Cyber Granular-Enabled Intelligent Sensing for High Performance Flocking Control in Continuous Flat Pressing System

The deviation type cyber of granular construction satisfies the multi-objective optimisation requirements of flexible process and agile production of MDF in intelligent manufacturing, and the ultimate goal is to realise the intelligent perception of cluster control of MDF continuous flat pressing system. At present, the research on the formation mechanism of plate thickness deviation and quality control of deviation used to ensure product quality and improve production efficiency in MDF continuous flat pressing is an urgent problem to be solved in the forestry intelligent equipment industry, therefore, innovatively carried out research on the distributed control mode and deviation control collaborative method of MDF continuous hot pressing, and for the “constant thickness stage” of MDF continuous flat pressing process, the “viscoelastic multi-field control granularity” of MDF continuous flat pressing process is the best way to achieve intelligent sensing. “viscoelastic multi-field coupling distributed agile regulation and control problems, put forward a control grain structure based on the array of pressure cylinder, collaborative cluster control and intelligent perception technology of the fault type decision-making controller design scheme, designed three decision-making intelligent perception controller. By controlling the grain-driven intelligent perception technology, the data acquisition system of the position and pressure sensors of the MDF continuous flat pressing system carries out real-time monitoring and analysis of the various indicators of the sheet; the actual production test shows that the scheme has high feasibility and practicability, and the accurate quality control can improve the quality and efficiency of MDF sheet production.


