber heats the wall of the discharge chamber to a temperature that allows the tobacco leaves to exchange heat with the wall through conduction and radiation as they flow through the discharge end. This heat flux is proportional to the temperature difference, and the wall heat transfer coefficient is defined as β².

[0063] Based on the above analysis, the energy conservation differential equation for tobacco materials is constructed as shown in equation (3): (3) The standardized differential equation obtained after simplification is shown in equation (4): (4) (3) Establishment of continuous-time state-space model Combining equations (2) and (4), we select the state vector x = [x1, x2]. T These correspond to the return air temperature and the tobacco leaf temperature, respectively; the control input vector is u = [u1, u2, u3]. T These correspond to hot air temperature, steam flow rate, and discharge chamber temperature increase, respectively. A linear continuous-time state-space equation is constructed: (5) The observation equation is established by selecting the actual outlet temperature of the tobacco leaf, x2, as the system output y: (6) In this model, matrix A is the system matrix, matrix B is the input matrix, and C is the output matrix.

[0064] (4) Discretization of the model If the model needs to be discretized, the following methods can be used to process the continuous state equations.

[0065] Set the sampling period to T s Using the first-order Euler approximation, the derivative term can be approximated as: (7) Substituting into formula (5) and simplifying, we can obtain the discrete-time state-space model: (8) Let the discrete system matrix A d= I+A∙T s Discrete input matrix B d = B∙T s I is the identity matrix, and the final iterative model used to control the algorithm is: (9) (10) Step S4: Nonlinear modeling of humidity based on mass conservation This step establishes a dynamic model for the moisture content of tobacco leaves at the humidifier outlet. The humidity change process is influenced by the definition of wet basis moisture content and material flow rate, exhibiting significant nonlinear characteristics. Based on the law of conservation of mass, this invention neglects moisture evaporation under high humidity conditions and focuses on analyzing the dynamic process of moisture addition and mixing.

[0066] (1) Analysis of water sources and definition of variables First, source term analysis was performed on the mass flow affecting the moisture content change in the liquid control volume: The moisture gain term includes mechanically added water and steam condensate. Mechanically added water is defined as u4 (unit: kg), which is liquid water directly injected through the humidification water pipeline. Its contribution to moisture depends on the atomization efficiency and the tobacco leaf absorption rate, and is defined by the coefficient η. w High-temperature steam releases latent heat when heating materials, and condensation occurs during the phase change. The steam condensation coefficient is defined as η. s .

[0067] Moisture loss is the moisture carried away by exported tobacco leaves.

[0068] The tobacco industry uses wet basis moisture content, which is defined as M in this invention, and its calculation method is shown in formula (11).

[0069] (11) Where, m w The moisture content of tobacco leaves in the humidification drum, m d Let m be the mass of dry tobacco leaves in the wetting drum. Assume the tobacco leaf flow rate is constant during the wetting process, i.e., m. d The rate of change is approximately 0.

[0070] Total material retention m in the leaf moistening machine total Its instantaneous flow rate F of tobacco leaves tob The dynamic changes are calculated as shown in formula (12).

[0071] (12) (2) Construction of the differential equation for mass conservation Based on the above analysis, a humidity evolution model is derived. According to the law of conservation of mass: (13) in Substitute the moisture change rate for the ingested tobacco leaves. This represents the rate of change in moisture content carried over from exports. Add the total moisture change rate to the system.

[0072] Substitute specific physical quantities: (14) Where F tob M in F introduces moisture into the incoming tobacco leaves tob M out To remove moisture at the outlet, η w u4+η s u2 adds total water to the system.

[0073] Based on the definition of moisture content on a wet basis, formula (11) is used to differentiate it: (15) Because m d The rate of change is approximately 0, and formula (15) can be simplified to: (16) Substituting equation (14) into equation (16), we obtain the nonlinear differential equation describing the change in outlet moisture content: (17) (3) Discretization of the model To be suitable for digital control systems, it is possible to utilize Where τ is the average dwell time, the first-order Euler method is used, and the sampling period is T. s : (18) Among them, M k Let F be the moisture content of the wet basis at time k. tob (k) represents the instantaneous flow rate of tobacco leaves at time k, M in,k Let be the wet basis moisture content of the inlet tobacco leaves at time k. After simplification, the final nonlinear iterative model can be obtained: (19) Step S5: Design of Cascaded MPC Controller Based on the mechanistic model set generated in the previous steps, this step designs a cascade model predictive control architecture. This architecture considers the physical relationship between the thermodynamic and humidity responses of the leaf conditioner, employing a temperature-priority, humidity-coordinated control strategy. The entire controller design comprises three logical parts: a temperature controller design based on Linear Model Predictive Control (LMPC), a cascade information transmission mechanism, and a humidity controller design based on Nonlinear Model Predictive Control (NMPC).

