CN102319612B - Method for intelligently controlling pressure difference of cement raw meal vertical mill - Google Patents

Abstract

The invention discloses a method for intelligently controlling the pressure difference of a cement raw meal vertical mill. In the method, internal pressure difference of the mill is taken as a major controlled variable, external circulating current is taken as an auxiliary controlled variable, a feeding amount is taken as a major control variable, and a water spray amount is taken as an auxiliary control variable. The method comprises the following steps of: judging a current production condition according to the internal pressure difference of the mill and external circulating current information with a production rule method; and under the normal working condition, controlling the internal pressure difference of the mill by adopting GPC-PID (Generalized Predictive Control-Proportion Integration Differentiation). In the method, a mathematic model for describing the change of the pressure difference along with the feeding amount can be established only by using input and output data of a process without analyzing the complex material milling process of the raw meal vertical mill, the identifying process is simple, and online identification is available; the method has certain adaptability to the change of the production condition; and the advantages of multistep prediction, rolling optimization and feedback correction of a GPC algorithm as well as the advantages of easiness and high robustness of a PID algorithm are fully utilized, so that the labor intensity of field operating personnel is lowered greatly.

Description

A kind of intelligence control method of cement slurry Vertical Mill pressure reduction

Technical field

The present invention relates to the cement production process control field, relate in particular to a kind of intelligence control method of cement slurry Vertical Mill pressure reduction.

Background technology

Roll mill roller grinding machine (abbreviation Vertical Mill) is the crucial raw grinding equipment in System during New Style Dry-Process Production, is bearing the effect of grinding and material drying.Therefore, effectively control the interior pressure reduction of vertical mill and go out to grind gas temperature, most important to the stable operation of whole piece raw material production line.

Cement slurry grinding process flow process as shown in Figure 1.Lime stone, iron powder, sandstone and clay are delivered in mill by belt conveyor after the proportioning station batching.Material is pulverized grinding by the relative motion between mill and grinding roller.The thermal current of being sent here by kiln tail (or hot-blast stove) is blown into from grinding machine spout ring, acts on material, and the material after pulverizing is blown afloat by thermal current, and material is dried simultaneously; Bulky grain falls back and continues grinding on mill, the especially big particle that can not be blown afloat by the vane hot blast is outside slag-drip opening is discharged mill (outer circulation), rest materials is brought into the separator that grinds top and carries out the thickness separation, satisfying the particle of product requirement discharges outside mill with air-flow, underproof particle returns to mill and continues grinding (interior circulation), until qualified.

In grinding, what of interior inventory the big or small directly reaction of pressure reduction grind, and when the pressure reduction rising, the interior bed of material thickening of mill causes grinding machine to vibrate; When pressure reduction descends, grind interior bed of material attenuation, mill produces linear the contact with grinding roller, also can cause the grinding machine vibration.Good for guaranteeing raw material stoving, go out to grind the temperature general control at 80 ~ 90 ℃.If it is too high to go out to grind gas temperature, material is dried rapidly, makes the bed of material unstable, can cause the grinding machine vibration, grinding machine stop jumping when going out to grind gas temperature and reach 120 ℃; If it is too low to go out to grind gas temperature, illustrate that oven dry is not enough, the moisture content of finished product is large, and the mill efficiency of system reduces, and has a strong impact on output.Going out to grind gas temperature mainly controls by hot blast rate and cold blast rate that adjusting enters to grind, set up out the Mathematical Modeling of mill gas temperature and hot and cold air quantity by equation of heat balance, then utilize Dynamic array control algorithm (DMC) to control, control effect and can satisfy technological requirement fully; And in mill, the control of pressure reduction is the control difficult point of vertical mill system.Material process of grinding in mill is the process of non-linear a, large time delay, many disturbances.Measure through field trial, the ratio of the lag time of process and time constant is greater than 3, and in addition, pressure reduction also is subject to the interference of the factors such as material characteristic, systematic air flow, and this has brought very large difficulty for control of pressure reduction in mill.

At present, most domestic manufacture of cement factory adopts the manual adjustments feeding capacity to control pressure reduction in mill, and indivedual producers have adopted PID control.Studies show that, greater than 3 the time, PID controls and just is difficult to the control effect that obtains when the ratio of lag time of process and time constant.Manual adjustments has not only increased operator's labour intensity, and the control effect of pressure reduction is also steady not, fluctuation often occurs, is difficult to reach technological requirement.Therefore, be necessary to alleviate operator's work load from a kind of new control method of Vertical Mill object own characteristic set off in search, can improve the raw grinding Systems balanth again.