I. INTRODUCTION
Manufacturing systems are becoming more sophisticated and expensive, particularly with the development of the intelligent industry [1].The new generation of intelligent manufacturing systems requires a deep integration of humancyber-physical spaces.Visual analytics plays a critical role in effectively navigating humans through twin data, enabling The associate editor coordinating the review of this manuscript and approving it for publication was Min Wang .them to make better decisions and discover new knowledge, contributing to the evolution of digital twins [2].Intelligent manufacturing has become the main trend of transformation and upgrading of the wood-based panel industry, and intelligent manufacturing technology and intelligent factories have gradually become the focus of research by relevant scholars.
With the gradual upgrading of the wood-based panel industry in China, more and more MDF production lines adopt continuous flat-pressing production, while the preheating of slabs is conducive to the perfection of the hot pressing process and the improvement of production efficiency, and is widely used in the production [3].How to improve the uniformity of heat transfer during hot pressing and reduce the heating time through a reasonable preheating process, so as to enhance product quality, improve production efficiency, save energy, reduce production costs, and improve market competitiveness has become a highly important issue for all manufacturers [4].The practical application of microwave preheating of slabs and the combination of microwave preheating of slabs and steam preheating of slabs in MDF production lines is introduced [5].
How to improve the uniformity of heat transfer during hot pressing and reduce the heating time through a reasonable preheating process, so as to enhance product quality, improve production efficiency, save energy, reduce production costs, and improve market competitiveness has become a highly important issue for all manufacturers [6].The practical application of microwave preheating of slabs and the combination of microwave preheating of slabs and steam preheating of slabs in MDF production lines is introduced [7].
To date, Dieffenbacher has sold more than 70 sets of continuous press lines in China, including 14 sets of the new generation of CPS+ continuous presses, with seven production lines already in smooth operation [8].In addition to inheriting Dieffenbacher's safety, reliability and efficiency, this equipment also combines the needs of Industry 4.0, big data, artificial intelligence and integrated automation control in the new era, and has become a model for efficient continuous press systems.
The continuous flat press is the most critical process link of the man-made board production line, which is the technical bottleneck and the containment project for the formation of the domestic supporting capability of hot-pressing equipment [9].Zhu et al. [10] combined the structure of the inlet section of the continuous press and its deformation requirements of the hot pressed plate when pressing the slab, the exhaust requirements of the slab, and the various control requirements of the pressurized cylinder in the inlet section, applied the design method of the hydraulic proportional control system.
In order to investigate the key factors affecting the stability of multi-terminal DC systems, Zhu et al. [11] established the corresponding equivalent circuit for the low-voltage three-terminal DC system composed of the voltage-source converter, DC line, renewable energy and DC load, and combined the control strategies of each converter in the master-slave control mode to form a state-space model of the system as a whole in order to investigate the key factors affecting the stability of the multi-terminated DC system.
Guangwei et al. [12] have conducted research on the position and pressure master-slave control of hydraulic servo systems.The focus was primarily on position control with supplementary emphasis on pressure closed-loop control.The system is designed to convert the pressure signal, in real-time, when hydraulic cylinders interact with external loads, and then simultaneously apply feedback signals of the position closed-loop.
Combining fuzzy failure mode and effects analysis (FMEA) method, Liu et al. [13] proposed an FMEA deviation type classification method [14] that can classify deviation types into classes as a basis for deviation class classification [15].The method addresses the potential for plate thickness failure and thickness deviation defect adjustment in the continuous hot pressing (CHP) process.
Lv investigated the possibility of automatic quality control of the continuous flat pressing process on a continuous hot pressing process of MDF.A distributed coordination mode framework was constructed to propose a process control programming method for MDF continuous flat press quality control [16]; based on the MDF continuous flat press process mechanism and control criteria, a framework of five distributed flat press collaborative control mode elements for continuous panel system (CPS) press cylinder arrays was defined.[17].
He et al. [18] is concerned with sampled-data leaderfollowing consensus of a group of agents with nonlinear characteristics.A distributed consensus protocol with probabilistic sampling in two sampling periods is proposed [19] develops a novel synthesis approach for the synchronization of a network of singularly perturbed systems (SPSs) with a small singular perturbation parameter e via distributed impulsive control [20].
Gao put forward a cement decomposition furnace outlet temperature prediction control algorithm based on the optimisation of three-branch decision-making, adopts the theory of three-way decision-making to dynamically adjust the decomposition temperature operation indexes according to the change of working conditions, and realises the real-time dynamic optimisation of process indexes, at the same time, using the model prediction control method, builds the At the same time, using the model predictive control method, the decomposition furnace temperature control model is constructed, the objective function combining set value control and interval control is constructed, and the optimal solution is obtained by optimally solving the objective function to achieve the precise control of the decomposition furnace temperature.
In summary, although there is a large amount of research base on MDF continuous hot pressing and control direction, there is a lack of systematic analysis in the continuous hot pressing process, and there is a lack of research techniques on the deviation situation of the sheets in the process quality control.Therefore, this paper proposes a master-slave cooperative control method based on the control grain calculation for the process control method related to the deviation problem of MDF continuous hot pressing sheets with high practical significance, the proposed control decision method in the actual production of MDF continuous hot pressing process in the man-made board industry The new method has some feasible significance in the practical production of MDF continuous hot pressing process in the wood-based panel industry, The aim is to study the single-cylinder power execution control grain structure of MDF continuous hot press system and to design the controller based on the synergistic method among the press cylinder groups.The proposed method of master-slave synergistic control grain calculation based on three-way decisions will optimize and improve the design of the process decision rule-base and finally form the core controller component module of high performance continuous hot press machine, and it is intended to pass the application feasibility verification in the production line with an annual capacity of 300,000 m 3 .
In the MDF continuous hot pressing cooperative control provides a new method, which contributes to the development of the wood-based panel industry, which is reflected in: (1) it provides a new synergistic method, which can make the MDF continuous hot pressing process to achieve precise control.
(2) increasing the output of MDF and improving a certain degree of superiority in the production of MDF.

