CN107270283B - Multivariable constraint predictive control method based on circulating fluidized bed unit - Google Patents
Multivariable constraint predictive control method based on circulating fluidized bed unit Download PDFInfo
- Publication number
- CN107270283B CN107270283B CN201710579652.9A CN201710579652A CN107270283B CN 107270283 B CN107270283 B CN 107270283B CN 201710579652 A CN201710579652 A CN 201710579652A CN 107270283 B CN107270283 B CN 107270283B
- Authority
- CN
- China
- Prior art keywords
- formula
- control
- unit
- fluidized bed
- circulating fluidized
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 66
- 238000005457 optimization Methods 0.000 claims abstract description 25
- 230000004044 response Effects 0.000 claims abstract description 12
- 238000005096 rolling process Methods 0.000 claims abstract description 8
- 238000012937 correction Methods 0.000 claims abstract description 6
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims abstract description 4
- 229910052799 carbon Inorganic materials 0.000 claims abstract description 4
- 230000008021 deposition Effects 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 24
- 238000005259 measurement Methods 0.000 claims description 12
- 239000011541 reaction mixture Substances 0.000 claims description 12
- 239000003245 coal Substances 0.000 claims description 10
- 230000026676 system process Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 150000001875 compounds Chemical class 0.000 claims description 6
- 238000002485 combustion reaction Methods 0.000 claims description 5
- 238000002474 experimental method Methods 0.000 claims description 5
- 239000000446 fuel Substances 0.000 claims description 5
- 230000007246 mechanism Effects 0.000 claims description 3
- 208000011231 Crohn disease Diseases 0.000 claims description 2
- 238000013178 mathematical model Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims 1
- 238000011160 research Methods 0.000 abstract description 8
- 238000009825 accumulation Methods 0.000 abstract description 2
- 238000004088 simulation Methods 0.000 description 22
- 230000000694 effects Effects 0.000 description 8
- 230000009471 action Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 238000011217 control strategy Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011056 performance test Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000006477 desulfuration reaction Methods 0.000 description 1
- 230000023556 desulfurization Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000036039 immunity Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000010977 unit operation Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23C—METHODS OR APPARATUS FOR COMBUSTION USING FLUID FUEL OR SOLID FUEL SUSPENDED IN A CARRIER GAS OR AIR
- F23C10/00—Fluidised bed combustion apparatus
- F23C10/18—Details; Accessories
- F23C10/28—Control devices specially adapted for fluidised bed, combustion apparatus
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F23—COMBUSTION APPARATUS; COMBUSTION PROCESSES
- F23C—METHODS OR APPARATUS FOR COMBUSTION USING FLUID FUEL OR SOLID FUEL SUSPENDED IN A CARRIER GAS OR AIR
- F23C2206/00—Fluidised bed combustion
- F23C2206/10—Circulating fluidised bed
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Feedback Control In General (AREA)
- Medicines Containing Antibodies Or Antigens For Use As Internal Diagnostic Agents (AREA)
Abstract
A multivariable constraint predictive control method based on a circulating fluidized bed unit relates to the research field of AGC control of a circulating fluidized bed. The method comprises the following steps: determining an AGC operation condition interval of a control object, and dividing working condition points; obtaining the step response coefficients of the control object under 4 working condition points; obtaining approximate expression forms of the state space of the control object under 4 working condition points; acquiring a linear variable parameter model consisting of point models under different working conditions; obtaining a prediction model of a prediction control algorithm; performing rolling optimization of a predictive control algorithm, and calculating to obtain an optimal control quantity; and carrying out feedback correction of the predictive control algorithm. The invention fully utilizes the dynamic influence of primary air quantity on carbon deposition and heat accumulation of the circulating fluidized bed boiler, and improves the AGC load tracking rate of the unit. The method can improve the capability of describing the remarkable nonlinear characteristics of the circulating fluidized bed unit, can more accurately describe the characteristics of a control object, and reduces the influence of model mismatch on the performance of a control algorithm.
Description
Technical Field
The invention relates to the field of research of AGC (automatic gain control) of a circulating fluidized bed, in particular to a multivariable constraint predictive control method based on a circulating fluidized bed unit.
