CN106873376A - Glassed steel reaction vessels control based on network method based on scheduling with controller parameter dynamic restructuring - Google Patents

Abstract

Glassed steel reaction vessels control based on network method based on scheduling with controller parameter dynamic restructuring, comprises the following steps：1), the model of control object recognize according to actual condition and obtain object model parameter；According to system IAE performance indications parameter setting schedulers threshold value, scheduling strategy, system communication cycle, system transfers cycle initial value, sample sequence and renewal sequence, system model is (1)；2), according to current network service quality dynamic restructuring communication sequence and online analyzing computing controller parameter；3) exported to actuator after, being changed by D/A, controlled device is applied to by actuator, controlled device is operated in given scope.The present invention can be controlled the Single Controller of device dynamic state of parameters reconstruct according to network real-time status, overcome original controller architecture complicated and by communication sequence transmission cycle restricted problem, multiple controllers are avoided periodically to switch, it is convenient to be grasped and promoted the use of by engineers and technicians.The present invention can be effectively improved network environment, increase network utilization and system overall performance.

Description

Glassed steel reaction vessels control based on network based on scheduling with controller parameter dynamic restructuring Method

Technical field

The invention belongs to Industrial Ethernet Control technical field, refer to a kind of based on scheduling and controller parameter dynamic The glassed steel reaction vessels control based on network method of reconstruct.

Background technology

Synthetic reaction directly affects the quality of medicine as one of important procedure in medicinal chemicals preparation process, Glassed steel reaction vessels as pharmacy reaction visual plant, its technology development with equipment improve turned into medicinal chemicals manufacture work The key of industry transition and upgrade.The quality of reaction will directly affect yield, quality of bulk drug etc.；Meanwhile, the net based on Ethernet Network control method is also to improve the essential requirement of drug quality.

Network control system is a kind of full distributed, networking real-time feedback control system, refer to sensor, controller and The set of actuator and communication network, with the data transfer between communication network offer equipment, so as to realize System Resources Sharing And coordinated manipulation.At present, the method for designing for the limited network controller of multivariable communication generally using communication sequence and is mixed When legacy network control system is transformed into collection scheduling and is controlled in the discrete periodic of one by the concept of logical Dynamic Architecture Become switching system, and then design corresponding controller.The number of this kind of method for designing generally existing controller will receive original communication The constraint of sequence period number and complex structure；Further, when network service quality occurrence dynamics change, often need Multiple controller periodicity frequent switchings are wanted just to can ensure that the performance of control system；On the other hand, the phase used in this kind of method Close professional theory knowledge more, be not easy to be grasped and promoted the use of by engineers and technicians.

The content of the invention

Exist to overcome in the multiple input/multiple output glassed steel reaction vessels network control system with random short time-delay Control gain, communication sequence cannot dynamic restructuring problem, the present invention provide can be dynamically adjusted according to system IAE performance indications The transmission means of information, overcomes network transmission to dispatch the priori for being limited to current transmission network utilization, it is impossible to self adaptation The dynamic change of network, it is impossible to ensure that whole closed network networked control systems have good anti-interference and robustness, is disliking Even controller can be caused to fail in the case of bad, so that whole closed loop network control system does not reach the control being pre-designed Performance issue.

Meanwhile, the present invention can be controlled the Single Controller of device dynamic state of parameters reconstruct according to network real-time status, overcome Original controller architecture is complicated and receives communication sequence transmission cycle restricted problem, it is to avoid multiple controllers periodically switch, convenient Grasped and promoted the use of by engineers and technicians.The present invention can be effectively improved network environment, increase network utilization and system Overall performance.In order to solve the above-mentioned technical problem the technical scheme for using for：

Glassed steel reaction vessels control based on network method based on scheduling with controller parameter dynamic restructuring, including following step Suddenly：

Step 1, System Discrimination and parameter Estimation such as formula (1) are carried out to glassed steel reaction vessels according to actual condition：

