CN101329553A - Gasifying stove forecasting type PID control method - Google Patents

Abstract

The invention relates to a control method of predicative proportion integration differentiation of a gasifier; aiming at the characteristics of nonlinearity, long time delay and strong coupling of the gasifier in the controlled process, the method uses the advantages of robustness and easy realization of the conventional PID control algorithm, combines the effectiveness of the predicative control method to the object with long time delay, uses the predicative control idea into the conventional PID control algorithm, uses a variable predictor to predict the input of the controlled object of the gasifier after L time, leads the reference input and the output of the variable predictor to subtract, takes the obtained results as the deviation of the entrance of a PID regulator, carries out a plurality of times of calculation, leads output of the obtained controller by two times of calculation to be constant basically, outputs the control function, realizes the advanced action of the regulator and still obtains good control quality when the working condition of the gasifier occurs larger range change. The predicative PID algorithm of the gasifier is simple and is easy to be implemented on the software/hardware platform of the existing business control system.

Description

The forecasting type PID control method of gasification furnace

Technical field

The present invention is a kind of forecasting type PID control method at gasification furnace, realizes effective control to gasified boiler system belonging to thermal chemical reaction engineering, motility engineering and automation field in conjunction with conventional PID controller and forecast Control Algorithm.

Background technology

Coal gasification is a kind of important clean energy resource mode of production, and gasification furnace then is the core component of coal gasification course.Because relate to the chemical reaction process of many complexity, gasification furnace is multivariate, large time delay, the non-linear and high coupled system of a complexity, and all very sensitive to the influence of various disturbances, its control corresponding is theoretical and use not overripened.High-quality control method at gasification furnace is the assurance of gasification furnace safe and reliable operation, also is present gasification furnace problem demanding prompt solution in large-scale application and popularization process.

Existing gasification furnace control method mainly is divided into two kinds: (1) based on the conventional control method of classical control theory, as PID (proportion integration differentiation) control method of routine etc.; (2) based on the Advanced Control method of modern control theory, as Model Predictive Control (MPC) etc.These two kinds of methods all exist not enough in actual applications.For example, conventional PID control method, though implement easily, but because its algorithm just calculates according to the setting value of current time and preceding two sampling instants and the deviation between the output, for the such large time delay object of gasification furnace, control action is action in advance not, can't obtain gratifying control effect; And common Model Predictive Control (MPC), because the complicacy of algorithm, can occupy a large amount of controller resources, especially at the such height coupling multivariable process of gasification furnace, such control method is difficult to implement on the hardware and software platform of existing general-purpose control system.If can dope the output of following L sampling instant at current time

And the PID controller can be according to the control deviation in future $e\left(k\right)=R-\tilde{y}(k+L)$ Calculate, then the PID controller can shift to an earlier date L sampling period change control action, and this gasification furnace process to large time delay is vital.Based on such thought, the present invention proposes forecasting type PID controller (LP-PID) as shown in Figure 1.

Summary of the invention

Technical matters: the purpose of this invention is to provide a kind of gasification furnace forecasting type proportion integration differentiation (PID) control method, traditional PID control is combined with PREDICTIVE CONTROL, the following output of applied forcasting model prediction system, make the PID controller carry out computing to gasification furnace according to the control deviation in the moment in future, this invention is to be used to solve gasified boiler system large time delay object, control action is action in advance not, common control method is difficult to implement on the hardware and software platform of existing general-purpose control system, i.e. the gasification furnace controlled device method that is difficult to be effectively controlled.