A. CONSTRUCTION OF PRESSURE HYDRAULIC CYLINDER CYBER GRANULAR STRUCTURE UNIT AND FRAMEWORK
By studying the intrinsic constitutive equations and creep equations of the Burgers model for viscoelastic bodies, the expression for creep flexibility is obtained [21].The viscoelastic model is established based on the viscoelastic mechanism of the slab drawing stiffness-compression cylinder structure [22].The distributed synergistic model of cyber granular computing (several distributed synergistic approaches)consists of a single terminal power actuator unit and viscoelastic model as a single grain structure unit, the hydraulic power system is considered as a power mechanism with elastic load, and the single cylinder power actuator unit grain structure can be represented by the structure shown in Figure 1.
Figure 1 mainly consists of a current servo valve, a fourway valve-controlled hydraulic cylinder [23].A detection mechanism, a controller, and a slab viscoelastic model [24].In the slab viscoelastic structure, η 1 and η 2 denote the elasticity coefficient, and these two parameters can reflect the influence of viscoelasticity on the cyber granular in the single cylinder power actuation unit.According to the command (i.e.reference input), the load generates displacement or pressure along the direction of action.After detecting the real-time output data of the load, the controller calculates the control quantity according to the deviation between the real-time output and the reference input and controls the hydraulic cylinder to drive the load to generate the corresponding motion and eliminate the deviation.Where P L is the hydraulic propulsion force per unit area, the hydraulic cylinder as shown in Figure 1, the continuity equation of the hydraulic cylinder flow Q L (displacement output) is where F g is the hydraulic push force, B c is the viscous damping factor, and K is the load spring stiffness.Due to the viscoelastic properties [25] of the slab during hot pressing and the curing effect of the adhesive, the slab will have a rebound effect on the output of the hot-pressed plate.the MDF slab viscoelastic model principal equation.
Combining behavioral decision theory [26], three-way decision theory [27], and granular Computing Theory (GCT) [28] is a new theory of artificial intelligence and data mining that aims to introduce the uncertainty and ambiguity of human knowledge into computing.Granular Computing Theory opens up a new way to process complex information and to communicate between humans and computers, with the important feature that information, signals, and data of different granularity can be effectively fused [29].In the field of decision-making, granular computing theory provides a new approach to information processing and decision analysis, allowing for more accurate and effective classification of data in data analysis to help people make better decisions and slide mode.[22], a three-way decision model based on grain calculation for MDF continuous flat pressure is proposed through that three methods.
Equation (2) (3) ε denotes the elastic deformation of the slab, σ denotes the elastic stress, and E is the spring stiffness which η 1 , η 2 respectively, denotes the amount of adhesive curing and thermal mass transfer, by equation( 1) the principal equation can be derived as follows Figure 2 viscoelastic deformation relationship under different stresses.
It can be seen that, when applying the co-efficient principle to solve the single terminal control problem of plate thickness correction, the intervention quantity of the plate quality control process, which should start from 0 and grow to a positive value, behaves as a growth from a negative value to 0. The difference in the growth interval is caused by the special physical meaning expressed by the intervention quantity of the plate quality control process, while its change trend is consistent with that expressed by the co-efficient principle, The trend is consistent with the principle of the covariance, which is not contradictory to the principle of 'the system reaches order when the amount of intervention in the quality control process of the plate is maximum'.
f u1 (x) represents the initial state function when the viscoelastic slab without elastic deformation, and f u2 (x) represents the state function when the viscoelastic elastic deformation reaches a steady state.Equation ( 6) represents the output between the position and pressure and the relationship with the viscoelastic model under the action of viscoelastic gluing and coupling of multiple fields such as thermal mass transfer.

B. TO CONSTRUCT A THREE-WAY DECISION MODEL FOR DETERMINING THE TYPE OF CONTINUOUS FLAT PRESSURE DEVIATION OF MDF
For the determination of the deviation type of the MDF continuous flat pressing process, it uses the 5 * 5 press cylinder array as the target for testing, and the decision analysis of the deviation type is carried out by the grain calculation method to construct the three-way decision model, and the decision diagrams are shown in Figure 3, which indicate that the second determination and decision are generated only in the case that the intervention is applied; the target after the decision is finally presented with the form of the decision matrix.
Model output of the decision takes: (2) Pressure-pressure output: active terminal Master and slave terminal Slave output different physical quantities at the same time under the same input, Master terminal output pressure, Slave terminal according to Master terminal pressure sensor for pressure follow the output.
(3) Position-pressure output mode: active terminal Master and slave terminal Slave output different physical quantities at the same time under the same input, Master terminal output displacement, and Slave terminal output pressure according to the part of the pressure sensor accompanying the Master terminal position sensor.The controller needs to correct the position-pressure output of the actuator unit according to the feedback of the pressure value between the decision maker (Master) and the decision maker (Slave) to make the output meet the thickness requirement.

