CN104330972A - Comprehensive prediction iterative learning control method based on model adaptation - Google Patents

Abstract

The invention relates to a comprehensive prediction iterative learning control method based on model adaptation, and belongs to the technical field of product quality tracking control of batch-type production and machining processes. For most batch machining and production processes, the method solves the problems that the production processes have random noise and convergence is slow, and the method is applicable to the product quality tracking control of the batch-type production and machining processes. The method can solve the problem of errors of a batch process model to some extent by the aid of a comprehensive prediction iterative learning control algorithm based on model adaptation, and convergence of the algorithm is accelerated. The method is wide in application range, ingenious in conception, simple and practical, can be widely applied to high-accuracy control in the batch production process of an industrial production line and can eliminate noise caused by real environments and machining means for the production processes.

Description

A kind of integrated forecasting iterative learning control method based on model adaptation

Technical field

The present invention relates to a kind of based on model adaptation integrated forecasting iterative learning control method, especially for this production run with random noise situation and the slower situation of speed of convergence, belong to the follow-up of product quality control technology field of batch production process.

Background technology

Batch production process (Batch Process) is a kind of production by batch to operate, and batch between there is the mode of production of certain quiescent interval.It is extensively present in national product as every field such as biological products, pharmaceutical production, fine chemistry industry, SIC (semiconductor integrated circuit), dominate in the industrial products of short run, multi items and high added value.In order to ensure the high-quality of product and quality thereof stability and, process control seems particularly important.But on the one hand, batch process has the features such as uncontinuity, unstable state, strong nonlinearity and time variation usually, and the accurate model setting up batch process is very difficult, and traditional control method generally can not play good effect; On the other hand, when filling a prescription constant in batch process is produced, production run reruns substantially, and within each batch of cycle of operation, control variable and product quality run along certain operation variation track, has stronger repeatability.In addition, in the production run of reality, because actual environment and manufacturing process inevitably bring interference or noise to production run, in the Product processing that accuracy requirement is higher, process noise brings serious impact may to final product quality.

Iterative learning controls (Iterative Learning Control, ILC) be very suitable for that this class has periodically, the control of the production run of the characteristic that reruns, its thought starting point is the system for repeating identical control task in finite time interval, and its performance can by improving to the study of repetitive process in the past.Iterative Learning Control Algorithm expects to utilize the previous or information of multiple batches to upgrade the input trajectory of next batch, makes output trajectory converge on the target trajectory of expectation as soon as possible.

For most interval controlled process, object concrete knowledge obtain not a duck soup, sometimes or even impossible, as: the operation batch limited inputoutput data that makes of batch process is limited; Be in initially some batches rerun the stage, Data mutuality degree is higher; Process has stronger non-linear, cannot use linear process matching; Modeling cost is more high, can not set up accurate system model in these cases or the model of foundation has larger deviation.Simultaneously due to batch process itself likely in the process repeatedly reruned running status change or change in operational process to making model due to external disturbance.Thus bring puzzlement to the Selecting parameter of control algolithm, therefore simple Iterative Learning Control Algorithm is difficult to direct use, or result of use is not good, is still difficult to through multiple batches the control object reaching expection.

In addition, namely comparatively accurate process model is compared in enable acquisition, if there is interference or noise in controlled process, iterative learning control effects may be deteriorated in batch upper integral effect due to control algolithm thus not reach expection object.Most of current Iterative Algorithm all supposes that process noise does not exist and then algorithm character is discussed, but in actual production process, process noise is often difficult to avoid, when certain process noise how Process Control algorithm to ensure that batch process product quality has to be solved.

Model Predictive Control (Model Predictive Control, MPC) be a kind of feedback advanced control algorithm, the output of Kernel-based methods model and system input prediction in the past system in the future, and on this basis by input that optimization method correction is current.For process model, the uncertain controlled process with there is process noise has good control effects to Model Predictive Control.Therefore in iterative learning control method, combination model PREDICTIVE CONTROL is a natural approach solved the problem.

In general, when parameter cannot select to cause conventional iterative learning algorithm speed of convergence comparatively slow or there is process noise, how designing appropriate integration iterative learning control method is a problem highly studied.