Technical scheme: the present invention discloses a kind of forecasting type proportion integration differentiation (PID) control method of gasification furnace, and the concrete implementation step of this method is as follows:

Step 1: in the control loop of gasified boiler system, controller utilizes reference input to deduct the output quantity of variable fallout predictor according to the reference input of the output quantity and the control system of variable fallout predictor, and initial value is set to $\tilde{y}(k+L)=y\left(k\right),$ The inlet deviation of controlled device, controller is according to inlet deviation calculation output controlled quentity controlled variable u (k), and wherein k is current sampling instant; Y (k) is the output of current sampling instant controlled process;

Step 2: utilize gasification furnace controlled process mathematical model, after obtaining the object pure delay, determine prediction step L,,, obtain the prediction output valve of L sampling instant according to predictive control theory prediction controlled device according to the output controlled quentity controlled variable u (k) of step 1)

Step 3: with the output valve of step 2

As the output quantity of variable fallout predictor in the step 1, double counting output controlled quentity controlled variable u (k);

Step 4:, finish forecasting type proportion integration differentiation control to gasification furnace with the control input of final output controlled quentity controlled variable u (k) as the gasification furnace control loop.

Described output controlled quentity controlled variable u (k) is:

u(k)＝u(k-1)+K _p(1+T _d/T _s+T _s/T _i)e(k)

-K _p(1+2T _d/T _s)e(k-1)+K _pT _d/T _se(k-2)

In the formula, K _p, T _i, T _dBe respectively proportional gain, integral time and the derivative time of proportional integral derivative controller; U (k), u (k-1) are the proportion integration differentiation output of current and last sampling instant, and e (k), e (k-1), e (k-2) are respectively the proportion integration differentiation inlet deviation of current sampling instant, last sampling instant, preceding two sampling instants; T _sBe sampling time interval.

The mathematical model of the controlled process in the step 2 is the CARIMA model:

A(q ^-1)y(k)＝B(q ^-1)u(k-1)+ξ(k)/Δ

In the formula, it is zero white noise sequence in k output, input and average constantly that y (k), u (k), ξ (k) are respectively controlled process; Δ=1-q ^-1A (q ^-1), B (q ^-1) be following backward shift operator q ^-1Polynomial expression:

$\left.\begin{array}{c}A\left(q^{-1}\right)=1+a_1{q}^{-1}+\·\·\·+a_{\mathrm{na}}{q}^{-\mathrm{na}}\\ B\left(q^{-1}\right)=b_0+b_1{q}^{-1}+\·\·\·+b_{\mathrm{nb}}{q}^{-\mathrm{nb}}\end{array}\right\}$

a _iBe the polynomial coefficient of A; b _iBe the polynomial coefficient of B; n _aBe the polynomial exponent number of A; n _bBe the polynomial exponent number of B.

Above-mentioned steps 1) inlet deviation is obtained by the output quantity with reference to input and variable fallout predictor, and promptly following system output signal constantly subtracts each other acquisition.Step 3) double counting output controlled quentity controlled variable u (k) number of times is no less than secondary.

The computation process of forecasting type PID controller (LP-PID) is:

The output initial value that 1. the variable fallout predictor is set is: $\tilde{y}(k+L)=y\left(k\right).$

2. the output controlled quentity controlled variable of calculating the PID controller is: u (k).

The inlet deviation of PID controller is: $e\left(k\right)=R-\tilde{y}(k+L).$

The output controlled quentity controlled variable of PID controller is:

u(k)＝u(k-1)+K _p(1+T _d/T _s+T _s/T _i)e(k)

-K _p(1+2T _d/T _s)e(k-1)+K _pT _d/T _se(k-2) (1)

In the formula, K _p, T _i, T _dBe respectively proportional gain, integral time and the derivative time of PID regulator; U (k), u (k-1) are the PID output of current and last sampling instant, and e (k), e (k-1), e (k-2) are respectively the PID inlet deviation of current sampling instant, last sampling instant, preceding two sampling instants; T _sBe sampling time interval.

3. according to the control action u (k) of controlled process mathematical model and previous step calculating, forecasting process is in the output valve of following L sampling instant

The mathematical model of controlled process is following CARIMA (Controlled Auto-Regressive IntegratedMoving Average, a controlled autoregression integration running mean) model:

A(q ^-1)y(k)＝B(q ^-1)u(k-1)+ξ(k)/Δ (2)

In the formula, it is zero white noise sequence in k output, input and average constantly that y (k), u (k), ξ (k) are respectively controlled process; Δ=1-q ^-1A (q ^-1), B (q ^-1) be following backward shift operator q ^-1Polynomial expression.