C. DESIGN OF MDF CONTINUOUS FLAT-PRESS DEVIATION TYPE DECISION CONTROLLER
The controller is designed by the above-constructed three-way decision model of MDF continuous flat-press deviation type and level determination, and there exist position-position, pressure-pressure, and inter-positionpressure synergistic ways among the press cylinders, whose corresponding controller designs are shown in Figure 3-8:

1) THE CONTROLLER DESIGN FOR THE HYDRAULIC CYBER GRAIN PRESS CYLINDER OUTPUT MODE
The controller in MDF continuous flat-press distributed synergistic control is designed to design the grain structure actuator with intervention amount,which includes two kinds of actuators, including single group grain and group frame grain structure whose output forms of position-position, pressure-pressure, and inter-position-pressure.
Figure 3 shows the design of the controller for the position-position linkage of the single grain press cylinder,  D d is the distance information, and D 1 is the distance output formed by the single grain structure decision maker Master and controller 1 together under the action of the intervention quantity g (the value range is between plus and minus 0.05).The output D 1 is feedback to the comparator which D d is located through position sensor 1, and D is the position output formed by the single-grain structure decision maker Slave acting together with controller 2 to realize the position-position linkage control of D 1 and D 2 .
Figure 4 shows the controller design for the pressurepressure interval of the single-grain press cylinder.P d is the input pressure, P 1 is the position output formed by the single grain structure decision maker Master acting together with controller 1, and P 2 is the position output formed by the single grain structure decision maker Slave acting together with controller 2 to achieve P 1 P 2 pressure-pressure follower control.
Figure 5 shows the controller design for the positionpressure interval of the single-grain cylinder.D d is the input position information, and in the case of intervention g, D 1 is the position output of the single-grain structured decision maker Master acting together with controller 1. D 1 Output is feedback to D d the comparator via position sensor 1 and, at the same time, an accompanying pressure is  generated on pressure sensor 1. P 1 is the position output of the single-grain structured decision maker Slave acting together with controller 2 to achieve master-slave control of D 1 and P 1 .
Figure 6 shows the controller design of the positionposition linkage between the group frames of the press cylinder.A group frame with five sensors is defined as a group frame, and the output of the group frame is the position group information formed by the action of the intervening quantity g (the value of g is in the range of plus or minus 0.05).The output of D 1 is fed back to the comparator where D d is located via position sensor 1, and D 2 is the position output of the group frame particle structure decision maker Slave acting together with controller 2 to achieve position-position linkage control between the groups of D 1 and D 2 .