Summary of the invention

The object of the invention is to for prior art, cement slurry Vertical Mill pressure reduction be controlled the deficiency that exists, a kind of impact that effectively overcomes non-linear, the large time delay of grinding process, large inertia and many disturbances is provided, realizes the self-adaptation control method based on generalized predictive PI D of the steady control of pressure reduction in mill.

The technical scheme that the present invention adopts for achieving the above object is: a kind of intelligence control method of cement slurry Vertical Mill pressure reduction, it is characterized in that, choosing the interior pressure reduction of mill is main controlled variable, the outer circulation electric current is auxiliary controlled variable, feeding capacity is the major control variable, and injection flow rate is auxiliary control variables; Adopt the current production status of production rule method judgement according to pressure reduction and outer circulation current information in mill; Under nominal situation, adopt GPC-PID to control pressure reduction in mill;

Described GPC-PID control method is:

The A Model Distinguish: the model between pressure reduction and feeding capacity can be expressed as following form:

y(k)A(z ^-1)＝z ^-dB(z ^-1)u(k-1)+C(z ^-1)ξ(k)/Δ (2)

Wherein

A(z ^-1)＝1+a ₁z ^-1+…a _naz ^-na

B(z ^-1)＝b ₀+b ₁z ^-1+…+b _nbz ^-nb

C(z ^-1)＝c ₀+c ₁z ^-1+…+c _ncz ^-nc

In formula, A (z ^-1), B (z ^-1), C (z ^-1) be respectively the z on na, nb and na rank ^-1Multinomial, Δ=1-z ^-1Y (k) is pressure reduction in mill, and u (k) is feeding capacity, and ξ (k) is random disturbances, and d is lag time;

For the control channel between pressure reduction in feeding capacity and mill, based on the recursive least-squares discrimination method, utilize test data to obtain feeding capacity and the pulsed transfer function model that grinds interior pressure reduction;

B optimizes calculating: adopt Rolling optimal strategy.At current time k, utilize input/output information in the past and the following input message of prediction, by the transfer function model that identification obtains, pressure reduction y (k+j) in the output of predict future mill, and will predict that output and the reference locus yr (k) that sets output valve compare.Use quadratic performance index and carry out rolling optimization, performance index function is taken as:

J＝E[Y _r(k)-Y(k)][Y _r(k)-Y(k)] ^T+ΔUQΔU ^T (27)

Wherein, Q=Λ+diag (Q _j) z ^-1, Λ=diag (λ ₁λ _n), $Q_j=q_1^j{z}^{-1}+\&CenterDot;\&CenterDot;\&CenterDot;+q_{\mathrm{nb}}^j,$ $q_i^j=\mathrm{diag}({\mathrm{<figure-callout\; id="1"\; label="q\; j"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">q</figure-callout>}}_j^{},0\&CenterDot;\&CenterDot;\&CenterDot;,0),$ N is prediction time domain, j=1,2 ... N; I=1,2 ... nb;

By GPC-PID control law and traditional PID control rule are compared, by the recurrence relation of parameter in the GPC algorithm obtain three parameters of PID from tuning formulae, thereby obtain the controlling increment △ u (k) of current time, and then obtain the current controlled quentity controlled variable u (k) of the system that acts on original controlled quentity controlled variable addition.

Under unusual service condition, according to the detected value of outer circulation electric current and the deviation between desired value, regulate the size of injection flow rate according to Expert Rules, control pressure reduction and outer circulation current balance type in mill, realize the control to pressure reduction.

Described Expert Rules is the adjusting rule list of being determined by empirical value.

Describedly adopt the method for the current production status of production rule method judgement to be according to pressure reduction and outer circulation current information in mill: if pressure reduction higher than lower limit Plower and lower than upper limit Pupper, current working is nominal situation; If greater than upper limit Iupper, current working is unusual service condition to pressure reduction lower than lower limit Plower and outer circulation electric current.

The upper limit Pupper of described pressure reduction, lower limit Plower, it is given according to expertise that working condition is looked on the borders such as upper limit Iupper of outer circulation electric current.