$\left.\begin{array}{c}A\left(q^{-1}\right)=1+a_1{q}^{-1}+\·\·\·+a_{\mathrm{na}}{q}^{-\mathrm{na}}\\ B\left(q^{-1}\right)=b_0+b_1{q}^{-1}+\·\·\·+b_{\mathrm{nb}}{q}^{-\mathrm{nb}}\end{array}\right\}---\left(3\right)$

Process output in the predicted value of following L sampling instant is:

$\tilde{y}(k+L)=G_L\left(q^{-1}\right)\mathrm{\Δu}(k+L-1)+F_L\left(q^{-1}\right)y\left(k\right)---\left(4\right)$

In the formula:

G _L(q ^-1)＝E _L(q ^-1)B(q ^-1)＝g ₀+g ₁q ^-1+…+g _nb+L-1q ^-(nb+L-1)(5)

Polynomial expression E (q ^-1), F (q ^-1) can pass through the acquisition of following Diophantine (diophantus, Ancient Greek Mathematics man) equation:

1＝E _L(q ^-1)AΔ+q ^-LF _L(q ^-1) (6)

In the formula,

$\left.\begin{array}{c}{E}_L\left(q^{-1}\right)=1+e_1{q}^{-1}+\·\·\·+e_{L-1}{q}^{-(L-1)}\\ F_L\left(q^{-1}\right)=f_0+f_1{q}^{-1}+\·\·\·+f_{\mathrm{na}}{q}^{-\mathrm{na}}\end{array}\right\}---\left(7\right)$

For the controlled process that does not have unstable limit, the number of times of desirable following control increment is 1, promptly has:

$\left.\begin{array}{c}\mathrm{\Δu}\left(k\right)=u\left(k\right)-u(k-1)\\ \mathrm{\Δu}(k+i)=u(k+i)-u(k+i-1)=0\end{array}\right\}---\left(8\right)$

(i＝1，2，…，L-1)

Formula (8), (5) substitution formula (4) can directly be tried to achieve output in the predicted value of following L sampling instant is:

In the formula,

Δu(k)＝u(k)-u(k-1)，u(k)

Be the 2. step PID output of being calculated.

4. according to the top prediction of output of calculating

Turn back to the 2. the step carry out double counting, till the PID output that continuous quadratic calculates is constant substantially, generally need double counting 3 to 5 times.

5. final control signal u (k) is acted on real process, guarantee that forecasting type PID controller can shift to an earlier date L sampling period action.

Beneficial effect: the present invention is by the advantage in conjunction with existing gasification furnace control method, forecasting type PID control method is proposed, on traditional pid algorithm basis, be applied to the control of gasification furnace controlled process, control system is moved in advance, the PID controller can carry out computing according to control deviation constantly in future, can overcome gasification furnace large time delay characteristic effectively, significantly improve the control system quality, and bottom adopts conventional pid control algorithm, can implement at existing general-purpose control system hardware and software platform easily, simple, regulating effect is remarkable.

Description of drawings

Fig. 1 is a forecasting type PID controller architecture block diagram of the present invention.

Fig. 2 is the gasified boiler system embodiment block diagram that the present invention adopts forecasting type PID control method.

Embodiment

Concrete implementation step of the present invention is as follows:

Step 1) is in the control loop of gasified boiler system, controller is according to the reference input of the output quantity and the control system of variable fallout predictor, utilize reference input to deduct the output quantity of variable fallout predictor, the inlet deviation of controlled device, controller is according to inlet deviation calculation output controlled quentity controlled variable u (k), and wherein k is current sampling instant;

Step 2) utilizes gasification furnace controlled process mathematical model, after obtaining the object pure delay, determine prediction step L,,, obtain the output valve of L sampling instant according to predictive control theory prediction controlled device according to the output controlled quentity controlled variable u (k) of step 1)

Step 3) is with step 2) output valve As the output quantity of variable fallout predictor in the step 1), double counting output controlled quentity controlled variable u (k);

Step 4) is finished the forecasting type PID control to gasification furnace with the control input of final output controlled quentity controlled variable u (k) as the gasification furnace control loop.