2) DESIGN OF A MASTER-SLAVE STRUCTURE CONTROLLER FOR MULTIPLE GRAIN FRAMES BETWEEN PRESSURE CYLINDER GROUPS
Figure 7 shows the design of the controller for the pressure-pressure linkage between the frames of the press cylinders.Pd represents the position information of the input group of five sensors, and under the action of the intervention quantity g (which takes values between plus and minus 0.05), P 1 is the position group information output formed by the group frame grain structure decision maker Master and 41238 VOLUME 12, 2024 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.Figure 8 shows the controller design of the positionpressure inter-frame of the pressure cylinder group, D d is a set of distance input information of the group frame, under the action of the intervention amount g, D 1 is the distance output formed by the single grain structure decision maker Master and controller 1, the output of D 1 is feedback to the comparator where D d is located through position sensor 1, and at the same time an accompanying pressure will be generated on the pressure sensor 1, P 1 is the position output formed by the single grain structure decision maker Slave acting on the output of pressure sensor 2 and controller 2, to realize the master-slave cooperative control of D 1 P 1 .
According to the design scheme of the controller design diagram, the Master-Slave method is applied to simulate the design, and the resulting single group of pressure cylinder displacement and PID tracking effect is shown in Figure 10: Figure 9 shows the results of the single cylinder displacement simulation experiment, the figure can be seen in 3 seconds is a better tracking effect, the first two seconds before the slab viscoelastic conditions will have a certain   delay, but in the later after the viscoelastic treatment can maintain a steady state.
Figure 10 shows the effect of the output of the single cylinder pressure, set the step time to 5, and the effect of tracking the reference output at the completion of the step response to observe the difference, it can be seen that the error is small, the method is feasible.
Figure 11 shows that the M-S method in collaboration with the traditional PID tracking control method so that the reference input is a step signal, the generation step time is 1 second, the intervention amount of the initial value is set to 0.06mm, and the final value is 0.3mm.It means that the output thickness of the plate thickness needs to be adjusted to 30mm for plate deflection, and it can be seen that there is a certain delay in tracking, but it can be kept consistent after 5 seconds.
Figure 12 shows the output curve from the main pressure controller.The control quantity of the PID controller oscillates and has a large amplitude, and reaches a stable   output after 4s.The calculation volume is significantly larger than that of the synergistic controller, and its ability to guide the system to a steady state will be degraded if the plate thickness correction command changes frequently; the maximum value of the control quantity output of the two synergistic controllers is about 1/2 of the maximum value of the PI controller output, and there is no oscillation.
Figure 13 shows that, set the MS method compared with the traditional PID tracking control method, so that the reference input is a step signal, the generation step time is 1 second, the initial value is 100MPa, the final value is 500Mpa.indicates that the plate correction needs to adjust the displacement output of the hydraulic cylinder piston to 400-500Mpa.
Figure 14 shows the synergistic relationship between the dual outputs P1 and P2 for multiple pressure controllers.The  trend in the figure shows that the MS synergistic approach is superior to PID tracking and allows a faster approach to a steady state.
Figure 15 shows the pressure output in a synergistic way, the pressure in the figure indicates the input increased pressure intervention amount, the initial intervention amount is 0.05, the negative value represents the intervention amount is adjusted downward on the basis of the set value, the tracking effect is better after 3 seconds, and the synergistic is more consistent.
Figure 16 shows the position output in the cooperative mode.The vertical coordinate in the figure indicates the input increased position intervention amount, the initial intervention amount is 0.05, and the negative value means that the intervention amount is adjusted downward from the set value, and the tracking reaches synchronization after 5 seconds.
Figure 17 shows the output of the intervention volume for the synergistic pressure between multiple pressure cylinder groups.It can be seen that the tracking of multiple pressure cylinder groups can be consistent after the step response with the Master-Slave method of synergistic correction, and the tracking effect is satisfactory.

D. COOPERATIVE CONTROL METHOD FOR CONTINUOUS FLAT-PRESSING THREE-WAY DECISION MAKING IN MDF WITH INTERVENTION
The hydraulic equation of the state of the system between the MDF continuous flat-pressure distributed synergy is 41240 VOLUME 12, 2024 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
represented by the particle information.
where a 2 = − ), the hydraulic distance servo system is a third-order system, which F L mainly comes from the changing steam pressure inside the slab, and the external disturbance is uncertain.Since the parameters such as the modulus of elasticity of the fluid β e and the total leakage coefficient of the hydraulic cylinder are uncertain a 1 a 2 , there are parameters set to move, but in the actual system, the disturbance is considered to be bounded.
The distributed cooperative control in the MDF continuous hot pressing process is analyzed, and the grain structure state model between two press cylinders in a single press cylinder is shown as follows: S : x 1 x 2 x 3 represents the state of the press cylinder movement in the distributed control, P L represents the pressure, the displacement of the press cylinder piston,in equation ( 7)- (8) x M z M represents the position state output of the single grain structure using Master mode, x S z S represents the position state output of the single grain structure using Slave mode, Master and Slave work together to achieve the synergistic control between master and slave, so that the auxiliary cylinder can be better with the positioning linkage of the master cylinder.
Equation ( 9)- (10) represents the synergy between the two single-cylinder grain structures and the position pressure output, two or more groups of pressure cylinders between the group frame between the pressure cylinder grain structure synergy between the master and slave structure analysis.L represents the active control of the part -Leader is the leader, F represents the part of the slave control -Follower is the follower.