[0074] (1) Temperature LMPC controller based on minimizing temperature deviation and energy consumption The linear discrete state-space model constructed using step S3 Design a temperature predictive controller. The core task of this controller is to calculate the optimal hot air temperature and steam flow rate to maintain a stable outlet tobacco leaf temperature while meeting equipment constraints.

[0075] The discrete state-space equation obtained in step S3 is used as the prediction model, with the prediction time domain set to P (20-50 sampling periods) and the control time domain set to N (5-10 sampling periods). Then, a quadratic programming (QP) form optimization objective function J is constructed. t The aim is to minimize temperature tracking error and the rate of change of control variables. (20) Among them, y T (k+i|k) represents the predicted future exit temperature of the tobacco leaf at the i-th step based on the current moment, y set For the temperature setpoint, Δu T (k+j|k) represents the set of future control increment vectors for the j-th step based on the current moment, and Q T To output the weight matrix, R T This is the input weight matrix.

[0076] (2) Cascade control mechanism based on steam flow parameter transmission In the physical process of a leaf-humidifying machine, steam injection is the key link connecting the temperature and humidity fields. To address the mutual influence among multiple variables, this invention designs the following cascaded computation path: a. In each control cycle, the temperature LMPC controller is operated first; b. Parameter Locking and Transfer: After solving the QP problem, the temperature LMPC controller obtains the optimal control sequence for future time steps. The calculated optimal steam flow sequence u2 is then extracted from this sequence. * .

[0077] c. Feedforward compensation effect: This optimizes the steam flow sequence u2 * It is sent as an instruction to the actuator and simultaneously transmitted in real time to the humidity NMPC controller as a known look-ahead quantity. This allows the humidity controller to predict the increase in moisture brought by steam when calculating the amount of water to be added, thereby achieving precise coordinated control.

[0078] (3) Design a humidity NMPC controller based on minimizing the deviation between humidity and water addition. Using the nonlinear mass conservation model constructed in step S4, a humidity predictive controller is designed. Since the model includes time-varying characteristics of tobacco leaf flow and nonlinear terms defined on a wet basis, a nonlinear optimization algorithm is employed for this controller.

[0079] The nonlinear difference equation obtained in step S4 is used as the prediction kernel: (twenty one) The objective function is defined as: (twenty two) Among them, M out (k+i|k) represents the predicted future outlet moisture content, Δu3(k+j|k) represents the predicted future water addition flow increment for the j-th step based on the current time, and Q M To output the weight matrix, R M This is the input weight matrix.

[0080] Using Taylor series expansion, the nonlinear prediction model in step S4 is approximated with a first-order linearization near the operating point. The Jacobian matrix is ​​calculated, transforming the nonlinear model into a time-varying linear approximation model, and the original nonlinear objective function J is... h Approximately a standard quadratic form, construct a quadratic programming problem for solution.

[0081] Step S6: Online Scrolling Optimization and Feedback Execution This step involves deploying the cascaded MPC control strategy designed in step S5 into the execution phase of the leaf-conditioning machine's field control system. The system operates periodically online based on a sampling cycle, achieving precise temperature and humidity control through a closed-loop design involving real-time sampling, state observation, rolling optimization, and control deployment.

[0082] (1) Real-time sampling and data preprocessing At the beginning of each control cycle, real-time operating data of the leaf-humidifying machine is collected through field sensors. The collected raw signals are then processed by moving average filtering to remove high-frequency noise, yielding the observed values ​​used for calculation.

[0083] (2) State observation and feedback correction This invention utilizes a feedback correction mechanism to modify the internal state of the model, thereby ensuring the accuracy of the prediction.

[0084] For the temperature linear model, a discrete Kalman filter is designed, using the measured outlet temperature T. out_measure The deviation of (k) from the model prediction, combined with the Kalman gain matrix K kalman For the two-state vector of the system Perform the optimal estimate.

[0085] (twenty three) Among them, C T This is the output matrix in step S3.

[0086] For the nonlinear humidity model, calculate the measured moisture content M. out_measure The deviation e between (k) and the model prediction M (k). This deviation is applied as an additive perturbation to the subsequent prediction trajectory to shift and correct future moisture predictions, thereby eliminating steady-state errors.