According to forecasting type PID controller architecture block diagram shown in Figure 1, specific implementation method is: 1. according to gasification furnace design and running parameter, in conjunction with the site test curve, obtain gasification furnace controlled process mathematical model; 2. the programming of forecasting type pid control algorithm is provided according to the step that is provided in the technical scheme; 3. with program and pack in the control system storer; 4. carry out the off-line and the on-line debugging of forecasting type PID control method, and this control method puts into operation the most at last, guarantee that the gasification furnace controlled process can move with security and stability.

Fig. 2 is for adopting certain gasified boiler system synoptic diagram of forecasting type PID control method.The main technique flow process of this gasified boiler system is: coal in gasification furnace (or other carbon containing energy), water vapour, the air oxygen of air separation plant (perhaps from) under certain temperature and pressure condition through the chemical reaction process of a series of complexity, low-calorie coal gas output in being converted to, coal gas is through after purifying, and sends into the gas turbine combustion acting or reclaims as synthesis material.According to process characteristic, the manipulated variable of this gasified boiler system is chosen for: gasification furnace bed drain purge, intake air mass rate, coal supply mass rate, inlet water steam mass flow and inlet lime stone mass rate (WLS); Four controlled variables need regulating are respectively: gas pressure and gas temperature in gasification furnace outlet calorific value of gas, gasification siege material gross mass, the gasification furnace.Because going into the lime stone amount and the coal dust amount of stove is proportional (1: 10), is not an independently manipulated variable, therefore whole gasification furnace controlled process comes down to one 4 * 4 multivariate object.For gasified boiler system, main external disturbance comes from the variation of gas turbine operating mode, therefore, with the inlet pressure (PSINK) of gas turbine as a main external disturbance amount.

The forecasting type PID control method of the gasification furnace that the present invention proposes is carried out decentralised control to multivariable gasified boiler system.By RGA (relative gain matrix) analyze, effectively relative gain matrix (ERGA) is analyzed or svd methods such as (SVD) is determined pair relationhip between manipulated variable-controlled variable.The calculating of relative gain matrix can be carried out according to following short-cut method:

Λ＝K□(K ^-1) ^T

Wherein K represents system's open-loop gain matrix, the multiplying each other of representing matrix corresponding element.

For example, to this gasification furnace under a certain operating mode, to system four inputs do respectively 10% the instruction step increase test, by the system responses tracing analysis, can obtain system's open-loop gain matrix K

$K=\left[\begin{array}{cccc}14328& -46238& 22239& -54097\\ -5743.9& -1388.4& 4692& -692.55\\ -671.8& 13664& 1406.3& 7562.9\\ -2.283& 20.695& -18.503& -44.876\end{array}\right]$

Utilize following formula, the RGA that can calculate system under a certain operating mode is as shown in table 1 below:

Relative gain matrix analysis under a certain operating mode of table 1

According to the result of relative gain matrix analysis, can obtain the relation of the collocation between the manipulated variable and controlled variable in the scheme one, that is:

(1) the coal gas enthalpy is regulated by coal-supplying amount, and the change of coal amount influences the composition of product combustible, thereby influences the enthalpy of coal gas;

(2) the bed material is regulated by bed drain purge, and the variable effect of bed drain purge is to total material balance;

(3) gas pressure is regulated by air capacity, and the variation of air capacity can change gas pressure fast;

(4) gas temperature is regulated by quantity of steam, and the variation of quantity of steam can change the temperature of gasified boiler system fast.

The PID controller of the gasification furnace forecasting type PID control method bottom that the present invention proposes adopts increment type PID algorithm, and rudimentary algorithm is:

u(k)＝u(k-1)+K _p(1+T _d/T _s+T _s/T _i)e(k)

-K _p(1+2T _d/T _s)e(k-1)+K _pT _d/T _se(k-2)

U (k) wherein, u (k-1) they are current and last one constantly PID output, Kp, and Ti, Td are respectively proportional gain, integral time and derivative time; E (k), e (k-1), e (k-2) were respectively the PID inlet deviation in current time, a last moment and preceding two moment.This PID also has functions such as input dead band, feedforward, hand automatic switchover, tracking, amplitude limitation, rate limit and anti-integration are saturated.

