CN103610227B - Cut tobacco dryer head and tail section process variable optimizing control method - Google Patents

Abstract

The present invention discloses a head end section Drying process variable optimization method of controlling the temperature history data Drying cylinder head and tail segment basis, air temperature, moisture exhaust damper of the process variables and the like used as a cubic function of the radial basis function Cubic-RBF-ARX model for modeling the dynamic characteristics drying; the model having a self-adjusting ability, can reflect the influence of the inlet flow in different modes and the change of the water outlet of the inlet water, the head and tail sections may be different according to the mode change the water inlet flow and an inlet to predict future changes in the water outlet; optimized for each set of process variables according to the model, can control the water outlet head and tail cut tobacco segment achieve better results. The method of the present invention to consider the quantity and the dynamic characteristics among the input variables, can be more effectively overcome the incoming water flow and change on Drying head and tail sections for the inlet and the inlet flow cut tobacco in different modes when moisture control head and tail sections.

Description

Drying the head end section one kind of optimal control process variables

FIELD

[0001] The present invention relates to a head end section Drying optimal control process variables.

Background technique

[0002] Drying process cigarette silk production is the most important one processing step, it is mainly carried out by heating the dried cut tobacco, cut tobacco to reduce the moisture content, a moisture content of cut tobacco after baking, uniform temperature and controlled within a certain range of values, in order to meet the production process. Drying is divided into the preheating process, dry head, middle and tail dry process four portions. In the dry first stage, cut tobacco inlet flow is increasing, but the non-detection value cut tobacco outlet of water, it is difficult to perform feedback control, is likely to cause dry head phase difference between the outlet water quality control, dry feed more; in the dry end of the stage, since the cut tobacco inlet flow suddenly decreased and drying cartridge having a large heat capacity, the internal temperature of the cylinder wall is difficult to fall at the rate specified problems are likely to cause low moisture management properties of the dry end of the outlet stage and multiple dry material. Thus, "dry dry head-tail" is cut tobacco drying process where the outlet moisture control difficulties.

[0003] Dry the head end of the conventional dry process control methods are:

[0004] (1) the use of materials and drying medium into and output Drying machine as heat and mass balance Object mathematical model combined feedforward PID control barrel temperature regulation. But the feed-forward mathematical model only takes into account the water flow rate and feed, and did not take into account other factors such as hot air temperature has an important influence on the export of water, can not fully reflect the true process, causing a large head and tail section Drying outlet moisture fluctuations, It requires operator manual intervention, the head and tail sections for the different modes having different inlet flow and the inlet of the incoming water is difficult to obtain a satisfactory control effect.

[0005] (2) Based on the above-described feedforward control, an increase in the head and tail section of the vapor discharge humidifying device craniocaudal applying steam feed water to raise the moisture content of the material end to end, in order to reduce the quantity dry. However, this method only cut tobacco humidify the surface, only to improve the silk leaf surface humidity, will still result in lower intrinsic quality of tobacco, and increasing the difficulty and stability of export control moisture.

[0006] (3) through several tests, to seek the best process parameters craniocaudal phase adjustment value and the temperature of hot air moisture exhaust valve opening degree is reduced quantity dry. When this method is the lack of self-regulation, is not guaranteed for the different modes having different incoming water inlet and an inlet flow rate, the optimal set of process parameters are set values;

[0007] (4) On the basis of PID control strategy based on fuzzy control theory applied to the tobacco drying machine moisture control.

[0008] in just a simple two-dimensional fuzzy controller to solve the problem of export control moisture Drying process still can not get the optimum process parameter settings, and change the inlet flow rate and inlet water for different modes, the need fuzzy control rule table adjustment, which brought inconvenience to industrial production.

SUMMARY

[0009] The present invention solves the technical problem, for the deficiencies of the prior art, there is provided a head end section Drying process control variable optimization method, and the dry stage cut tobacco outlet head increase of moisture as quickly as possible, and rapidly reach steady state, so that the tail dry stage cut tobacco decreases slowly as the water outlet, thus effectively reducing the quantity of dry head and tail section, improve the control performance drying process; more effectively overcome the incoming water flow and changes drying influence the process of head and tail section, to avoid inconvenience to enter manual tuning process variable parameters.