Described GPC-PID control procedure is

Step 1 initializes: the required parameter upper lower limit value of operating mode identification module is set;

Arrange under nominal situation, the GPC-PID algorithm is used the initial value of parameter;

Parameter value in Expert Rules is set under unusual service condition;

Step 2 reads pressure reduction y (k), feeding capacity u (k), the sampled value of outer circulation electric current I and injection flow rate;

Step 3, the data of utilizing step 2 to read, according to

On-line Estimation

Δy(k)＝-a ₁Δy(k-1)+bΔu(k-1-d)+ξ(k) （6）

In parameter a ₁And b;

Step 4, the parameter of utilizing step 3 to estimate is by losing a kind equation

1＝E _j(z ^-1)A(z ^-1)Δ+z ^-jF _j(z ^-1) （7）

E _j(z ^-1)B(z ^-1)＝G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein,

E _j(z ^-1)＝1+e ₁z ^-1+...+e _j-1z ^-j+1

$F_j\left(z^{-1}\right)=f_0^j{z}^{-1}+...+f_{\mathrm{na}}^j{z}^{-\mathrm{na}}$

G _j(z ^-1)＝g ₀+g ₁z ^-1+...+g _j-1z ^-j+1

$H_j\left(z^{-1}\right)=h_0^j{z}^{-1}+...+h_{\mathrm{nb}-1}^j{z}^{-\mathrm{nb}+1}$

With

E _j+1(z ^-1)＝E _j(z ^-1)+e _jz ^-j （21）

S _j+1(z ^-1)=z[S _j(z ^-1)-A(z ^-1)Δ] （22）

g _j＝h _j，0+b ₀e _j （24）

h _j+1，i＝h _j，i+1+b _i+1e _j-g _j （25）

Recursive Solution E _j(z ^-1), G _j(z ^-1), F _j(z ^-1) and H _j(z ^-1);

Step 5 is utilized

$K_p=-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j({{f_1}^j}^{}+{2f}_2^j)$

$K_i=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j---\left(18\right)$

$K_d=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j$

Find the solution the K of GPC-PID controller _p, K _i, K _dParameter;

Step 6 is utilized

$\mathrm{\&Delta;u}\left(k\right)=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j(y_r\left(k\right)-y\left(k\right))-[\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j\left({1-z}^{-1}\right)-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j({f_1}^j+{2f}_2^j)]y\left(k\right)---\left(19\right)$

Calculate feeding capacity u (k);

Step 7 makes k=k+1, returns to step 2.

Beneficial effect of the present invention is embodied in:

1. need not to analyze the complex process of raw material vertical-grinding grinding material, only need utilize the process input and output data just can set up description pressure reduction with the Mathematical Modeling of feeding capacity size variation, identification process is simple, and can on-line identification;

2. adopt different control models according to different production status, the variation of production status is had certain adaptability;

3. take full advantage of the advantage of GPC algorithm multi-step prediction, rolling optimization and feedback compensation, the advantage of, strong robustness simple with pid algorithm, generalized predictive control is controlled with PID combined, the impact that makes pressure reduction in raw material vertical-grinding mill well overcome pure hysteresis, many disturbances and become when slow, steadily reach rapidly setting value, tracking performance is good, strong robustness, and greatly alleviated site operation people's labour intensity.

Description of drawings

Fig. 1 is cement slurry grinding process flow chart of the present invention;

Fig. 2 is GPC-PID control structure figure of the present invention;

Fig. 3 is the Based Intelligent Control schematic diagram of pressure reduction in mill of the present invention;

Fig. 4 is under nominal situation of the present invention, Vertical Mill pressure reduction control procedure flow chart;

Fig. 5 is Vertical Mill pressure reduction control procedure flow chart of the present invention.

The specific embodiment

Below in conjunction with drawings and the specific embodiments, the present invention is elaborated.

Variable is chosen: as shown in Figure 1, at t constantly, in mill, the material loading is G (t), and feeding capacity is G _in(t), the outer circulation inventory is G _c(t), going out the grind materials amount is G _oInventory G (t when (t), driving mill in mill ₀).In the mill, the loading of material is:

$G\left(t\right)=\underset{0}{\overset{t}{\&Integral;}}(G_{\mathrm{in}}\left(t\right)-G_c\left(t\right)-G_o\left(t\right))\mathrm{dt}+G\left(t_0\right)---\left(1\right)$

In mill, what of inventory in mill chamber the variation of pressure reduction directly react, what of outer circulation current response outer circulation inventory.The outer circulation electric current is also more stable when pressure reduction in feeding capacity, mill is all more stable.Therefore under nominal situation, choosing the interior pressure reduction of mill is main controlled variable.Certain when feeding capacity, when the outer circulation current fluctuation was larger, pressure reduction also can fluctuate thereupon.Therefore under unusual service condition, choose the outer circulation electric current and be auxiliary controlled variable.Feeding capacity is maximum to pressure reduction and outer circulation the influence of peak current, is exactly secondly injection flow rate in mill.Therefore, choosing feeding capacity is the major control variable, and in mill, injection flow rate is auxiliary control variables.