WhereinRepresent that r ties up the state vector of glassed steel reaction vessels；Represent m dimension enamel reactions The valve opening input vector of kettle, τ_kThe stochastically bounded network short time-delay in each sampling period is represented, and meets 0≤τ_k≤eh (0 ＜ e ＜ 1) h is the sampling period of control system,Portion's disturbing signal of q dimensions is represented,Represent that enamel is anti- Answer the vectorial by tune output temperature, pressure signal of kettle, A_p,B_p,C_pIt is the sytem matrix of appropriate dimension；

Step 2：Sampling period h, step-length N dispatching cycle, T dispatching cycle of control system are set, and are met：T=Nh； The scheduling coefficient θ of glassed steel reaction vessels kT+ α h instance sample sequences is set_i(kT+ α h) threshold value η_iWith kT+ β h moment renewal sequences Scheduling coefficient δ_j(kT+ β h) threshold valueAnd set shown in scheduling strategy such as formula (2)：

(1 ..., r), (1 ..., m), β ∈ (0 ..., N-1), IAE represents the exhausted of signal to α ∈ (1 ..., N) j ∈ to wherein i ∈ To error performance.

As k=1, initial schedule matrix Λ is set_α=I, Ξ_β=I, otherwise generates current dispatching cycle according to scheduling strategy Sampling schedules matrix in TWith renewal dispatch matrix

WhereinFor kT+ α h instance sample sequences scheduling coefficient composition to Amount,It is the vector of kT+ β h moment renewal sequences scheduling coefficient composition；

Step 3:System sets up following mapping relations with T discretizations dispatching cycle according to the scheduling strategy of step 2：

WhereinIt is the input state vector of controller after system call,It is to be controlled after system call The output vector of device；

Constructing system augmentation vector matrix includes：State vector matrix X (kT), controller exports U (kT), controller to be had Effect inputEffective input of actuatorAdjusted output Z (kT), external disturbance signal W (kT), sampling schedules square Battle arrayUpdate dispatch matrixWithAnd it is as follows to set up new mapping relations：

Wherein：

Step 4:System mode feedback controller is setReconfigure new state vectorM, N, L are as follows：

L=H₁' and set up collection scheduling The closed-loop model (5) being integrated with control：

Wherein：

K_kIt is controller parameter, τ_kd----represent d-th time delay in sampling period, A, B in k-th dispatching cycle₀(τ_k)、 B₁(τ_k1)、H₀、A_e、B₀′(τ_k)、B₁′(τ_k)、H₀′、D₀′(τ_k)、C_e′、D₁′(τ_k)、H₁' it is intermediate computations variable,

Step 5：Reset calculating variable And the optimization problem below line solver:

Using feasible solution obtained in the previous step, state feedback controller parameter can be obtainedWhereinRepresenting matrix Pseudoinverse；Work as k=k+1, come back to Step 3 and solve controller parameter；Temperature control signals are exported to heat after being changed by D/A The magnetic valve of oily device, pressure controling signal is exported to pressure-regulating valve after being changed by D/A, is finally applied to enamel reaction Kettle, makes the temperature and pressure of glassed steel reaction vessels operate in given scope.

Beneficial effects of the present invention are mainly manifested in：Single Controller, system transfers cycle dynamic restructuring, controller parameter Dynamic restructuring simple structure.

Brief description of the drawings

Fig. 1 is the glassed steel reaction vessels closed loop controlling structure figure that the present invention is given

The structural representation used when Fig. 2 is actual motion of the present invention

Implementation method

The invention will be further described below in conjunction with the accompanying drawings.