[0010] To solve the above technical problem, the technical solution employed in the present invention is: A head end section Drying process control variable optimization method is:

[0011] 1) The operational flow Drying machine, establish Drying process middle wire inlet flow rate, inlet water, barrel temperature, air temperature, moisture exhaust damper, the timing relationship between the outlet of water, while the dry head stage without accordance Drying Process leaf wire exit moisture measurement value, the dry end of the wire non-leaf stage with the inlet water flow characteristics of the inlet of the detection value, using the cubic function cubic-RBF-ARX model as radial basis functions are established drying process with the dry tail dry stage head Cubic-RBF-ARX model stage;

[0012] 2) The historical operating data Drying head tail section, structured nonlinear parameter optimization procedure were optimized Drying Dry head tail dry stage and stage Cubic-RBF-ARX model;

[0013] 3) Drying process according to the optimized phase and dry head Cubic-RBF-ARX model tail dry stage, moisture exhaust damper using double S phase function describes the dry head, optimal input air temperature curve, the temperature of the tube ; described step function using a dry phase inlet flow header optimum input curve; optimal input curve exponential function tail dry stage described moisture exhaust damper, air temperature, motor cylinder barrel moderate frequency;

[0014] 4) The method 列维布格奈奎 Stewart, minimum error calculated by Cubic-RBF-ARX model allows optimization of the dry tail dry stage head outlet water phase predicted value of the set value and the outlet water , to find out the optimal parameter input stage dry head curve drying tail dry stage of the process to accommodate variations in the incoming case to reduce the quantity of dry tail dry stage.

[0015] step 1), the first stage Dryer Drying Cubic-RBF-ARX model:

[0019] wherein, yH (tH) represents the water outlet head Dryer Drying stage Cubic-RBF-ARX model;

Respectively represent moisture exhaust throttle opening degree Cubic-RBF-ARX model dry first stage, air temperature, cylinder temperature, the inlet flow rate and an inlet water; XH (tH-l) is the state the inlet flow and the inlet water variables; np H, nqH , dH and orders have indicated mH dry Cubic-RBF-ARX model head stage; Zf, Zf are first dry stage

RBF neural network Cub i c-RBF-ARX model output in terms of entry.

Scalar weights Cubic-RBF-ARX model dry first stage; M · I | F represents Frobenius matrix norm; | H (tH) is a modeling error Cubic-RBF-ARX model dry first stage, is Gaussian white noise; T QH dryer drying of the head stage Cubic-RBF-ARX model modeling sampling time, T1 is an inlet flow rate detection value from the inlet moisture detection time value, T 2 is the detected value from the inlet water has outlet moisture detection time value, T 3 is the detected value of the moisture from the inlet tube to the inlet drying time! \ Leaf silk yarn in the drying cylinder drying time.

[0020] step 1), the tail Dryer Drying stage Cubic-RBF-ARX model:

[0021]

"00241 where, vT (tT) illustrates co hereby Dryer drying stage tail Cubic-RBF-ARX model water outlet;

Denote cylinder temperature tail dry stage Cub i c-RBF-ARX model, hot air temperature, moisture exhaust throttle opening, the inlet flow rate, inlet water and the frequency of the motor cylinder; XT (tT_l) to moderate hot air cylinder motor frequency state variables; order npT, nqT, dT and m% represents Cubic-RBF-ARX model dry tail stage; <; << were dry tail phase cub i c-RBF-ARX model output term to the input terms RBF Center for neural network;

Scalar weights Cubic-RBF-ARX model tail dry stage; f (tT) tail dry stage Cubic-RBF-ARX modeling error is Gaussian white noise; IV to the tail dry stage cut tobacco drying machine Cubic-RBF- ARX ​​modeling sampling time.

[0025] step 2), the first stage Dryer Drying Cubic-RBF-ARX model optimization as follows:

[0027] wherein, Yan (Yan) is the actual value of the outlet water Drying Dryer first stage, yield (Yan) under actual input action, calculated by Drying Dryer first stage Cub i c-RBF-ARX model for an the predicted value of the water outlet;

Drying machine head% dry stage linear parameter Cubic-RBF-ARX model;

A nonlinear parameter Cubic-RBF-ARX model head Dryer Drying stage; Nh Drying machine for the Cubic-RBF-ARX model modeling data length header dry stage.