Summary of the invention

The object of this invention is to provide a kind of integrated forecasting iterative learning control method based on model adaptation, the method restrains the problem that there is noise in comparatively slow or process for batch process, can control batch process end product quality very well.The method calculates simple, calculates consuming time less, has good generalization.

The integrated forecasting iterative learning control method based on model adaptation that the present invention proposes, method comprises the following steps:

1) combine with actual production process, arrange a batch of data acquisition of producing and store link, this link can utilize the equipment such as the existing industrial control computer of manufacturing enterprise, PLC;

2) according to production process data in the past in the production history database that collects, after carrying out data prediction, the simple mathematical model of suitable method establishment production run is adopted;

3) data acquisition link collects the inputoutput data of Product processing in industrial production line, and according to target following trajectory calculation tracking error curve;

4) according to step 3) tracking error that obtains, adopt integrated forecasting Iterative Learning Control Algorithm to calculate the real-time controlled quentity controlled variable of next batch;

5) when starting for new batch, the model of production run is upgraded adaptively according to new service data;

6) at each new sampled point, implementation step 4), realize the effective tracking to exporting target trajectory.

Described step 2 in above-mentioned control method) in, the mathematical model method setting up production run is as follows:

1. according to production process data in the past in historical data base, after carrying out data prediction, suitable method establishment process mathematical model is adopted:

Assuming that certain moment input amendment collection U=(u ₁, u ₂..., u _m) ^t∈ R ^m, represent that m monitoring sensor is in historical data sometime, m represents the number of monitoring sensor, R ^mrepresent m dimensional vector; u _jrepresent in sample U, the single sampled data values of a jth sensing data, j=1,2 ..., m; The output sample in this moment integrates as Y=(y ₁, y ₂..., y _n) ^t∈ R ⁿ, represent that n monitoring sensor is in historical data sometime, n represents the number of monitoring sensor, R ⁿrepresent n dimensional vector; y _jrepresent in sample Y, the single sampled data values of a jth sensing data, j=1,2 ..., n, supposes to take N group historical data, and the total sample set of input data obtained is as follows: Q _u={ U ₁..., U _m, the total sample set of input data is: Q _y={ Y ₁..., Y _m; try to achieve average and the variance of inputoutput data collection respectively, reject undesirable sample point according to the data limit of setting, finally obtain total sample set Q; in data processing, the key of data prediction is the rejecting of unreasonable data and the normalized of data;

2. founding mathematical models.Assuming that process model can represent with following discrete equation:

y(t)+a ₁·y(t-1)+...+a _p·y(t-p)＝b ₁·u(t-1)+...+b _q·u(t-q)+v(t) (1)

Data in set of data samples Q are substituted into respectively the two ends of discrete equation, the proper method such as least square is adopted to obtain the approximate value of parameter in discrete equation and find the state space realization of discrete equation, in this, as the mathematical model of batch process, the approximation state spatial model that it obtains is:

$\left\{\begin{array}{c}x(t+1)=A\·x\left(t\right)+B\·u\left(t\right)\\ y\left(t\right)=C\·x\left(t\right)+d\left(t\right)\end{array}\right.---\left(2\right)$

Consider the repetitive nature of batch process, note k is a batch direction, then the state-space model of batch process can be expressed as:

$\left\{\begin{array}{c}x(t+1,k)=A\·x(t,k)+B\·u(t,k)\\ y(t,k)=C\·x(t,k)+d(t,k)\end{array}\right.---\left(3\right)$

Assuming that sampled point is N number of in one batch, the input and output track of individual batch of kth is made to be respectively U _k=[u _k(0), u _k(1) ..., u _k(N-1)] ^t, Y _k=[y _k(1), y _k(2) ..., y _k(N)] ^t, the discrete model that can obtain process is:

Y _k＝GU _k+d _k(4)

Wherein:

$G=\left[\begin{array}{cccc}{g}_{\mathrm{1,0}}& 0& ...& 0\\ g_{\mathrm{2,0}}& g_{\mathrm{2,1}}& ...& 0\\ .& .& & \\ .& .& ...& 0\\ .& .& & \\ g_{N,0}& g_{N,1}& ...& g_{N,N-1}\end{array}\right]\∈R^{N\×N},g_{i,j}=C\·A^{i-j}\·B---\left(5\right)$

Obtain the mathematical model of production run thus.