L :
x x L z L represents the distance state output of the in Leader mode adopting cylinder group frame, x F z F represents the   position state output of the adopted cylinder group frame in Follower mode, Leader and Follower work together to achieve the synergistic control between master and slave cylinder group frame so that the slave cylinder group frame grain structure can better match the distance linkage of the master cylinder and cylinder groups frame grain structure.
From the analysis of the above equations ( 9)-( 10), it can be obtained that among the four deviation types, different deviation types correspond to different decision methods.With the known deviation types, the information on their deviation levels is used for decision-making, and the mapping relationship and adjustment scheme is derived.The state analysis of the four deviation levels is shown in Figure 18 For the four universal deviation types, it is possible to determine the direct implementation of the decision scheme for the deviation types.In the case of types that do not belong to a certain deviation type but fall within the boundary domain granularity space, further calculations are made to determine the deviation types and levels and give the decision scheme through the three-branch decision analysis of the four deviation types.
According to the waveform analysis of the simulation analysis in Figure 19-20, the specific deviation level is determined by the three-way decision method, and the combination of the decision scheme and the control mode is given.Under the square wave signal with different wave values, different deviation types will be generated (the determination of deviation type is determined by expert rules and field experience for plate determination), and different displacement amounts are input for correction at 8, 15, and 32 seconds, respectively.By adjusting the position value of the MS synergy method and inputting the given intervention amount for decision-making, it can be seen that the correction effect of the three-branch decision synergy method is obvious, and it can be a decision and intervention to solve the plate thickness problem caused by deviation.
Figure 21 shows that the pressure value in the vertical coordinate is not the actual pressure value, but the difference required in the adjustment process.It can be seen in the variable square wave signal that the pressure-pressure interval adjustment follows better and can meet the online adjustment mechanism in continuous production.

A. DETERMINATION OF THE TYPE OF DEVIATION IN THE CONTINUOUS HOT PRESSING PROCESS OF MDF AND CONTROL DECISIONS
The fault information in the FMEA table is completely expressed and the decision analysis of failure information is conducted.The fault information of MDF continuous hot pressing is listed as shown in Table 1 and Table 2, and the matrix data in the table are expressed as decision matrices of different deviation levels, and control mode mapping is made for each decision matrix.The forms of deviation types from left to right are expressed by the matrix information in turn, and the corresponding control mode decision mapping is made for each form to realize the distributed cooperative control among MDF continuous flat press cylinder groups.Among them, 3 + and 3 ++ represent different degrees in control mode 3. Since the deviation of MDF is relatively large and cannot be adjusted in an over-intensive way, there will be different degree control methods made for each control mode.To ensure the accuracy of correcting deviation, fine tuning, S indicates the sensor, the front bit of the number represents the position of the sensor, the back bit represents the information of the number of levels built by the sensor, through the different values of the five sensors, seen as five points, the points will be connected to determine the type of deviation initially, D + indicates the amount of thickness intervention, the intervention value has positive and negative, in different cases of thickness demand in accordance with the positive and negative intervention, can be added or subtracted P + P − indicate the pressure intensity value (increase or decrease) of the adjustment, and the more the plus or minus sign, the greater the intensity.
In the actual production process, 17-21 sets of pressure and displacement data were intercepted and sampled every five minutes, due to the large size of the sampling data, the table only shows the sampling data for every 5 minutes from 0-85 minutes as shown in Table 1 and Table 2, to observe the situation of dynamic data in different time periods, so as to facilitate the adjustment at any time.
According to the simulation analysis of Table 2 data, it can be seen that the middle part of the pressure value is small, the simulation results are shown in Figure 25, the red line represents the pressure value of the middle part of each system group, it can be seen that the pressure value is small, in 17-20 group the pressure is almost close to 0, in 21th group through the adjustment can basically be consistent with the  left and right side.This is the result of the synergistic way of cooperation between the pressure cylinder groups.
Simulation results as shown in Figure 23, the red line represents the middle part of the displacement, in the 17-20 group of the middle displacement has obvious fluctuations, can be determined at this time for the depression, therefore, need to find the deviation type and level of control decision program against for 4-3 ++ -2-2-4 mode.After the control mode and thickness intervention amount adjustment in group 21, there is a significant improvement to meet the requirement of producing a plate thickness of 5.5mm, which proves the effectiveness of the proposed synergistic method and control mode scheme.