Operating mode identification: choosing the interior pressure reduction of mill and outer circulation electric current is the condition of operating mode judgement, and production status is divided into nominal situation and unusual service condition.Adopt production rule to divide operating mode, the operating mode judgment rule is as follows:

If pressure reduction is higher than lower limit Plower and lower than upper limit Pupper, current working is nominal situation; If greater than upper limit Iupper, current working is unusual service condition to pressure reduction lower than lower limit Plower and outer circulation electric current.In the operating mode judgment rule, the upper limit Pupper of pressure reduction, lower limit Plower, it is given according to expertise that working condition is looked on the borders such as upper limit Iupper of outer circulation electric current.

Operating mode identification is given an example: suppose, the scope that pressure reduction allows is [6000Pa, 8000Pa], is limited to 110A in the permission of outer circulation electric current.At t constantly, the detected value of pressure reduction is 5500Pa, and the outer circulation electric current is 118A.At this moment, the upper limit 110A that pressure reduction allows higher than the outer circulation electric current lower than the lower limit 6000Pa of pressure reduction and outer circulation electric current, current working is unusual service condition so.

Nominal situation:

A Model Distinguish: the main Mathematical Modeling of setting up between the interior pressure reduction of mill and feeding capacity.

According to Track character, the model between pressure reduction and feeding capacity can be expressed as following form:

y(k)A(z ^-1)＝z ^-dB(z ^-1)u(k-1)+C(z ^-1)ξ(k)/Δ (2)

Wherein

A(z ^-1)＝1+a ₁z ^-1+…a _naz ^-na

B(z ^-1)＝b ₀+b ₁z ^-1+…+b _nbz ^-nb

C(z ^-1)＝c ₀+c ₁z ^-1+…+c _ncz ^-nc

In formula, A (z ^-1), B (z ^-1), C (z ^-1) be respectively the z on na, nb and na rank ^-1Multinomial, Δ=1-z ^-1Y (k) is pressure reduction in mill, and u (k) is feeding capacity, and ξ (k) is random disturbances, and d is lag time.

As C (z ^-1)=1 o'clock, formula (1) can be write as:

Δy(k)＝-a ₁Δy(k-1)-…-a _naΔy(k-na)+b ₀Δu(k-d-1)+ (3)

…+b _nbΔu(k-d-1-nb)+ξ(k)

Following formula can be expressed as:

Wherein,

θ＝[a ₁,…,a _na,b ₀,…,b _nb] ^T

Consider when controlled device is to become, adopt the Recursive Least-square estimation model parameter θ with forgetting factor here:

In following formula, λ is forgetting factor, usually gets 0.9＜λ＜1; The initial value of P (k) is taken as σ ²I, σ are enough large positive number numbers.

The data that collect by analyzing the production scene, the Mathematical Modeling in mill between pressure reduction and feeding capacity can add pure lag system with one order inertia

Expression with its discretization, obtains difference form and is:

Δy(k)＝-a ₁Δy(k-1)+bΔu(k-1-d)+ξ(k) (6)

Lag time, d obtained according to step response curve; Add pure lag system for one order inertia, model parameter is θ=[a ₁, b] ^T

In certain cement plant, locate at steady operation point (410kg/h, 6200Pa), gather 200 groups of data of grinding interior pressure reduction and feeding capacity in production process with sampling time Ts=5s, utilize above-mentioned (4) formula and (5) formula identification to obtain model parameter a ₁And b, then with a ₁With b substitution (6) formula, and with (6) formula serialization, thereby the Mathematical Modeling that obtains between pressure reduction and feeding capacity is

B optimizes calculating: according to prediction theory, utilize until k is inputoutput data extremely constantly, during to k+j, the output of etching system is predicted, introduces and loses kind (Diophantine) equation:

1＝E _jz ^-1)A(z ^-1)Δ+z ^-jF _jz ^-1) (7)