With gas temperature in the gasification furnace-this dynamic link of inlet water vapor quality flow is example, at first by the method for System Discrimination, sets up the transfer function model of this dynamic link

$G_{T-\mathrm{STM}}\left(s\right)=\frac{-44.88(245s+1)}{(168s+1){(24s+1)}^3}$

Choosing the sampling time is 1s, and this system is carried out discretize, the dynamic link model that obtains dispersing

y(k)-3.8716y(k-1)+5.6206y(k-2)-3.6262y(k-3)+0.87726y(k-4)＝-0.00076u(k-1)-0.0022u(k-2)+0.0022u(k-3)+0.000715u(k-4)+e(k)/Δ

Just obtained the variable fallout predictor of this dynamic link thus, can be used to predict that this link is in output constantly in future.By choosing suitable prediction step, just can carry out forecasting type PID control to this loop.Rudimentary algorithm is:

The output initial value that 1. the variable fallout predictor is set is: $\tilde{y}(k+L)=y\left(k\right).$

2. the output controlled quentity controlled variable of calculating the PID controller is: u (k).

The inlet deviation of PID controller is: $e\left(k\right)=R-\tilde{y}(k+L);$

The output controlled quentity controlled variable of PID controller is:

u(k)＝u(k-1)+K _p(1+T _d/T _s+T _s/T _i)e(k)

-K _p(1+2T _d/T _s)e(k-1)+K _pT _d/T _se(k-2)

In the formula, K _p, T _i, T _dBe respectively proportional gain, integral time and the derivative time of PID regulator; U (k), u (k-1) are the PID output of current and last sampling instant, and e (k), e (k-1), e (k-2) are respectively the PID inlet deviation of current sampling instant, last sampling instant, preceding two sampling instants; T _sBe sampling time interval.

3. according to the control action u (k) of controlled process mathematical model and previous step calculating, forecasting process is in the output valve of following L sampling instant

The mathematical model of controlled process is that following CARIMA model is top calculating gained.

A(q ^-1)y(k)＝B(q ^-1)u(k-1)+ξ(k)/Δ

In the formula, it is zero white noise sequence in k output, input and average constantly that y (k), u (k), ξ (k) are respectively controlled process; Δ=1-q ^-1A (q ^-1), B (q ^-1) be following backward shift operator q ^-1Polynomial expression.

$\left.\begin{array}{c}A\left(q^{-1}\right)=1+a_1{q}^{-1}+\·\·\·+a_{\mathrm{na}}{q}^{-\mathrm{na}}\\ B\left(q^{-1}\right)=b_0+b_1{q}^{-1}+\·\·\·+b_{\mathrm{nb}}{q}^{-\mathrm{nb}}\end{array}\right\}$

Process output in the predicted value of following L sampling instant is:

Process output in the predicted value of following L sampling instant is:

$\tilde{y}(k+L)=G_L\left(q^{-1}\right)\mathrm{\Δu}(k+L-1)+F_L\left(q^{-1}\right)y\left(k\right)$

In the formula:

G _L(q ^-1)＝E _L(q ^-1)B(q ^-1)＝g ₀+g ₁q ^-1+…+g _nb+L-1q ^-(nb+L-1)

Polynomial expression E (q ^-1), F (q ^-1) can obtain by following Diophantine equation:

1＝E _L(q ^-1)AΔ+q ^-LF _L(q ^-1)

In the formula,

$\left.\begin{array}{c}{E}_L\left(q^{-1}\right)=1+e_1{q}^{-1}+\·\·\·+e_{L-1}{q}^{-(L-1)}\\ F_L\left(q^{-1}\right)=f_0+f_1{q}^{-1}+\·\·\·+f_{\mathrm{na}}{q}^{-\mathrm{na}}\end{array}\right\}$

For the controlled process that does not have unstable limit, the number of times of desirable following control increment is 1, promptly has:

$\left.\begin{array}{c}\mathrm{\Δu}\left(k\right)=u\left(k\right)-u(k-1)\\ \mathrm{\Δu}(k+i)=u(k+i)-u(k+i-1)=0\end{array}\right\}$

(i＝1，2，…，L-1)

According to top formula, can directly try to achieve output and be in the predicted value of following L sampling instant:

In the formula,

Δu(k)＝u(k)-u(k-1)，u(k)

Be the 2. step PID output of being calculated.