The step 4 of above-mentioned control method) in, adopt the controlled quentity controlled variable method of integrated forecasting Iterative Learning Control Algorithm calculating next batch as follows:

1. getting control law is:

$\begin{array}{c}{u}_k\left(t\right)=u_k^{\mathrm{ILC}}\left(t\right)+u_k^{\mathrm{MPC}}\left(t\right)\\ u_k^{\mathrm{ILC}}\left(t\right)=u_{k-1}\left(t\right)+K^{\mathrm{ILC}}\·e_{k-1}(t+1)\end{array}---\left(6\right)$

Wherein for iterative learning controlled quentity controlled variable, for PREDICTIVE CONTROL amount, integrated forecasting Iterative Learning Control Algorithm object is intended to find make the output of next batch of batch process closer to target trajectory;

2. according to the discrete model of system, the prediction of system exports and is:

${\hat{Y}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)=G_{\mathrm{pt}}\·U_k(t-1)+G_{\mathrm{mt}}\·U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1)---\left(7\right)$

Wherein:

$\begin{array}{c}{G}_{\mathrm{pt}}=\left[\begin{array}{ccccccc}{g}_{t+\mathrm{1,0}}& g_{t+\mathrm{1,1}}& ...& g_{t+1,t-1}& 0& ...& 0\\ g_{t+\mathrm{2,0}}& g_{t+\mathrm{2,1}}& ...& g_{t+2,t-1}& 0& ...& 0\\ .& .& .& .& .& .& .\\ g_{N,0}& g_{N,1}& ...& g_{N,t-1}& 0& ...& 0\end{array}\right]\∈R^{(N-t)\×N}\\ G_{\mathrm{mt}}=\left[\begin{array}{cccc}{g}_{t+1,t}& 0& ...& 0\\ g_{t+2,t}& g_{t+2,t+1}& ...& 0\\ .& .& .& .\\ g_{N,t}& g_{N,t+1}& ...& g_{N,N-1}\end{array}\right]\∈R^{(N-t)\×(N-t)}\end{array}---\left(8\right)$

Then tracking error can be estimated by following formula:

${\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)=Y_d\left({}_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\·U_k(t-1)-G_{\mathrm{mt}}\·(U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1)+U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1))---\left(9\right)$

3. introduce Model Predictive Control and consider following criterion function:

$\underset{U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)}{\min }{J}_k\left({}_{t+m}^t\right|t)={\left({\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)\right)}^{T}Q{\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)+{\left(U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)\right)}^T{\mathrm{RU}}_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)---\left(10\right)$

4. can obtain control law by above formula calculating analytic solution is:

$U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)={[G_{\mathrm{mt}}^{T}Q{G}_{\mathrm{mt}}+R]}^{-1}{G}_{\mathrm{mt}}Q{\hat{E}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)---\left(11\right)$

Wherein: ${\hat{E}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)=Y_d\left(F_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\·U_k(t-1)-G_{\mathrm{mt}}\·U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1),$ At each time point t, the Section 1 of analytic application solution is as current Introduced Malaria amount.

In each batch, along with the change of time t, the detailed algorithm flow process of control law is as follows:

1. according to the approximate model of system, iterative learning control law and controling parameters thereof is selected;

2. when starting for new batch, according to new batch service data adaptive updates system model, iterative learning controlled quentity controlled variable is calculated, and setting-up time t=1;

3. the moment t in each batch, real-time computational prediction controls correction;

If 4. t < N, makes t=t+1 and returns 3., otherwise making k=k+1, returning 2..

The integrated forecasting iterative learning control method based on model adaptation that the present invention proposes, owing to taking above technical scheme, has the following advantages:

1, the present invention utilizes the integrated forecasting Iterative Learning Control Algorithm based on model adaptation of proposition, can overcome batch process model error problem to a certain extent, accelerates convergence of algorithm speed, is of wide application.

2, this method can overcome the noise brought to production run by actual environment and manufacturing process.This method is skillfully constructed, simple and practical, and the high precision that can be widely used in industrial production line batch production process controls.