[0028] Dryer Drying stage tail Cubic-RBF-ARX model optimization as follows:

[0030] wherein, / (r) is the actual value of the tail Dryer Drying process water outlet; R & lt (r〇 under actual input action, calculated by the tail Dryer Drying stage Cub i C-RBF-ARX model water outlet predicted value;

Drying machine for the tail dry stage linear parameter Cubic-RBF-ARX model,

A nonlinear parameter Cubic-RBF-ARX model tail dry stage cut tobacco drying machine; Nt of tobacco drying machine as Cubic-RBF-ARX model modeling data length header dry stage.

[0031] step 3):

Double S function expression [0032] Drying machine for describing the first stage of the dry moisture exhaust damper, air temperature, barrel temperature optimum curve for the input:

[0034] wherein, TS is the input of the time, in units of s; λ i, λ4, \ 5 are double S starting function, the turning point and the end point value; λ2, λ6 are two pairs of S-shaped function symmetry axis center position; λ 3, λ 7 are double-S-shaped function increase or decrease speed; λ 3, when λ 7 is greater than 0 represents S type function rising, λ 3, λ 7 is less than 0 represents S type function decline; c = l, 2, 3, Usl (ts) is a set value of moisture exhaust damper; Us2 (ts) is a set value of the air temperature; U s3 (ts) is a set value of the barrel temperature.

Step function expression [0035] used to describe the optimal inlet flow input curve Dryer Drying head stage is:

[0037] wherein, tT is the time input unit s; κ κ 2, κ 3 is a step function increase in velocity, the rise time and the final values.

[0038] step 4), the first stage Dryer Drying Cub i c-RBF-ARX model to calculate the predicted values ​​of the outlet water

By setting the input variable optimization curve Dryer Drying substituting first stage of the process variables of each dry first stage constructed Cub i c-RBF-ARX of

Obtained. By dry stage head Cubic-RBF-ARX model to calculate the predicted values ​​of the outlet water and the outlet water setpoint ysrt (ta) error eH (ta) the smallest, i.e., a method using optimization problem is solved 列维布格奈奎 Technology

Finding the first stage of the dry moisture exhaust damper, Parameter λχ inlet flow input curve and the curve of the input air temperature, a barrel temperature of κ κ2, κ3; wherein ^ 1,2, ..., 7 = 2,3 "dry first phase duration.

[0039] step 3), the end of the stage is used to describe dry moisture exhaust damper, the optimal expression of an exponential function of the input air temperature profile, the cylinder barrel mild motor frequency is:

[0041] wherein Uzl (tz), Uz2 (tz), Uz3 (tz), Uz4 (tz) represent the end stage of the dry moisture exhaust damper, optimal input air temperature curve, the motor cylinder barrel moderate frequency.

[0042] step 4), Cubic-RBF-ARX model tail dry stage calculated predicted value of the water outlet

By setting the input variable optimization curve substituting tail Dryer Drying stages of process variables constructed tail dry stage Cub i c-RBF-ARX model "6V) obtained; tail dry stage by Cubic-RBF- water outlet ARX model predictive value calculated Chan P) and the water outlet of the minimum set value y'srt (tb) error eT (tb), i.e., using the method of optimization problem is solved 列维布格奈奎 Technology

Find out the tail dry stage moisture exhaust damper, air temperature, motor cylinder barrel moderate frequency input parameters apg optimal curve; wherein, g = l, 2, 3; M 'is the duration of the tail dry stage.

[0043] Compared with the prior art, the present invention has beneficial effects: dry method of the present invention allows the first leaf stage wire exit moisture increased as quickly as possible, and rapidly reach a steady state, the tail dry stage cut tobacco can water was slowly lowered as the outlet, thus effectively reducing the quantity of dry head and tail section, improve the control performance drying process with great economic value; the method of the present invention is considered to quantity between the respective input variables dynamic characteristics, can be more effectively overcome the effects of changes in water flow and incoming drying head and tail section, the control section when the head and tail of the inlet flow to the inlet water applied in different patterns cut tobacco; method of the present invention is based on the model identification optimization of the optimal input setting curves, avoiding the manual tuning process variable input parameters inconvenience.

BRIEF DESCRIPTION

[0044] FIG. 1 is a schematic process Drying machine.