Background
The Circulating Fluidized Bed Boiler (CFBB) has the advantages of high combustion efficiency, wide fuel adaptability, low combustion pollution and high desulfurization efficiency, and has very wide application prospect under the large background of energy shortage and environmental protection. However, the circulating fluidized bed unit is a very complex multivariable system, has characteristics of nonlinearity, large delay, large inertia, strong coupling and the like, and has a great difference from the conventional industrial pulverized coal boiler unit in structure, and currently, the AGC control research for the CFBB unit mainly has the following two problems: firstly, AGC control structure research is less, the control structure of the traditional pulverized coal boiler unit is directly applied in engineering practice, and the characteristics of CFBB cannot be fully utilized; secondly, the control strategy needs to be improved, the traditional control scheme of PID plus feedforward decoupling can not obtain ideal control effect, and especially the load tracking performance of AGC is poor.
Aiming at the second problem, many scholars carry out intensive research, apply the predictive control algorithm to the AGC control problem of the CFBB unit and achieve certain effect. However, for the application of the predictive control algorithm in the aspect of AGC control of the CFBB set, the following problems mainly exist:
(1) the research of the predictive control algorithm is carried out based on a single working condition, and the nonlinear characteristic of a CFBB object is not considered;
(2) the related input and output constraint problems are not considered;
(3) the prediction model mainly adopts a transfer function form, and the capability of eliminating the influence of the immeasurable disturbance needs to be enhanced.
In summary, for the control research of the circulating fluidized bed unit, it is important to provide a novel control structure that fully utilizes the characteristics of the unit, and design a novel control strategy based on the novel control structure to cope with the characteristics of object nonlinearity, large delay, large inertia, strong coupling, and the like.
Disclosure of Invention
Aiming at the technical problems, the invention provides a multivariable constraint predictive control method based on a circulating fluidized bed unit.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a multivariable constraint predictive control method based on a circulating fluidized bed unit comprises the following steps:
s1: determining AGC operation condition interval of control object, dividing operating condition points
Determining AGC operation interval according to the characteristics of the control object of the specific circulating fluidized bed unit, and taking
Θ＝{Ne|Ne∈[40％ECR,110％ECR]} (1)
In the formula (1), theta is an AGC operation interval, Ne is unit power (MW), ECR is a unit rated working condition, 40% ECR is a low-load operation state, and 110% ECR is an output super-power operation state;
in the interval range, 4 specific working condition points, namely 40% ECR, 60% ECR, 80% ECR and 100% ECR, are selected at intervals of 20% ECR;
s2: obtaining the step response coefficients of the control object under 4 working condition points
Performing a step response dynamic experiment on the actual power plant object at the 4 working condition points; taking the l-th working condition point as an example, if the object has m-dimensional control input and p-dimensional output, the obtained corresponding step response coefficient matrix Al,iAs shown in formula (2):
in the formula (2), m is the number of input quantities, p is the number of output quantities, N is the length of the model, and l is the number of working condition points;
s3: obtaining approximate expression form of control object state space under 4 working condition points
Similarly, taking the ith operating point as an example, the approximate expression form of the state space of the control object is shown in formula (3):
in the formula (3), Xl(k) A state variable of a control object of the first working point, y (k) is an output variable, delta u (k) is a control quantity increment, and Sl、AlAnd ClAre respectively shown as formulas (4), (5) and (6);
in the formula (5), Al,iAs shown in formula (2);
s4 obtaining linear variable parameter Model (L initial-Parameters-Varying Model, L PV) composed of sub-models of different working conditions
The L PV model is shown in equation (7):
in the formula (7), the reaction mixture is,
C(w)＝[λ1(w)C1λ2(w)C2… λl(w)Cl](9)