Embodiment

The self-adaptation consolidated forecast iterative learning control method that the present invention proposes, comprises the following steps:

1) combine with actual production process, arrange a batch of data acquisition of producing and store link, this link can utilize the equipment such as the existing industrial control computer of manufacturing enterprise, PLC;

2) according to production process data in the past in the production history database that collects, after carrying out data prediction, the simple mathematical model of suitable method establishment production run is adopted;

3) data acquisition link collects the inputoutput data of Product processing in industrial production line, and according to target following trajectory calculation tracking error curve;

4) according to step 3) tracking error that obtains, adopt integrated forecasting Iterative Learning Control Algorithm to calculate the real-time controlled quentity controlled variable of next batch;

5) when starting for new batch, the model of production run is upgraded adaptively according to new service data;

6) at each new sampled point, implementation step 4), realize the effective tracking to exporting target trajectory.

Described step 2) in, the mathematical model method setting up production run is as follows:

1. according to production process data in the past in historical data base, after carrying out data prediction, suitable method establishment process mathematical model is adopted.

Assuming that certain moment input amendment collection U=(u ₁, u ₂..., u _m) ^t∈ R ^m, represent that m monitoring sensor is in historical data sometime, m represents the number of monitoring sensor, R ^mrepresent m dimensional vector; u _jrepresent in sample U, the single sampled data values of a jth sensing data, j=1,2 ..., m; The output sample in this moment integrates as Y=(y ₁, y ₂..., y _n) ^t∈ R ⁿ, represent that n monitoring sensor is in historical data sometime, n represents the number of monitoring sensor, R ⁿrepresent n dimensional vector; y _jrepresent in sample Y, the single sampled data values of a jth sensing data, j=1,2 ..., n.Suppose to take N group historical data, the total sample set of input data obtained is as follows: Q _u={ U ₁..., U _m, the total sample set of input data is: Q _y={ Y ₁..., Y _m.Try to achieve average and the variance of inputoutput data collection respectively, reject undesirable sample point according to the data limit of setting.Finally obtain total sample set Q.In data processing, the key of data prediction is the rejecting of unreasonable data and the normalized of data.

2. founding mathematical models.Assuming that process model can represent with following discrete equation:

y(t)+a ₁·y(t-1)+...+a _p·y(t-p)＝b ₁·u(t-1)+...+b _q·u(t-q)+v(t) (1)

Data in set of data samples Q are substituted into respectively the two ends of discrete equation, the proper method such as least square is adopted to obtain the approximate value of parameter in discrete equation and find the state space realization of discrete equation, in this, as the mathematical model of batch process, the approximation state spatial model that it obtains is:

$\left\{\begin{array}{c}x(t+1)=A\&CenterDot;x\left(t\right)+B\&CenterDot;u\left(t\right)\\ y\left(t\right)=C\&CenterDot;x\left(t\right)+d\left(t\right)\end{array}\right.---\left(2\right)$

Consider the repetitive nature of batch process, note k is a batch direction, then the state-space model of batch process can be expressed as:

$\left\{\begin{array}{c}x(t+1,k)=A\&CenterDot;x(t,k)+B\&CenterDot;u(t,k)\\ y(t,k)=C\&CenterDot;x(t,k)+d(t,k)\end{array}\right.---\left(3\right)$

Assuming that sampled point is N number of in one batch, the input and output track of individual batch of kth is made to be respectively U _k=[u _k(0), u _k(1) ..., u _k(N-1)] ^t, Y _k=[y _k(1), y _k(2) ..., y _k(N)] ^t, the discrete model that can obtain process is:

Y _k＝GU _k+d _k(4)

Wherein:

$G=\left[\begin{array}{cccc}{g}_{\mathrm{1,0}}& 0& ...& 0\\ g_{\mathrm{2,0}}& g_{\mathrm{2,1}}& ...& 0\\ .& .& & \\ .& .& ...& 0\\ .& .& & \\ g_{N,0}& g_{N,1}& ...& g_{N,N-1}\end{array}\right]\&Element;R^{N\&times;N},g_{i,j}=C\&CenterDot;A^{i-j}\&CenterDot;B---\left(5\right)$

Obtain the mathematical model of production run thus.