Detailed ways

[0045] Drying process unit as shown in FIG. Before entering the cut tobacco cut tobacco drying step, the inlet flow is first detected yarn leaves the water inlet U4 and U5. After time T3, leaf tobacco drying machine wire to reach the entrance. Moisture exhaust throttle opening when the cut tobacco Drying cylinder drying, the system periodically sampling cylinder U 1, air temperature U2, U3 cylinder temperature and other process variable parameters. The drying cycle duration time T4, after the cut tobacco drying Drying decanted from the barrel outlet, and the measured water content value y leaf wire exit at the outlet. A flow detection value from the inlet to the outlet moisture detection value to be experienced a longer period of time, for example, a production line Drying takes about 340s. Further, between the input / output variables Drying machine also has a large delay.

[0046] When detecting the inlet flow rate, indicating Drying process runs. Drying processes have the initial operation leaves yarn inlet and the inlet water flow rate detection value, there is no outlet water detected value of cut tobacco, when the first dry Drying process stage begins. Drying processes according to characteristics of the first stage of the dry, model structure Cubic-RBF-ARX:

[0050] wherein, yH (tH) represents the water outlet head Dryer Drying stage CubiC-RBF-ARX model;

Respectively represent moisture exhaust throttle opening degree Cubic-RBF-ARX model dry first stage, air temperature, cylinder temperature, the inlet flow rate and an inlet water; XH (tH-l) is the state the inlet flow and the inlet water variables; np H, nqH , dH have indicated mH and the order of the first stage of the dry Cubic-RBF-ARX model;

Stem head center stage Cub i c-RBF-ARX model output items are the entry and RBF neural network

Scalar weights Cubic-RBF-ARX model dry first stage; M · I | F represents Frobenius matrix norm; | H (tH) is a modeling error Cubic-RBF-ARX model dry first stage, is Gaussian white noise; T QH dryer drying of the head stage Cubic-RBF-ARX model modeling sampling time, T1 is an inlet flow rate detection value from the inlet moisture detection time value, T 2 is the detected value from the inlet water has outlet moisture detection time value, T 3 is the detected value of the moisture from the inlet tube to the inlet drying time! \ Leaf silk yarn in the drying cylinder drying time.

[0051] When the normal inlet flow becomes 0, marking the start of the dry end of the process, when the outlet water dropped to 3%, marking the end of the whole machine Drying Drying process. Tail dry process without the inlet flow rate detection value, but with an outlet moisture detection value. The characteristics of the tail Dryer Drying process section, following the establishment of Cubic-RBF-ARX model:

[0052]

[0055] wherein, yT (tT) represents the outlet water Cubic-RBF-ARX model Drying Dryer tail stage; denote cylinder temperature of the dry tail phase Cub i c-RBF-ARX model, hot air temperature, moisture exhaust damper opening, the inlet flow rate, inlet water and the frequency of the motor cylinder; XT (tT_l) state of mild hot air cylinder motor variable frequency; order npT, nqT, dT and m% represents Cubic-RBF-ARX model tail dry stage views; zP 'are centered on the end of a dry RBF neural network stage Cubic-RBF-ARX model output term to the input terms;

Scalar weights Cubic-RBF-ARX model tail dry stage; f (tT) tail dry stage Cubic-RBF-ARX modeling error is Gaussian white noise; IV to the tail dry stage cut tobacco drying machine Cubic-RBF- ARX ​​modeling sampling time.

[0056] The present invention is structured nonlinear parameter optimization method (SNPOM) method to estimate the model. In order to make Cubic-RBF-ARX model can be constructed in the above described global dynamic characteristics of head and tail Drying section, we first use the model to optimize SNPOM method, step prediction error parameters the minimum case, and as long as the parameters optimize the initial predicted model parameters at the target value. Then, using 列维布格奈奎 Gault method (LMM) to optimize the optimal model parameters to predict long-term performance.

[0057] The first stage Dryer Drying Cubic-RBF-ARX model (1) the parameters of the optimization problem as follows:

[0059] wherein the yield (Yan) is the actual value of the outlet water Drying Dryer first stage, r (Yan) under actual input action, calculated from the drying wire Dryer first stage Cub i c-RBF-ARX model the predicted value of the water outlet;

Linear parameter Cubic-RBF-ARX model head Dryer Drying Phase

A nonlinear parameter Cubic-RBF-ARX model head Dryer Drying stage; Nh Drying machine for the Cubic-RBF-ARX model modeling data length header dry stage.