1~Fig. 2 of reference picture, a kind of glassed steel reaction vessels control based on network based on scheduling with controller parameter dynamic restructuring Method, the present embodiment glassed steel reaction vessels temperature, pressure network control process：

The first step, on configuration interface set Model Distinguish relevant parameter, determine the identification model of glassed steel reaction vessels It is as follows：

Wherein：X (t) is the state vector of system, u (t)=[u₁ u₂]^TIt is control input vector, u₁Represent steam in tank Valve opening, u₂Represent condensed water valve opening.Z (t)=[z₁ z₂]^TIt is controlled output vector, z₁Represent temperature in tank, z₂Generation Table pressure inside the tank.W (t) is external disturbance vector.The data such as target model identification parameter are sent to memory cell RAM by industrial computer In；And it is " offline " adjustment state that system is set on configuration interface.One is set up according to the closed loop controlling structure figure shown in accompanying drawing 2 Individual closed-loop control system.

Second step：System communication cycle h=1s, step-length N=2, the IAE controlling of dispatching cycle are set on configuration interface The threshold value of energy index is respectively η₁=0.01, η₂=0.05,

3rd step：" RUN " key is clicked on configuration interface, in communication sequence, the controller parameter of line computation scheduling.

4th step：Amplitude limit is carried out to u (k), prevents from integrating saturation, exported to actuator after then being changed by D/A, by performing Device is applied to controlled device, controlled device is operated in given scope；Online situation is now shown on configuration interface Under system closed-loop response curve, observation curve carry out on-line fine.

5th step：It is " online " adjustment state that system is set on configuration interface, starts control system parameter regulation, again Perform the controlled quentity controlled variable that " On-line Control program " obtains current time.

Described above is the excellent effect of optimization that one embodiment that the present invention is given shows, it is clear that the present invention is not only Above-described embodiment is limited to, without departing from essence spirit of the present invention and the premise without departing from scope involved by substance of the present invention Under it can be made it is a variety of deformation be carried out.

Claims (1)

1., based on the glassed steel reaction vessels control based on network method dispatched with controller parameter dynamic restructuring, it includes following step Suddenly：

Step 1, System Discrimination and parameter Estimation such as formula (1) are carried out to glassed steel reaction vessels according to actual condition：

$\left\{\begin{array}{c}\stackrel{\·}{x}\left(t\right)=A_{p}x(t,{\mathrm{\τ}}_k)+B_{p}u(t,{\mathrm{\τ}}_k)\\ z\left(t\right)=C_{p}x(t,{\mathrm{\τ}}_k)+w\left(t\right)\end{array}\right.---\left(1\right)$

WhereinRepresent that r ties up the state vector of glassed steel reaction vessels；Represent m dimension glassed steel reaction vessels Valve opening input vector, τ_kThe stochastically bounded network short time-delay in each sampling period is represented, and meets 0≤τ_k≤ eh (0 ＜ e ＜ 1) h for control system sampling period,Portion's disturbing signal of q dimensions is represented,Represent glassed steel reaction vessels By tune output temperature, pressure signal vector, A_p,B_p,C_pIt is the sytem matrix of appropriate dimension, R^x(r, m, q, n ∈ x) represents x dimensions Set of real numbers；

Step 2：Sampling period h, step-length N dispatching cycle, T dispatching cycle of control system are set, and are met：T=Nh；Set The scheduling coefficient θ of glassed steel reaction vessels kT+ α h instance sample sequences_i(kT+ α h) threshold value η_iWith the tune of kT+ β h moment renewal sequences Degree coefficient δ_j(kT+ β h) threshold valueAnd set shown in scheduling strategy such as formula (2)：

(1 ..., r), (1 ..., m), β ∈ (0 ..., N-1), IAE represents the absolute mistake of signal to α ∈ (1 ..., N) j ∈ to wherein i ∈ Difference performance.