Described step 4) in, adopt the controlled quentity controlled variable method of integrated forecasting Iterative Learning Control Algorithm calculating next batch as follows:

1. getting control law is:

$\begin{array}{c}{u}_k\left(t\right)=u_k^{\mathrm{ILC}}\left(t\right)+u_k^{\mathrm{MPC}}\left(t\right)\\ u_k^{\mathrm{ILC}}\left(t\right)=u_{k-1}\left(t\right)+K^{\mathrm{ILC}}\&CenterDot;e_{k-1}(t+1)\end{array}---\left(6\right)$

Wherein for iterative learning controlled quentity controlled variable, for PREDICTIVE CONTROL amount.Integrated forecasting Iterative Learning Control Algorithm object is intended to find make the output of next batch of batch process closer to target trajectory.

2. according to the discrete model of system, the prediction of system exports and is:

${\hat{Y}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)=G_{\mathrm{pt}}\&CenterDot;U_k(t-1)+G_{\mathrm{mt}}\&CenterDot;U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1)---\left(7\right)$

Wherein:

$\begin{array}{c}{G}_{\mathrm{pt}}=\left[\begin{array}{ccccccc}{g}_{t+\mathrm{1,0}}& g_{t+\mathrm{1,1}}& ...& g_{t+1,t-1}& 0& ...& 0\\ g_{t+\mathrm{2,0}}& g_{t+\mathrm{2,1}}& ...& g_{t+2,t-1}& 0& ...& 0\\ .& .& .& .& .& .& .\\ g_{N,0}& g_{N,1}& ...& g_{N,t-1}& 0& ...& 0\end{array}\right]\&Element;R^{(N-t)\&times;N}\\ G_{\mathrm{mt}}=\left[\begin{array}{cccc}{g}_{t+1,t}& 0& ...& 0\\ g_{t+2,t}& g_{t+2,t+1}& ...& 0\\ .& .& .& .\\ g_{N,t}& g_{N,t+1}& ...& g_{N,N-1}\end{array}\right]\&Element;R^{(N-t)\&times;(N-t)}\end{array}---\left(8\right)$

Then tracking error can be estimated by following formula:

${\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)=Y_d\left({}_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\&CenterDot;U_k(t-1)-G_{\mathrm{mt}}\&CenterDot;(U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1)+U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1))---\left(9\right)$

3. introduce Model Predictive Control and consider following criterion function:

$\underset{U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)}{\min }{J}_k\left({}_{t+m}^t\right|t)={\left({\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)\right)}^{T}Q{\hat{E}}_k\left({}_{t+m}^{t+1}\right|t)+{\left(U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)\right)}^T{\mathrm{RU}}_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)---\left(10\right)$

4. can obtain control law by above formula calculating analytic solution is:

$U_k^{\mathrm{MPC}}\left({}_{t+m-1}^t\right|t-1)={[G_{\mathrm{mt}}^{T}Q{G}_{\mathrm{mt}}+R]}^{-1}{G}_{\mathrm{mt}}Q{\hat{E}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)---\left(11\right)$

Wherein: ${\hat{E}}_k^{\mathrm{ILC}}\left({}_{t+m}^{t+1}\right|t)=Y_d\left(F_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\&CenterDot;U_k(t-1)-G_{\mathrm{mt}}\&CenterDot;U_k^{\mathrm{ILC}}\left({}_{t+m-1}^t\right|t-1),$ At each time point t, the Section 1 of analytic application solution is as current Introduced Malaria amount.

In each batch, along with the change of time t, the detailed algorithm flow process of control law is as follows:

1. according to the approximate model of system, iterative learning control law and controling parameters thereof is selected;

2. when starting for new batch, according to new batch service data adaptive updates system model, iterative learning controlled quentity controlled variable is calculated, and setting-up time t=1;

3. the moment t in each batch, real-time computational prediction controls correction;

If 4. t < N, makes t=t+1 and returns 3., otherwise making k=k+1, returning 2..

In step 4) in, key is the choosing of each controling parameters, control algolithm, and the adaptive method of employing is revised model before batch starting.