[0060] Dryer Drying stage tail Cubic-RBF-ARX model (3) of the parameter optimization problem as follows:

[0062] wherein, / (r) is the actual value Drying machine dry end of the process the outlet water; r (r) is the actual input action, calculated by Drying Dryer End stage Cub i c-RBF-ARX model water outlet predicted value;

Linear parameter Cubic-RBF-ARX model stage tail Dryer Drying

A nonlinear parameter Cubic-RBF-ARX model tail dry stage cut tobacco drying machine; Nt of tobacco drying machine as Cubic-RBF-ARX model modeling data length header dry stage.

[0063] Drying Dry head stage Cubic-RBF-ARX model of the process according to the estimated optimal design input curve for each process variable, to adapt to changes in the incoming case to minimize the quantity of dry dry phase of the head. The present invention uses a double S-shaped function will be described first stage of the dry moisture exhaust damper, optimal input air temperature curve, the barrel temperature, using step-function input curve to describe the optimal inlet flow.

[0064] S-shaped double curve following formula:

[0066] wherein, TS is the input of the time, in units of s; λ i, λ4, \ 5 are double S starting function, the turning point and the end point value; λ 2, λ 6 are two pairs of S-shaped function symmetry axis center position; λ 3, λ 7 are double-S-shaped function increase or decrease speed; λ 3, λ 7 represents S type function rising, λ 3, when λ 7 is smaller than the square represents S-shaped function decline greater than square; c = l, 2, 3, Usl (ts) is a set value of moisture exhaust damper; Us2 (ts) is a set value of the air temperature; U s3 (ts) is a set value of the barrel temperature.

[0067] The step-described function formula inlet flow input curve as follows:

[0069] wherein, tT is the time input unit s; κ ρ κ 2, κ 3 is a step function increase in velocity, the rise time and the final values.

[0070] The optimized set of process variables for each curve (7-8) into constructed CubiC-RBF-ARX model (1) 4 input variables (square, 4 (square, 4 (square, 4 (square in ? a prediction value obtained dry phase outlet header 37 water / (○:

[0072] The 列维布格奈奎 Technology (Levenberg-Marquardt Method, LMM) method, calculated by the model to the water outlet of the prediction error value and the minimum water content set value of the outlet, to find out the first stage of the dry moisture exhaust damper , air temperature, cylinder temperature optimal input curve AiQ = Ij, " '?) the input parameters and the optimal inlet flow curve Kj (j = l, 2,3) parameter set values ​​and the water outlet head on a dry phase Stem head dynamic error between the model predictions (9):

[0074] yset (ta) is the water outlet setpoint.

[0075] The first stage of the optimization problem dry optimal set of process variables as follows:

[0077] M is the duration of the first stage dry. Setting parameter values ​​to obtain optimal curve by solving the above optimization problem to design the optimal input curve Dryer Drying head at various stages of the process variables.

[0078] Drying process tail dry stage Cubic-RBF-ARX model estimated according to the optimal design input curve for each process variable, to adapt to changes in the incoming case to minimize the quantity of dry tail dry stage. Using the exponential function tail dry stage described moisture exhaust damper, optimal input air temperature curve, the motor cylinder barrel moderate frequency, the exponential curve formula is as follows:

[0080] wherein Uzl (tz), Uz2 (tz), U z3 (tz), Uz4 (tz) represent the end stage of the dry moisture exhaust damper, optimal input air temperature curve, the motor cylinder barrel moderate frequency. The optimized set of process variables for each curve (12) into constructed Cubic-RBF-ARX model (3) of the input variable "ff), the obtained dry water outlet end of stage prediction value:

[0082] The 列维布格奈奎 Stewart (the LMM) method, calculated by the model to the water outlet of the prediction error value and the minimum outlet water setpoint, locate a discharge end of tidal damper dry, air temperature, cylinder optimal motor cylinder moderate frequency input apg curve; wherein, g = l, 2,3. Water outlet tail dry stage based on an error between the setting value and the dry end of the dynamic model prediction value (13) as:

[0084] ysrt⑴ export water setpoint.