[0087] Cascade Scrolling Optimization Solution Based on the corrected state and current operating parameters, the optimization problem is solved in cascade order to calculate the optimal control quantity at the current moment.

[0088] 1) Solving for temperature LMPC Corrected temperature status Substitute the values ​​into the temperature LMPC controller. Call the online quadratic programming solver to solve for the control increment sequence in the future control time domain, while satisfying the valve opening and increment constraints. Extract the first element of the optimal sequence to obtain the optimal hot air temperature u1 required at the current moment. * (k), Steam flow rate u2 * (k), discharge chamber wall temperature u3 * (k).

[0089] 2) Parameter transfer and model update The calculated steam flow rate u2 * (k) Lock, and match with the real-time collected tobacco leaf flow rate F tob (k) is input into the humidity model. At this point, using Taylor expansion, based on the current u2... * (k) and F tob (k) updates the Jacobian matrix of the humidity model, completing the local linearization of the nonlinear model.

[0090] 3) Solve for humidity NMPC Based on the updated linearized humidity model, a quadratic programming subproblem is constructed. The QP solver is then used to calculate the optimal mechanical water supply flow rate u4 required to achieve the target humidity after deducting the contribution of steam condensate. * (k).

[0091] 4) Control command output and execution According to the rolling time-domain control principle, only the calculated control quantity (u1) at the current moment is used. * (k), u2 * (k), u3 * (k), u4 * (k) is the final command output. The calculated physical quantity is converted into a standard signal that can be recognized by the field actuator. Before the signal is issued, an engineering limit check is performed to prevent actuator overtravel or malfunction due to calculation errors. 5) Control command output and execution After completing this control action, the system enters the next sampling time k+1 and repeats steps 1) to 4). Through online rolling optimization, the control system can sense fluctuations in tobacco leaf flow and changes in steam pressure in real time, and adjust the control strategy accordingly, thereby maintaining a high degree of stability in the temperature and humidity at the humidifier outlet throughout the entire production process.

[0092] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A cascade control method for tobacco leaf re-drying and moistening equipment, characterized in that, Includes the following steps: Step S1: Leaf wetting process management and physical mechanism analysis to determine the modeling object of the leaf wetting machine; Step S2: System boundary determination and variable definition; Step S3: Temperature state-space modeling based on energy conservation; Step S4: Nonlinear modeling of humidity based on mass conservation; Step S5: Design of cascade MPC controller. Considering the physical relationship between the thermodynamic response and humidity response of the leaf humidifier, a temperature-priority and humidity-coordinated control strategy is adopted. Design a temperature controller based on linear model predictive control, a cascade information transmission control, and a humidity controller based on nonlinear model predictive control. Step S6: Online rolling optimization and feedback execution. The cascade MPC control strategy is deployed in the execution phase of the leaf humidifier's field control system. The system runs online periodically based on the sampling period. Through a closed-loop design of real-time sampling, status observation, rolling optimization, and control issuance, precise control of the temperature and humidity of the leaf humidifier is achieved.

2. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 1, characterized in that, Step S1 includes: clarifying the working principle, physical structure boundary, and energy and mass transmission path of the leaf humidifier; the leaf humidifier includes a feeding chamber, a rotating drum, a discharging chamber, a frame, a hot air system, a pipeline system, and a dehumidification system, and the modeling object is the internal space of the drum of the leaf humidifier and its associated actuators.

3. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 2, characterized in that, Step S2 includes: analyzing the energy flow and mass flow transmission mechanism inside the leaf humidifier system, dividing the thermodynamic boundary and control volume of the system accordingly, and clearly defining the state variables, control input variables and disturbance variables.

4. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 3, characterized in that, Step S3 includes: analyzing the thermodynamic mechanism inside the leaf-humidifying machine, constructing a set of differential equations describing the dynamic characteristics of the system, establishing energy conservation equations for the internal air and tobacco material respectively, and converting them into a standard state-space form.

5. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 4, characterized in that, Step S4 includes: establishing a dynamic model for the moisture content of tobacco leaves at the outlet of the leaf conditioning machine; the humidity change process is affected by the definition of wet basis moisture content and material flow rate, exhibiting significant nonlinear characteristics, and thus, based on the law of conservation of mass, the dynamic process of moisture addition and mixing is analyzed.

6. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 5, characterized in that, In step S5, a temperature LMPC controller based on minimizing temperature deviation and energy consumption is designed, including: Using the discrete state-space equation obtained in step S3 as the prediction model, and setting the prediction time domain as P and the control time domain as N, an optimization objective function of the form of quadratic programming QP is constructed to minimize the temperature tracking error and the rate of change of the control quantity. Under the premise of satisfying the equipment constraints, the optimal hot air temperature and steam flow rate are calculated to maintain the stability of the outlet tobacco temperature.

7. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 6, characterized in that, In step S5, designing the cascade calculation path based on steam flow parameter transmission includes: a) Within each control cycle, the temperature LMPC controller will be operated first; b) Parameter locking and transmission: After solving the above QP problem, the temperature LMPC controller obtains the optimal control sequence for future time moments, from which the calculated optimal steam flow sequence is extracted; c) Feedforward compensation: The optimal steam flow sequence is sent as an instruction to the actuator and simultaneously transmitted in real time to the humidity NMPC controller as a known look-ahead quantity. This allows the humidity controller to predict the amount of moisture increase brought by the steam when calculating the water addition, thereby achieving precise coordinated control.

8. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 7, characterized in that, In step S5, a humidity NMPC controller based on minimizing the deviation between humidity and water addition is designed, including: Using the nonlinear mass conservation model constructed in step S4, a humidity prediction controller is designed. The nonlinear difference equation obtained in step S4 is used as the prediction kernel. The objective function is defined, and the nonlinear prediction model in step S4 is expanded using Taylor series. The nonlinear prediction model near the operating point is approximated with a first-order linearization. The Jacobian matrix is ​​calculated, and the nonlinear model is transformed into a time-varying linear approximation model. The original nonlinear objective function is approximated as a standard quadratic form, and a quadratic programming problem is constructed for solution.

9. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 8, characterized in that, Step S6 includes: (1) Real-time sampling and data preprocessing At the beginning of each control cycle, real-time operating data of the leaf-wetting machine is collected through field sensors. The collected raw signals are then processed by moving average filtering to remove high-frequency noise and obtain the observation values ​​used for calculation. (2) State observation and feedback correction The internal state of the model is corrected by a feedback correction loop to ensure the accuracy of the prediction. For the temperature linear model, a discrete Kalman filter is designed. The deviation between the measured outlet temperature and the model prediction is used, combined with the Kalman gain matrix, to make the optimal estimate of the dual state vector of the system. For the nonlinear humidity model, the deviation between the measured moisture content and the model prediction is calculated. This deviation is then applied as an additive perturbation to the subsequent prediction trajectory to shift and correct the future moisture prediction values, thereby eliminating steady-state errors. (3) Cascade rolling optimization solution Based on the corrected state and current operating parameters, the optimization problem is solved in cascade order to calculate the optimal control quantity at the current moment.

10. The cascade control method for tobacco leaf re-drying and moistening equipment as described in claim 9, characterized in that, The cascade rolling optimization solution process includes: 1) Solving for temperature using LMPC Substitute the corrected temperature state into the temperature LMPC controller, call the online quadratic programming solver, and under the premise of satisfying the valve opening and incremental constraints, solve the control increment sequence in the future control time domain, extract the first element of the optimal sequence, and obtain the optimal hot air temperature, steam flow rate, and discharge chamber wall temperature required at the current moment. 2) Parameter transfer and model update The calculated steam flow rate is locked and input into the humidity model along with the real-time collected tobacco flow rate. Using the Taylor expansion method, the Jacobian matrix of the humidity model is updated according to the current steam flow rate and tobacco flow rate to complete the local linearization of the nonlinear model. 3) Solve for humidity NMPC Based on the updated linearized humidity model, a quadratic programming subproblem is constructed, and the QP solver is called to calculate the optimal mechanical water supply flow rate required to achieve the target humidity after deducting the contribution of steam condensate. 4) Control command output and execution Based on the rolling time domain control principle, only the calculated control quantity at the current moment is used as the final command output. The calculated physical quantity is converted into a standard signal that can be recognized by the field actuator. Before the signal is issued, an engineering limit check is performed to prevent the actuator from overtravel or malfunction due to calculation abnormalities. 5) After the control command is output and executed, the system enters the next sampling time k+1 and repeats the steps 1) to 4) above. Through online rolling optimization, the control system can sense the fluctuation of tobacco leaf flow and the change of steam pressure in real time, and adjust the control strategy in time, thereby maintaining a high degree of stability of the temperature and humidity at the outlet of the leaf humidifier throughout the entire production process.

Images