Claims (4)

1., based on an integrated forecasting iterative learning control method for model adaptation, it is characterized in that the method comprises the following steps:

1) combine with actual production process, arrange a batch of data acquisition of producing and store link, this link can utilize the equipment such as the existing industrial control computer of manufacturing enterprise, PLC;

2) according to production process data in the past in the production history database that collects, after carrying out data prediction, the simple mathematical model of suitable method establishment production run is adopted;

3) data acquisition link collects the inputoutput data of Product processing in industrial production line, and according to target following trajectory calculation tracking error curve;

4) according to step 3) tracking error that obtains, adopt integrated forecasting Iterative Learning Control Algorithm to calculate the real-time controlled quentity controlled variable of next batch;

5) when starting for new batch, the model of production run is upgraded adaptively according to new service data;

6) at each new sampled point, implementation step 4), realize the effective tracking to exporting target trajectory.

2. a kind of integrated forecasting iterative learning control method based on model adaptation as claimed in claim 1, is characterized in that: described step 2) in, the mathematical model method setting up production run is as follows:

1. according to production process data in the past in historical data base, after carrying out data prediction, suitable method establishment process mathematical model is adopted:

Assuming that certain moment input amendment collection U=(u ₁, u ₂..., u _m) ^t∈ R ^m, represent that m monitoring sensor is in historical data sometime, m represents the number of monitoring sensor, R ^mrepresent m dimensional vector; u _jrepresent in sample U, the single sampled data values of a jth sensing data, j=1,2 ..., m; The output sample in this moment integrates as Y=(y ₁, y ₂..., y _n) ^t∈ R ⁿ, represent that n monitoring sensor is in historical data sometime, n represents the number of monitoring sensor, R ⁿrepresent n dimensional vector; y _jrepresent in sample Y, the single sampled data values of a jth sensing data, j=1,2 ..., n, supposes to take N group historical data, and the total sample set of input data obtained is as follows: Q _u={ U ₁..., U _m, the total sample set of input data is: Q _y={ Y ₁..., Y _m, try to achieve average and the variance of inputoutput data collection respectively, reject undesirable sample point according to the data limit of setting.Finally obtain total sample set Q, in data processing, the key of data prediction is the rejecting of unreasonable data and the normalized of data;

2. founding mathematical models.Assuming that process model can represent with following discrete equation:

y(t)+a ₁·y(t-1)+...+a _p·y(t-p)＝b ₁·u(t-1)+...+b _q·u(t-q)+v(t) (1)

Data in set of data samples Q are substituted into respectively the two ends of discrete equation, the proper method such as least square is adopted to obtain the approximate value of parameter in discrete equation and find the state space realization of discrete equation, in this, as the mathematical model of batch process, the approximation state spatial model that it obtains is:

$\left\{\begin{array}{c}x(t+1)=A\&CenterDot;x\left(t\right)+B\&CenterDot;u\left(t\right)\\ y\left(t\right)=C\&CenterDot;x\left(t\right)+d\left(t\right)\end{array}\right.---\left(2\right)$

Consider the repetitive nature of batch process, note k is a batch direction, then the state-space model of batch process can be expressed as:

$\left\{\begin{array}{c}x(t+1,k)=A\&CenterDot;x(t,k)+B\&CenterDot;u(t,k)\\ y(t,k)=C\&CenterDot;x(t,k)+d(t,k)\end{array}\right.---\left(3\right)$

Assuming that sampled point is N number of in one batch, the input and output track of individual batch of kth is made to be respectively U _k=[u _k(0), u _k(1) ..., u _k(N-1)] ^t, Y _k=[y _k(1), y _k(2) ..., y _k(N)] ^t, the discrete model that can obtain process is:

Y _k＝GU _k+d _k(4)

Wherein:

$G=\left[\begin{array}{cccc}{g}_{\mathrm{1,0}}& 0& ...& 0\\ g_{\mathrm{2,0}}& g_{\mathrm{2,1}}& ...& 0\\ .& .& & \\ .& .& ...& 0\\ .& .& & \\ g_{N,0}& g_{N,1}& ...& g_{N,N-1}\end{array}\right]\&Element;R^{N\&times;N},g_{i,j}=C\&CenterDot;A^{i-j}\&CenterDot;B---\left(5\right)$

Obtain the mathematical model of production run thus.