E _jz ^-1)B(z ^-1)＝G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein,

E _j(z ^-1)＝1+e ₁z ^-1+...+e _j-1z ^-j+1

$F_j\left(z^{-1}\right)=f_0^j{z}^{-1}+...+f_{\mathrm{na}}^j{z}^{-\mathrm{na}}$

G _j(z ^-1)＝g ₀+g ₁z ^-1+...+g _j-1z ^-j+1

$H_j\left(z^{-1}\right)=h_0^j{z}^{-1}+...+h_{\mathrm{nb}-1}^j{z}^{-\mathrm{<figure-callout\; id="1"\; label="nb\; +"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">nb</figure-callout>}+1}$

The optimum prediction value that can be got k+j moment y (k+j) by (2), (7) two formulas is:

$\tilde{y}(k+j)=G_j\left(z^{-1}\right)\mathrm{\&Delta;u}(k+j-1)+F_j\left(z^{-1}\right)y\left(k\right)+H_j\left(z^{-1}\right)\mathrm{\&Delta;u}(k-1)---\left(8\right)$

When j is increased to N from 1, (8) formula is expressed as vector form:

$\tilde{y}(k+1)=\mathrm{G\&Delta;U}+\mathrm{Fy}\left(k\right)+\mathrm{H\&Delta;u}(k-1)---\left(9\right)$

Wherein

$\tilde{Y}(k+1){=[\tilde{y}(k+1),\tilde{y}(k+2),\&CenterDot;\&CenterDot;\&CenterDot;\tilde{y}(k+N)]}^T$

ΔU=[Δu(k),Δu(k-1),…Δu(k+N-1)] ^T

F＝[F ₁,F ₂,…,F _N] ^T

H＝[H ₁,H ₂,…,H _N] ^T

If given input

$T_r^T(k+1)=[y_r(k+1),\&CenterDot;\&CenterDot;\&CenterDot;,y_r(k+N)]---\left(10\right)$

Performance index function adopts the quadratic performance index to output error and controlled quentity controlled variable weighting:

$J=E[Y_r(k+1)-\tilde{Y}(k+1)][Y_r(k+1)-\tilde{Y}(k+1){]}^T+{\mathrm{\&Delta;UQ\&Delta;U}}^T---\left(11\right)$

Wherein

Q＝Λ+diag(Q _j(z ^-1)

Λ＝diag(λ ₁,...,λ _N)

$\begin{array}{cc}{Q}_j\left(z^{-1}\right)=q_1^j{z}^{-1}+...,q_{\mathrm{nb}}^j{z}^{-\mathrm{nb}}& j=1\&CenterDot;\&CenterDot;\&CenterDot;N\end{array}$

$\begin{array}{cc}{q}_i^j=\mathrm{diag}\{{\mathrm{<figure-callout\; id="1"\; label="q\; i"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">q</figure-callout>}}_i^{},0,...,0\};& i=\mathrm{1,2},...,\mathrm{nb}\end{array}$

With (9) and (10) formula substitutions (11) formula, and with object function (11) formula to controlling increment Δ U differentiate, making it is zero, obtains optimal control law:

$\mathrm{\&Delta;U}={(G^{T}G+\mathrm{\&Lambda;})}^{-1}{G}^T[Y_r-\mathrm{Fy}\left(k\right)-\mathrm{H\&Delta;u}(k-1)+G^{-T}\underset{i=1}{\overset{\mathrm{<figure-callout\; id="1"\; label="nb\; q\; i"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">nb</figure-callout>}}{\mathrm{\&Sigma;}}}{q}_i^{}{}_1^{}\mathrm{\&Delta;u}(k-i)]---\left(12\right)$

Following formula is got first trip, obtain instant controlling increment

$\mathrm{\&Delta;u}\left(k\right)=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{y}_r\left(k\right)-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{F}_j\left(z^{-1}\right)y\left(k\right)-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{H}_j\left(z^{-1}\right)\mathrm{\&Delta;u}(k-1)+G^{-\mathrm{<figure-callout\; id="1"\; label="T\; p"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">T</figure-callout>}}{p}_{}{}_1\underset{i=1}{\overset{\mathrm{<figure-callout\; id="1"\; label="nb\; q\; i"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">nb</figure-callout>}}{\mathrm{\&Sigma;}}}{q}_i^{}{}_1^{}\mathrm{\&Delta;u}(k-i)---\left(13\right)$