As k=1, initial schedule matrix Λ is set_α=I, Ξ_β=I, otherwise generates in current dispatching cycle T according to scheduling strategy Sampling schedules matrixWith renewal dispatch matrix

WhereinIt is the vector of kT+ α h instance sample sequences scheduling coefficient composition,It is the vector of kT+ β h moment renewal sequences scheduling coefficient composition；

Step 3:System sets up following mapping relations with T discretizations dispatching cycle according to the scheduling strategy of step 2：

$\left\{\begin{array}{c}\stackrel{\‾}{x}(kT+\mathrm{\α}h)={\mathrm{\Λ}}_{\mathrm{\α}}x(kT+\mathrm{\α}h)\\ \stackrel{\‾}{u}(kT+\mathrm{\β}h)={\mathrm{\Ξ}}_{\mathrm{\β}}u(kT+\mathrm{\β}h)+(I-{\mathrm{\Ξ}}_{\mathrm{\β}})\stackrel{\‾}{u}\[kT+(\mathrm{\β}-1)h\]\end{array}\right.---\left(3\right)$

WhereinIt is the input state vector of controller after system call,It is the controller after system call Output vector；

Constructing system augmentation vector matrix includes：State vector matrix X (kT), controller export U (kT), controller it is effective defeated EnterEffective input of actuatorAdjusted output Z (kT), external disturbance signal W (kT), sampling schedules matrixUpdate dispatch matrixWithAnd it is as follows to set up new mapping relations：

$\left\{\begin{array}{c}\stackrel{\‾}{X}\left(kT\right)=\stackrel{\‾}{\mathrm{\Λ}}\left(kT\right)X\left(kT\right)\\ \stackrel{\‾}{U}\left(kT\right)={\stackrel{\‾}{\mathrm{\Ξ}}}_0\left(kT\right)U\left(kT\right)+{\stackrel{\‾}{\mathrm{\Ξ}}}_1\left(kT\right)\stackrel{\‾}{U}\left(kT\right)\end{array}\right.---\left(4\right)$

Wherein：

$X\left(kT\right)=\left[\begin{array}{c}x\[(k-1)T+h\]\\ x\[(k-1)T+2h\]\\ .\\ .\\ .\\ x\left(kT\right)\end{array}\right],U\left(kT\right)=\left[\begin{array}{c}u\left(kT\right)\\ u(kT+h)\\ .\\ .\\ .\\ u\[kT+(N-1)h\]\end{array}\right]$

$\stackrel{\‾}{X}\left(kT\right)=\left[\begin{array}{c}\stackrel{\‾}{x}\[(k-1)T+h\]\\ \stackrel{\‾}{x}\[(k-1)T+2h\]\\ .\\ .\\ .\\ \stackrel{\‾}{x}\left(kT\right)\end{array}\right],\stackrel{\‾}{U}\left(kT\right)=\left[\begin{array}{c}\stackrel{\‾}{u}\left(kT\right)\\ \stackrel{\‾}{u}(kT+h)\\ .\\ .\\ .\\ \stackrel{\‾}{u}\[kT+(N-1)h\]\end{array}\right],Z\left(kT\right)=\left[\begin{array}{c}z\left(kT\right)\\ z(kT+h)\\ .\\ .\\ .\\ z\[kT+(N-1)h\]\end{array}\right],$

${\stackrel{\‾}{\mathrm{\Ξ}}}_1\left(kT\right)=\left[\begin{array}{cccc}0& \mathrm{...}& 0& I-{\mathrm{\Ξ}}_0\\ 0& \mathrm{...}& 0& (I-{\mathrm{\Ξ}}_1)(I-{\mathrm{\Ξ}}_0)\\ .& & .& .\\ .& \mathrm{...}& .& .\\ .& & .& .\\ 0& \mathrm{...}& 0& {\mathrm{\Π}}_{i=1}^{N-1}(I-{\mathrm{\Ξ}}_i)\end{array}\right],$

Step 4:System mode feedback controller is setReconfigure new state vectorG、M、 N, L are as follows：

$\tilde{X}\left(kT\right)={\left[\begin{array}{cc}{X}^T\left(kT\right)& {\stackrel{\‾}{U}}^T\[(k-1)T\]\end{array}\right]}^T,$