B. TESTING OF AN MDF CONTINUOUS FLAT PRESS LINE
The method was tested on a 37-group press cylinder array press in the actual production site, and the MDF with a production thickness of 5.5mm was analyzed as an example.
Figure 24 takes the left slope deviation level 3 as an example for decision making, using 4-3-2-2-4 control mode decision, it can be seen that in groups 2-3 for thickness intervention, increase the amount of intervention, 11-14 groups for pressurization, 17-21, pressurization intensity for P + .
Figure 25 shows that the right slope deviation grade 3 is used as an example for decision making, and the 0-2-3-1-4 control mode decision can be seen in groups 18-19 for 41244 VOLUME 12, 2024 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
thickness intervention, increasing the amount of intervention, and groups 6-9 for pressurization, with pressurization intensity P + .
Figure 26 shows the decision making with bump deviation level 5 as an example, using 1-3-2-2-4 control mode decision making it can be seen that decompression is used to adjust the thickness of the current position in groups 2-3, thickness intervention is performed in groups 9-10 pressure cylinders with increased intervention amount D + , 11-13 are pressurized with pressurization intensity P + and P ++ , and groups 17-21 are decompressed with decompression intensity P − .

IV. CONCLUSION
Tested on a press with 37 sets of press cylinders on a production line with an annual production capacity of 300,000 m 3 , an intervenable MDF continuous flat press three-way decision cooperative control approach is proposed, which has a better improvement on the type of plate deviation problem, which is practical in the process of continuous production, increases the product quality of the production, improves the rate of superior products, and solves the difficulties in the direction on continuous production in agile production.
An intervenable three-way cooperative control method is proposed to solve the problem of distributed agile control of viscoelastic multi-field coupling in the ''constant thickness section'' of the MDF continuous flat-pressing process.The method constructs a process control sequence planning model for single and multi-group press cylinder arrays by defining a digital twin distributed grain structure model for MDF continuous press cylinder arrays, and designs an intervenable MDF continuous flat-press three-branch decision collaborative controller to cope with four types of typical deviations in sheet production: thickness, slope, depression, and bump.This is reflected in the following aspects: (1) it is innovative to redefine the single terminal power unit of MDF hydraulic cylinder in the way of grain structure; (2) it adopts the idea of three-way decision to construct the master-slave cooperative distributed relationship of the press cylinder, which makes such a situation with uncertain deviation types and levels in the boundary domain controllable; (3) it designs three controllers with different outputs, which is more intuitive to find the deviation problem (4) The distributed control of MDF continuous flattening process is analyzed by a three-branch decision method, which takes into account the slab viscoelasticity and manual intervention to decide the type and level of deviation of MDF continuous flattening process, and the amount of intervention plays a decisive role in the Master-Slave synergy and Leader-Following control modes.and based on expert rules makes the production of sheets more accurate in thickness control and meets the requirements of agile MDF production.The method is tested in a press with 37 groups of press cylinders on a production line with an annual capacity of 300,000 m 3 , and it has a better improvement on the problem of the type of deviation of the plates, increases the quality of the produced products, improves the rate of superior products, solves the problem of continuous production direction in agile production, provides a certain realistic value in the forestry intelligent equipment industry of forest resources, and makes a better diary for the research of forest direction for Agile control and intelligent production of wood-based panel production has an important reference value.
YAQIU LIU (Student Member, IEEE) received the Ph.D. degree from Harbin Institute of Technology.He is currently a Professor.His research interests include intelligent and optimal control, intelligent soft measurement, intelligent information processing, system modeling and simulation, embedded computing, intelligent equipment and CPS technology, and robotics and intelligent systems.