Wherein,

[p ₁,...,p _N]＝[1,0,...0]·(G ^TG+Λ) ^-1G ^T

Make in formula (13) $G^{-\mathrm{<figure-callout\; id="1"\; label="T\; p"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">T</figure-callout>}}{p}_{}{}_1\underset{i=1}{\overset{\mathrm{<figure-callout\; id="1"\; label="nb\; q\; i"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">nb</figure-callout>}}{\mathrm{\&Sigma;}}}{q}_i^{}{}_1^{}\mathrm{\&Delta;u}(k-i)=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{H}_j\left(z^{-1}\right)\mathrm{\&Delta;u}(k-1),$

Obtain

$\begin{array}{cc}{\mathrm{<figure-callout\; id="1"\; label="q\; i"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">q</figure-callout>}}_i^{}=\frac{{\mathrm{<figure-callout\; id="1"\; label="G\; T\; p"\; filenames="HDA0000073767840000021.png"\; state="\{\{state\}\}">G</figure-callout>}}^T}{p_{}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{h}_i^j& i=\mathrm{1,2},...,\mathrm{nb}---\left(14\right)\end{array}$

(13) formula becomes:

$\mathrm{\&Delta;u}\left(k\right)=\underset{j}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{y}_r\left(k\right)-\underset{j}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{F}_j\left(z^{-1}\right)y\left(k\right)---\left(15\right)$

The control law of traditional PI D can be write as:

Δu(k)＝K _iy _r(k)-[(K _p+K _i+K _d)-(K _p+2K _d)z ^-1+K _dz ^-2]y(k) (16)

Relatively traditional PID control rule (16) formula and improved GPC control law (15) formula, make in (15)

$\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j=K_i$ (17)

$\underset{j}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{F}_j\left(z^{-1}\right)=(K_p+K_i+K_d)-{(K_p+{2K}_d)z}^{-1}+{K_{d}z}^{-2}$

Separate following formula and get the PID controller parameter:

$K_p=-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{P}_j\left({f_1}^j\text{+}{2f}_2^j\right)$

$K_i=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{P}_j---\left(18\right)$

$K_d=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j$

Will be based on the pid parameter K of GPC _p, K _i, K _dBe brought into (16) formula, namely get the control law based on the self-adaptive PID of GPC:

$\mathrm{\&Delta;u}\left(k\right)\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j(y_r\left(k\right)-y\left(k\right))-[\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j\left({1-z}^{-1}\right)-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j({f_1}^j+{2f}_2^j)]y\left(k\right)---\left(19\right)$

K controlled quentity controlled variable constantly is

u(k)＝u(k-1)+Δu(k) (20)

Lose kind Recursive Solution of (Diophantine) equation:

When prediction output length j changes, the E in (8) formula _j(z ^-1), F _j(z ^-1), G _j(z ^-1), H _j(z ^-1) value also can change, thereby therefore (19) formula Δ u (k) also can change, and need to recomputate by losing kind (Diophantine) equation (7) formula.For E _j(z ^-1), F _j(z ^-1) recurrence formula, can be pushed away by (7) formula:

E _j+1(z ^-1)＝E _j(z ^-1)+e _jz ^-j (21)

S _j+1(z ^-1)=z[S _j(z ^-1)-A (z ^-1) Δ] (22) initial value

e ₀＝1(23)

S ₁(z ^-1)＝z[1-A(z ^-1)Δ]

For G _j(z ^-1), H _j(z ^-1) recurrence formula, can be pushed away by (7) formula:

g _j＝h _j,0+b ₀e _j (24)

h _j+1，i＝h _j，i+1+b _i+1e _j-g _j (25)

Initial value

G ₁(z ^-1)＝b ₀ (26)

H ₁(z ^-1)＝z[B(z ^-1)-b ₀]

Feedback compensation: although obviously do not provide the expression of feedback or closed loop in the derivation of GPC-PID, it has emphasized that the basic point of optimizing is consistent with real system when carrying out rolling optimization.