3., as claimed in claim 1 based on the integrated forecasting iterative learning control method of model adaptation, it is characterized in that: described step 4) in, adopt the controlled quentity controlled variable method of integrated forecasting Iterative Learning Control Algorithm calculating next batch as follows:

1. getting control law is:

$u_k\left(t\right)=u_k^{\mathrm{ILC}}\left(t\right)+u_k^{\mathrm{MPC}}\left(t\right)$

(6)

$u_k^{\mathrm{ILC}}\left(t\right)=u_{k-1}\left(t\right)+K^{\mathrm{ILC}}\&CenterDot;e_{k-1}(t+1)$

Wherein for iterative learning controlled quentity controlled variable, for PREDICTIVE CONTROL amount, integrated forecasting Iterative Learning Control Algorithm object is intended to find make the output of next batch of batch process closer to target trajectory;

2. according to the discrete model of system, the prediction of system exports and is:

${\hat{Y}}_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m}^{t+1}\right|t)=G_{\mathrm{pt}}\&CenterDot;U_k(t-1)+G_{\mathrm{mt}}\&CenterDot;U_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)---\left(7\right)$

Wherein:

$G_{\mathrm{pt}}=\left[\begin{array}{ccccccc}{g}_{t+\mathrm{1,0}}& g_{t+\mathrm{1,1}}& ...& g_{t+1,t-1}& 0& ...& 0\\ g_{t+\mathrm{2,0}}& g_{t+\mathrm{2,1}}& ...& g_{t+2,t-1}& 0& ...& 0\\ .& .& .& .& .& .& .\\ g_{N,0}& g_{N,1}& ...& g_{N,t-1}& 0& ...& 0\end{array}\right]\&Element;R^{(N-t)\&times;N}$

(8)

$G_{\mathrm{mt}}=\left[\begin{array}{cccc}{g}_{t+1,t}& 0& ...& 0\\ g_{t+2,t}& g_{t+2,t+1}& ...& 0\\ .& .& .& .\\ g_{N,t}& g_{N,t+1}& ...& g_{N,N-1}\end{array}\right]\&Element;R^{(N-t)\&times;(N-t)}$

Then tracking error can be estimated by following formula:

${\hat{E}}_k\left({\mathrm{}}_{t+m}^{t+1}\right|t)=Y_d\left({\mathrm{}}_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\&CenterDot;U_k(t-1)-G_{\mathrm{mt}}\&CenterDot;(U_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)+U_k^{\mathrm{MPC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1))---\left(9\right)$

3. introduce Model Predictive Control and consider following criterion function:

$\underset{U_k^{\mathrm{MPC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)}{\min }{J}_k\left({\mathrm{}}_{t+m}^t\right|t)={\left({\hat{E}}_k\left({\mathrm{}}_{t+m}^{t+1}\right|t)\right)}^T{\hat{E}}_k\left({\mathrm{}}_{t+m}^{t+1}\right|t)+{\left(U_k^{\mathrm{MPC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)\right)}^T{\mathrm{RU}}_k^{\mathrm{MPC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)---\left(10\right)$

4. can obtain control law by above formula calculating analytic solution is:

$U_k^{\mathrm{MPC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1)={[G_{\mathrm{mt}}^T{\mathrm{QG}}_{\mathrm{mt}}+R]}^{-1}{G}_{\mathrm{mt}}Q{\hat{E}}_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m}^{t+1}\right|t)---\left(11\right)$

Wherein: ${\hat{E}}_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m}^{t+1}\right|t)=Y_d\left({\mathrm{}}_{t+m}^{t+1}\right|t)-G_{\mathrm{pt}}\&CenterDot;U_k(t-1)-G_{\mathrm{mt}}\&CenterDot;U_k^{\mathrm{ILC}}\left({\mathrm{}}_{t+m-1}^t\right|t-1),$ At each time point t, the Section 1 of analytic application solution is as current Introduced Malaria amount.

4., as claimed in claim 3 based on the integrated forecasting iterative learning control method of model adaptation, it is characterized in that: in each batch, along with the change of time t, the detailed algorithm flow process of control law is as follows:

1. according to the approximate model of system, iterative learning control law and controling parameters thereof is selected;

2. when starting for new batch, according to new batch service data adaptive updates system model, iterative learning controlled quentity controlled variable is calculated, and setting-up time t=1;

3. the moment t in each batch, real-time computational prediction controls correction;

If 4. t < N, makes t=t+1 and returns 3., otherwise making k=k+1, returning 2..