FIGURE 1 .
FIGURE 1. Cyber granular structure of the MDF press single hydraulic cylinder power actuation unit.

FIGURE 2 .
FIGURE 2. Three-way decision model for the determination of MDF continuous flattening deviation grade grain structure.

FIGURE 3 .
FIGURE 3. Single grain press cylinder position-position interlocking controller design.

FIGURE 4 .
FIGURE 4. Pressure-pressure follower controller design for single grain hydraulic cylinder.

FIGURE 5 .
FIGURE 5. Position-pressure master-slave structure controller design for single grain hydraulic cylinder.

FIGURE 6 .
FIGURE 6. Position-pressure master-slave structure controller design for single grain hydraulic cylinder.

FIGURE 7 .
FIGURE 7. Pressure-pressure control follower controller design between the frames of the pressure cylinder group.

FIGURE 8 .
FIGURE 8. Positon-pressure control follower controller design between the frames of the pressure cylinder group.

FIGURE 9 .
FIGURE 9. Single cylinder displacement and reference output tracking.

FIGURE 10 .
FIGURE 10.Single cylinder pressure and reference output tracking.

FIGURE 13 .
FIGURE 13.Pressure-pressure leader between pressure cylinder groups following the output of the synergistic system.

FIGURE 14 .
FIGURE 14. Pressure output of master-slave synergistic controller with step volume input under pressure cylinder.

FIGURE 15 .
FIGURE 15.Synergistic pressure output between multiple cylinders at step volume.

FIGURE 16 .
FIGURE 16.Single-cylinder position adjustment for the type of deviation under the three-way decision.

FIGURE 17 .
FIGURE 17. Synergistic pressure output between multiple cylinders at step volume.

FIGURE 18 .
FIGURE 18. Single-cylinder position adjustment for the type of deviation under the three-way decision.

FIGURE 19 .
FIGURE 19.Single-cylinder pressure adjustment for the type of deviation.

FIGURE 20 .
FIGURE 20.Master-slave position cooperative correction for deviation type.

FIGURE 21 .
FIGURE 21.Master-slave position cooperative correction for deviation type.

FIGURE 22 .
FIGURE 22. Trend of actual pressure data for the three groups of left-center-right of system 17-21.

FIGURE 23 .
FIGURE 23.Trend of actual distance data for the three groups of left-center-right of system 17-21.

FIGURE 25 .
FIGURE 25.Right slope type deviation decision diagram.
YUNLEI LV (Member, IEEE) received the master's degree from Northeast Forestry University, where she is currently pursuing the Ph.D. degree with the College of Computer and Control.Her current research interests include forestry intelligent equipment and intelligent control.ADIL MALIK (Member, IEEE) received the master's and Ph.D. degrees from Harbin Engineering University, China.He has over 20 years' experience of teaching, operating and maintaining integrated platform management system (IPMS) that controls and monitors all aspects of the vessel's propulsion plant and systems, and interfacing seamlessly with the Combat Systems and Navigation Suit.His main research interests include intelligent control and automation of energy conversion systems.

TABLE 1 .
Pressure continuous sampling data during continuous hot pressing of MDF.

TABLE 2 .
Distance continuous sampling data during continuous hot pressing of MDF.
FIGURE 24.Left slope type deviation decision diagram.