[0085] Optimization tail dry stage optimal set of process variables as follows:

[0087] M is the duration of the tail dry stage. Setting parameter values ​​to obtain optimal curve by solving the above optimization problem to design the optimal input tail Dryer Drying curves at various stages of the process variables.

Claims (3)

A head end section Drying process variables optimization control method, characterized in that the method: 1) The operational flow Drying machine, establish the middle silk Drying process flow inlet, water inlet, barrel temperature, wind temperature, moisture exhaust damper, the timing relationship of the outlet water, the first stage while dry wire exit moisture detecting no leaf tobacco drying process value based on the dry non-leaf stage tail yarn inlet and the inlet flow characteristics of the moisture detection values, as a function of the radial cubic Cubic-RBF-ARX model basis functions are established drying process stage and dry head Cubic-RBF-ARX model tail dry stage; 2) drying the historical operating data head tail section, structured nonlinear parameters optimization methods optimization drying process Stem head stage and Cubic-RBF-ARX model dry end of stage; 3) according to drying process Stem head stage and Cubic-RBF-ARX model dry end of stage optimized, double S-shaped function optimal input curve moisture exhaust damper head described dry phase, air temperature, the cartridge temperature; described step function using a dry phase inlet flow header optimum input curve; described exponential function tail dry stage moisture exhaust damper , Air temperature, optimal input curve cylindrical barrel moderate frequency motor; 4) using the method of Stewart 列维布格奈奎, by optimizing the first stage and the dry stage of the dry tail Cubic-RBF-ARX model calculated water outlet predicted value and the set value of the water outlet smallest error to find out the optimal parameter input stage dry head curve drying tail dry stage of the process to accommodate variations in the incoming case to reduce the quantity of dry tail dry stage ; step 1), the first stage dryer drying Cubic-RBF-ARX model:

Wherein, yH (tH) denotes an outlet water Cubic-RBF-ARX model Drying Dryer first stage; <(yield), Oii), "f (square," fC <(yield) represent dry first stage Cubic-RBF moisture exhaust throttle opening -ARX model, air temperature, cylinder temperature, and the inlet water flow inlet; XH (t Hl) and the inlet of the inlet flow of water state variables; npH, nqH, d H m H and each represents a dry head order Cubic-RBF-ARX model stage; $ ', Zf are dry first stage RBF neural network center Cubic-RBF-ARX model output term to the input terms; dry head phase scalar weights Cubic_RBF_ARX model; II · I If represented Frobenius matrix norm; ξ H (tH) is the first phase of the modeling error dry Cubic-RBF-ARX model is Gaussian white noise; Tci11 dry first stage Cubic-RBF-ARX model drying machine building mode sampling time, T from the flow rate detecting values ​​inlet to time the inlet water detection value, T2S from the inlet moisture detection value the time with an outlet moisture measurement value, T3 for the inlet of water detected value to a drying wire tube from the inlet time, T 4 is the cut tobacco drying cylinder drying time; step 1), the tail dryer drying stage Cubic-RBF -ARX model:

among them:

Wherein, yT (tT) represents a water outlet Cubic-RBF-ARX model tail dry stage cut tobacco drying machine; <(iK (square, <(/), :), respectively tail dry stage Cub i c-RBF-ARX model the barrel temperature, hot air temperature, moisture exhaust throttle opening, the inlet flow rate, inlet water and the frequency of the motor cylinder; XT (tT-l) is a gentle hot air cylinder motor frequency state variables; npT, nqT, dT and m order% represents Cubic-RBF-ARX model dry tail stage; ζί; ν, ζί; "respectively, the center RBF neural network dry end of stage Cubic-RBF-ARX model output term to the input terms;, <^ () , '6Clg scalar weights Cubic-RBF_ARX model dry tail stage; f (tT) is the dry end of stage Cubic-RBF-ARX modeling error is Gaussian white noise; IV is a tobacco drying machine dry end of stage Cubic-RBF -ARX modeling sampling time;) in the step 2, the first stage dryer drying Cubic-RBF-ARX model optimization as follows:

Wherein ,, (P) is the actual value of the water outlet head Dryer Drying stage, ^ V) is under the action of the actual input, the predicted value calculated by the first stage the dry Cubic-RBF-ARX model Drying water outlet ; rape gas two wide Λ '<= 4 »strict = 1, ·, called = 1 ,, ..,.."} is produced dryer drying head II, 0 f, _ /, 0 "J stage! linear parameter Cubic-RBF-ARX model; Si Z (,, non-linear parameter Cubic-RBF-ARX model head dryer drying stage; Nh is the head dryer drying stage Cubic-RBF-ARX model modeling data length; tail dry stage drying machine Cubic-RBF-ARX model optimization as follows:

Wherein ,, (square is the actual value of the tail Dryer Drying process water outlet; under action of the actual input, the predicted value calculated by the tail dry stage Cubic-RBF-ARX model Drying water outlet; [Theta] Bu <| linear parameter Cubic-RBF-ARX model 4'0, play qe, gas = 1, ..., n / 7ry stage, [theta] "ν = (Zf 'Zf BIAN = 1, ·," /} of drying dryer Nonlinear parameter Cubic-RBF-ARX model tail stage, Nt is the first stage dryer drying Cubic-RBF-ARX model modeling data length;) in the step 3: drying dryer for describing a first stage moisture exhaust damper, air temperature, optimum barrel temperature profile double S input function expression is:

Wherein, TS is the input of the time, in units of s; λ λ4, \ 5 are double S starting function, the turning point and the end point value; λ2, \ 6 are axial center position of two pairs of symmetrical S-shaped function; [lambda] 3, λ 7 are pairs of S-shaped function of the speed increase or decrease; λ 3, λ 7 when the S-shaped function represents Taiyu square rise, λ 3, λ 7 when the square is smaller than the S-shaped function represents decreased; c = 1,2, 3, Usl (ts) is a set value of moisture exhaust damper; Us2 (ts) is a set value of the air temperature; Us3 (ts) is a set value of the barrel temperature; dryer drying for describing the first stage inlet flow the optimal input function expression is a step profile:

Wherein, tT is the time input unit is S; K K2, K3 are rising speed of a step function, the rise time and the final value; 3) in the step for describing tail dry stage moisture exhaust damper, the air temperature expression exponential function curve of the optimum input frequency of the motor cylinder barrel is moderate:

P = 1,2, 3, 4 wherein Uzl (tz), Uz2 (tz), Uz3 (tz), Uz4 (tz) represent the end stage of the driest moisture exhaust damper, the air temperature, the cylinder drum motor frequency moderate preferably input curve.

2. Drying the head end section of the process variables according to claim 1, optimal control, wherein said step 4), the head Dryer Drying optimization stage set curve Mt s), Ut (tT ) into said cut tobacco drying machine dry head stage Cubic-RBF-ARX model input variables <(Π, "f (0, (0, 4 (0, the obtained drying dryer first stage Cubic-RBF-ARX model calculated Outputs water outlet of the predicted value (o; head by a dry phase Cubic-RBF-ARX model to calculate the predicted values ​​of the outlet water Chan (O moisture set value and the minimum outlet y srt (ta) error eH (ta), i.e., a method using column M 维布格奈奎 Technology optimization problems ininJ =, find the first stage of the dry moisture exhaust damper, and an inlet flow parameters λχ input curve input air temperature curve, the cartridge Av. κσ temperature parameters KK 2 , K 3; wherein, X = 1,2, ..., 7; g = 1,2, 3; M is a first phase duration dry.

The head end section Drying process variables of the optimization control method according to claim 1, wherein said step 4), the end of the optimization phase Dryer Drying curves set Uzp (tz) substituting the input variables Cubic-RBF-ARX model drying machine dry end of phase "" (wide) "[(Canton)," [#), to give Cubic-RBF-ARX model drying dryer tail phase calculated water outlet predicted value of the Y Yin); calculated by the tail dry stage Cubic-RBF-ARX model predicted value of the water outlet / f) and the outlet water setpoint y 'srt (tb) error eT (tb) minimum , i.e. using 列维布格奈奎 Technology M square method to solve the optimization problem min./ '= (^), locate a discharge end of tidal damper dry, air temperature, motor cylinder barrel moderate frequency aps k = \ optimal input curve parameters a pg; wherein, g = 1,2, 3; M 'is the duration of the tail dry stage.