Under nominal situation, the control structure of Vertical Mill pressure reduction and control principle as shown in Figures 2 and 3, the execution in step of specific algorithm as shown in Figure 4 and Figure 5:

Step 1 initializes: set controlled device order na=1, nb=1 and model structure

Δy(k)＝-a ₁Δy(k-1)+bΔu(k-1-d)+ξ(k)；

Prediction time domain N=17 and control time domain M=1; In the recursive least-squares method

θ(0)＝0，P(0)＝10 ⁵；

Lose initial value in kind (Diophantine) equation by (23) formula and (26) formula setting;

Step 2 reads sampled value y (k), the u (k) of pressure reduction and feeding capacity;

Step 3 is utilized the data of step 2, according to the parameter a of (5) formula On-line Estimation (6) formula ₁And b;

Step 4: the parameter of utilizing step 3 to estimate, by losing kind (Diophantine) equation (7) formula, (21) formula, (22) formula, (24) formula and (25) formula Recursive Solution E _j(z ^-1), G _j(z ^-1), F _j(z ^-1) and H _j(z ^-1);

Step 5: the K that calculates the GPC-PID controller with (18) formula _p, K _i, K _dParameter;

Step 6: calculate feeding capacity u (k) with (19) formula;

Step 7: make k=k+1, return to step 2.

Under unusual service condition, adopt pressure reduction and outer circulation current balance type in expert rule control mode tuning mill.According to the detected value of outer circulation electric current and the deviation delta I between desired value, regulate the size of injection flow rate according to expertise.

The below illustrates the course of work of Expert Rules take an example: the scope of establishing the pressure reduction permission is [6000Pa, 8000Pa], is limited to 110A in the permission of outer circulation electric current.At t constantly, the detected value of pressure reduction is 5500Pa, and the outer circulation electric current is 118A, and this moment is according to the rule of operating mode judgement: pressure reduction lower than lower limit 6000Pa and outer circulation electric current higher than upper limit 110A, current working is unusual service condition so, adopts the method for Expert Rules to regulate injection flow rate in mill.According to the detected value of outer circulation electric current and the deviation between desired value, regulate the size of injection flow rate according to Expert Rules.

Regulate rule as shown in table 1.

Table 1 expert regulates rule

Wherein, Δ I is the detected value of outer circulation electric current and the deviation between desired value; Tj(j=0,1 ... 5) be the deviation limit, Tj belongs to [110A, 125A]; a _i(i=0,1 ... 6) correction value that sets value for injection flow rate, a _iBelong to [0m3/h, 1m3/h]; The occurrence of each parameter is provided by expertise.

Claims (6)

1. the intelligence control method of a cement slurry Vertical Mill pressure reduction, is characterized in that, choosing the interior pressure reduction of mill is main controlled variable, and the outer circulation electric current is auxiliary controlled variable, and feeding capacity is the major control variable, and injection flow rate is auxiliary control variables; Adopt the current production status of production rule method judgement according to pressure reduction and outer circulation current information in mill; Under nominal situation, adopt GPC-PID to control pressure reduction in mill;

Described GPC-PID control method is:

The A Model Distinguish: the model between pressure reduction and feeding capacity can be expressed as following form:

y(k)A(z ^-1)＝z ^-dB(z ^-1)u(k-1)+C(z ^-1)ξ(k)/Δ (2)

Wherein

A(z ^-1)＝1+a ₁z ^-1+…a _naz ^-na

B(z ^-1)＝b ₀+b ₁z ^-1+…+b _nbz ^-nb

C(z ^-1)＝c ₀+c ₁z ^-1+…+c _ncz ^-nc

In formula, A (z ^-1), B (z ^-1), C (z ^-1) be respectively the z on na, nb and na rank ^-1Multinomial, Δ=1-z ^-1Y (k) is pressure reduction in mill, and u (k) is feeding capacity, and ξ (k) is random disturbances, and d is lag time;

For the control channel between pressure reduction in feeding capacity and mill, based on the recursive least-squares discrimination method, utilize test data to obtain feeding capacity and the pulsed transfer function model that grinds interior pressure reduction;

B optimizes calculating: adopt Rolling optimal strategy, at current time k, utilize input/output information in the past and the following input message of prediction, the transfer function model that obtains by identification, pressure reduction y (k+k) in the output mill of predict future, and will predict that output and the reference locus yr (k) that sets output valve compare, and to use quadratic performance index and carry out rolling optimization, performance index function is taken as:

J=E[Y _r(k)-Y (k)] [Y _r(k)-Y (k)] ^T+ Δ UQ Δ U ^T(27) wherein, Q=Λ+diag (Q _j) z ^-1, Λ=diag (λ ₁λ _n), $Q_j=q_1^j{z}^{-1}+\&CenterDot;\&CenterDot;\&CenterDot;+q_{\mathrm{nb}}^j,$ $q_i^j=\mathrm{diag}(q_i^1,0\&CenterDot;\&CenterDot;\&CenterDot;,0),$ N is prediction time domain, j=1,2...N; I=1,2...nb;

By GPC-PID control law and traditional PID control rule are compared, by the recurrence relation of parameter in the GPC algorithm obtain three parameters of PID from tuning formulae, thereby obtain the controlling increment △ u (k) of current time, and then obtain the current controlled quentity controlled variable u (k) of the system that acts on original controlled quentity controlled variable addition.