4. according to the top prediction of output of calculating

Turn back to the 2. the step carry out double counting, till the PID output of calculating for continuous four times is constant substantially.

5. final control signal u (k) is acted on real process, guarantee that forecasting type PID controller can shift to an earlier date L sampling period action.

Just finish thus gas temperature in the gasification furnace-this control loop of inlet water vapor quality flow has been carried out forecasting type PID control.

The rest may be inferred, and all the other control loops of gasification furnace are implemented forecasting type pid control algorithm proposed by the invention.

Claims (3)

1, a kind of forecasting type PID control method of gasification furnace is characterized in that this method comprises the steps:

Step 1: in the control loop of gasified boiler system, controller utilizes reference input to deduct the output quantity of variable fallout predictor according to the reference input of the output quantity and the control system of variable fallout predictor, and initial value is set to $\tilde{y}(k+L)=y\left(k\right),$ The inlet deviation of controlled device, controller is according to inlet deviation calculation output controlled quentity controlled variable u (k), and wherein k is current sampling instant; Y (k) is the output of current sampling instant controlled process;

Step 2: utilize gasification furnace controlled process mathematical model, after obtaining the object pure delay, determine prediction step L,,, obtain the prediction output valve of L sampling instant according to predictive control theory prediction controlled device according to the output controlled quentity controlled variable u (k) of step 1

Step 3: with the output valve of step 2 As the output quantity of variable fallout predictor in the step 1, double counting output controlled quentity controlled variable u (k);

Step 4:, finish forecasting type proportion integration differentiation control to gasification furnace with the control input of final output controlled quentity controlled variable u (k) as the gasification furnace control loop.

2, the forecasting type PID control method of gasification furnace as claimed in claim 1 is characterized in that described output controlled quentity controlled variable u (k) is:

u(k)＝u(k-1)+K _p(1+T _d/T _s+T _s/T _i)e(k)

-K _p(1+2T _d/T _s)e(k-1)+K _pT _d/T _se(k-2)

In the formula, K _p, T _i, T _dBe respectively proportional gain, integral time and the derivative time of proportional integral derivative controller; U (k), u (k-1) are the proportion integration differentiation output of current and last sampling instant, and e (k), e (k-1), e (k-2) are respectively the proportion integration differentiation inlet deviation of current sampling instant, last sampling instant, preceding two sampling instants; T _sBe sampling time interval.

3, the forecasting type PID control method of gasification furnace as claimed in claim 1, the mathematical model that it is characterized in that the controlled process in the step 2 is the CARIMA model:

A(q ^-1)y(k)＝B(q ^-1)u(k-1)+ξ(k)/Δ

In the formula, it is zero white noise sequence in k output, input and average constantly that y (k), u (k), ξ (k) are respectively controlled process; Δ=1-q ^-1A (q ^-1), B (q ^-1) be following backward shift operator q ^-1Polynomial expression:

$\left.\begin{array}{c}A\left(q^{-1}\right)=1+a_1{q}^{-1}+\·\·\·+a_{\mathrm{na}}{q}^{-\mathrm{na}}\\ B\left(q^{-1}\right)=b_0+b_1{q}^{-1}+\·\·\·+b_{\mathrm{nb}}{q}^{-\mathrm{nb}}\end{array}\right\}$

a _iBe the polynomial coefficient of A; b _iBe the polynomial coefficient of B; n _aBe the polynomial exponent number of A; n _bBe the polynomial exponent number of B.

Images