$G=\left[\begin{array}{cc}{A}_e+B_0^{\′}\left({\mathrm{\τ}}_k\right){\stackrel{\‾}{\mathrm{\Ξ}}}_0\left(kT\right)K_k\stackrel{\‾}{\mathrm{\Λ}}\left(kT\right)& B_0^{\′}\left({\mathrm{\τ}}_k\right){\stackrel{\‾}{\mathrm{\Ξ}}}_1\left(kT\right)+B_1^{\′}\left({\mathrm{\τ}}_k\right)\\ {\stackrel{\‾}{\mathrm{\Ξ}}}_0\left(kT\right)K_k\stackrel{\‾}{\mathrm{\Λ}}\left(kT\right)& {\stackrel{\‾}{\mathrm{\Ξ}}}_1\left(kT\right)\end{array}\right],$

L=H '₁And set up collection scheduling and control It is made as the closed-loop model (5) of one：

$\left\{\begin{array}{c}\tilde{X}\[(k+1)T\]=G\tilde{X}\left(kT\right)+MW\left(kT\right)\\ Z\left(kT\right)=N\tilde{X}\left(kT\right)+LW\left(kT\right)\end{array}\right.---\left(5\right)$

Wherein：

K_kIt is controller parameter, τ_kd----represent d-th time delay in sampling period, A, B in k-th dispatching cycle₀(τ_k)、B₁ (τ_k1)、H₀、A_e、B₀′(τ_k)、B₁′(τ_k)、H₀′、D₀′(τ_k)、C_e′、D₁′(τ_k)、H₁' it is intermediate computations variable,

$A=e^{A_{p}h},B_0\left({\mathrm{\τ}}_k\right)={\∫}_0^{h-{\mathrm{\τ}}_{k1}}{e}^{A_{p}s}{\mathrm{dsB}}_p,B_1\left({\mathrm{\τ}}_{k1}\right)={\∫}_{h-{\mathrm{\τ}}_{k1}}^h{e}^{A_{p}s}{\mathrm{dsB}}_p,H_0={\∫}_0^h{e}^{A_{p}s}{\mathrm{dsH}}_p,$

$A_e=\left[\begin{array}{cccc}0& \mathrm{...}& 0& A\\ 0& \mathrm{...}& 0& A^2\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ 0& \mathrm{...}& 0& A^N\end{array}\right]$

Step 5：Reset calculating variable And the optimization problem below line solver:

$\begin{array}{c}\min r\\ \begin{array}{cc}s.t.& \left[\begin{array}{ccccc}-X& 0& {(C_e^{\&prime;}X+{\tilde{B}}_{1}Y)}^T& {(A_e^{\&prime;}X+{\tilde{B}}_{0}Y)}^T& {({\tilde{E}}_{0}Y+{\tilde{E}}_{1}X)}^T\\ *& -rI& L^T& M^T& 0\\ *& *& -I+\mathrm{\&epsiv;}{\tilde{D}}_1{\tilde{D}}_1^T& {\tilde{D}}_1{\tilde{D}}_0^T& 0\\ *& *& *& -X+\mathrm{\&epsiv;}{\tilde{D}}_0{\tilde{D}}_0^T& 0\\ *& *& *& *& -\mathrm{\&epsiv;}I\end{array}\right]<0\end{array}\\ X>0\end{array}---\left(6\right)$

Using feasible solution obtained in the previous step, state feedback controller parameter can be obtainedWhereinThe puppet of representing matrix It is inverse；Work as k=k+1, come back to Step 3 and solve controller parameter；Temperature control signals are exported to deep fat after being changed by D/A and filled The magnetic valve put, pressure controling signal is exported to pressure-regulating valve after being changed by D/A, is finally applied to glassed steel reaction vessels, is made The temperature and pressure of glassed steel reaction vessels is operated in given scope.