2. the intelligence control method of a kind of cement slurry Vertical Mill pressure reduction according to claim 1, it is characterized in that, under unusual service condition, according to the detected value of outer circulation electric current and the deviation between desired value, regulate the size of injection flow rate according to Expert Rules, control pressure reduction and outer circulation current balance type in mill, realize the control to pressure reduction.

3. the intelligence control method of a kind of cement slurry Vertical Mill pressure reduction according to claim 2, is characterized in that, described Expert Rules is the adjusting rule list of being determined by empirical value.

4. the intelligence control method of a kind of cement slurry Vertical Mill pressure reduction according to claim 1, it is characterized in that, describedly adopt the method for the current production status of production rule method judgement to be according to pressure reduction and outer circulation current information in mill: if pressure reduction higher than lower limit Plower and lower than upper limit Pupper, current working is nominal situation; If greater than upper limit Iupper, current working is unusual service condition to pressure reduction lower than lower limit Plower and outer circulation electric current.

5. the intelligence control method of a kind of cement slurry Vertical Mill pressure reduction according to claim 4, is characterized in that, the upper limit Pupper of described pressure reduction, lower limit Plower, and it is given according to expertise that working condition is looked on the upper limit Iupper border of outer circulation electric current.

6. the intelligence control method of a kind of cement slurry Vertical Mill pressure reduction according to claim 1, is characterized in that, described GPC-PID control procedure is

Step 1 initializes: the required parameter upper lower limit value of operating mode identification module is set;

Arrange under nominal situation, the GPC-PID algorithm is used the initial value of parameter;

Parameter value in Expert Rules is set under unusual service condition;

Step 2 reads pressure reduction y (k), feeding capacity u (k), the sampled value of outer circulation electric current I and injection flow rate;

Step 3, the data of utilizing step 2 to read, according to

On-line Estimation

Δ y (k)=-a ₁Δ y (k-1)+b Δ u (k-1-d)+ξ k) the parameter a in (6) ₁And b;

Step 4, the parameter of utilizing step 3 to estimate is by losing a kind equation

1＝E _j(z ^-1)A(z ^-1)Δ+z ^-jF _j(z ^-1) （7）

E _j(z ^-1)B(z ^-1)＝G _j(z ^-1)+z ^-jH _j(z ^-1)

Wherein,

E _j(z ^-1)＝1+e ₁z ^-1+...+e _j-1z- ^j+1

$F_j\left(z^{-1}\right)=f_0^j{z}^{-1}+...+f_{\mathrm{na}}^j{z}^{-\mathrm{na}}$

G _j(z ^-1)＝g ₀+g ₁z ^-1+...+g _j-1z ^-j+1

$H_j\left(z^{-1}\right)=h_0^j{z}^{-1}+...+h_{\mathrm{nb}-1}^j{z}^{-\mathrm{nb}+1}$

With

E _j+1(z ^-1)E _j(z ^-1)+e _jz ^-j （21）

S _j+1(z ^-1)=z[S _j(z ^-1)-A(z ^-1)Δ] （22）

g _j＝h _j，0+b ₀e _j （24）

h _{J+1, i}=h _{J, i+1}+ b _i+1e _j-g _j(25) Recursive Solution E _j(z ^-1), G _j(z ^-1), F _j(z ^-1) and H _j(z ^-1);

Step 5 is utilized

$K_p=-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j({f_1}^j+{2f}_2^j)$

$K_i=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j---\left(18\right)$

$K_d=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j$

Find the solution the K of GPC-PID controller _p, K _i, K _dParameter;

Step 6 is utilized

$\mathrm{\&Delta;u}\left(k\right)=\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j(y_r\left(k\right)-y\left(k\right))-[\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j{f}_2^j\left({1-z}^{-1}\right)-\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{p}_j({{f_1}^j}^{}+{2f}_2^j)]y\left(k\right)---\left(19\right)$ Calculate feeding capacity u (k);

Step 7 makes k=k+1, returns to step 2.

Images