in the formula (9), w is a scheduling variable of an L PV model, taken as a power parameter Ne of the circulating fluidized bed unit, λ is a weight function, and is a potential function of the scheduling variable w;
s5: obtaining a predictive model of a predictive control algorithm
The prediction model is shown in equation (11):
y(k)＝FX(k)+ΦΔu(k) (11)
in the formula (11), the reaction mixture is,
in the formula (12), yn(k +1| k) is an output predicted value at the time k +1 obtained by calculation based on data at the time k, wherein n is 1. In the formula (13), Δ uz(k) For control increment, z is 1, m is the number of input quantities; in the formula (14), P is a prediction time domain of the prediction control algorithm;
s6: rolling optimization of the predictive control algorithm is carried out, and the optimal control quantity is obtained through calculation
The control object is a 3-input 3-output structure, wherein 3 input quantities are respectively a coal supply quantity (B, Kg/s) and a main steam regulating valve (u)T%) and primary wind frequency modulation command (u)W%); the 3 output quantities are respectively set power (Ne, MW) and main steam pressure (p)tMPa) and hearth bed temperature (Tb, deg.c); according to the specific requirements of AGC control of the circulating fluidized bed unit, the power of the unit and the pressure of main steam are set values for control, and the bed temperature of the hearth is interval control, so that the performance index established at the moment k is as shown in a formula (16):
in the formula (16), J (k) is a performance index function, w1,2(k) Is set value of unit power and main steam pressure, y1,2(k) The actual output values of the unit power and the main steam pressure, delta u (k) is the increment of the control quantity, q1,q2,q3Is an error weight coefficient, r1,r2,r3Is a control weight coefficient;P(k) the deviation of the bed temperature output from the set value is shown in formula (17):
in the formula (17), the compound represented by the formula (I),
in formula (18), TbminAnd TbmaxIn the bed temperature rangeUpper and lower limits of control;
meanwhile, constraint conditions are set for the performance indexes, including control quantity and control quantity increment, as shown in formula (19):
in the formula (19), Bmin、Bmax，uT,min、uT,max，uW,min、uW,maxRespectively, control quantity B, uT、uWMinimum and maximum values of; dBmin、dBmax，duT,min、duT,max，duW,min、duW,maxRespectively control quantity increment dB and duT、duWMinimum and maximum values of;
based on the performance index and the constraint condition, calculating to obtain the corresponding optimal control increment delta uM(k) The corresponding instantaneous control increment Δ u (k) is:
Δu(k)＝LΔuM(k) (20)
in the formula (20), the reaction mixture is,
in the formula (21), Δ uz,M(k) As shown in formula (22), z is 1.
In the formula (23), the compound represented by the formula,
L0＝[1 0 … 0](1×M)(24)
in the formula (19) -formula (24), M is a control time domain of the predictive control;
thereby obtaining Δ u (k), and calculating the value of u (k) by u (k) ═ u (k-1) + Δ u (k);
s7: and carrying out feedback correction of the predictive control algorithm.
Preferably, the state variable x (k) in step S4 is obtained by a rolling time domain estimation method, including the steps of:
s4.1: the state space description of the discrete-time linear time-invariant system considering the multiple-input multiple-output control object is shown as equation (25):
in the formula (25), v (k) is measurement noise, w (k) is system process noise, and is a system noise input matrix; the system process noise w (k) and the measurement noise v (k) are gaussian white noise sequences, and both have statistical properties shown in formula (26):
in the formula (26), Q0Is a non-negative definite matrix and represents a variance matrix of system process noise W (k); r0Is positive definite matrix, and represents variance matrix of measurement noise V (k), (k, j) is a Crohn's Nike symbol;
s4.2: acquiring input and output measurement data of the current time k, as shown in equation (27):
s4.3: calculating an optimal solution according to a constrained optimization problem, wherein the constrained optimization problem is represented by a formula (28):
in the formula (28), RMHE、QMHE、Π0Positive definite punishment matrixes respectively corresponding to the variables;
estimating an initial state estimator X (k-N) and an initial penalty matrix pi by an arrival cost functionk-NWherein the arrival cost functionAs shown in equation (29):
and pik-NIterative calculations based on the forward rica equation,
obtained based on kalman estimation;
the joint vertical type (25) -formula (29) calculates to obtain an optimal solution, as shown in formula (30):
answerop＝(X(k-N|k-1)op,W(N|k-1)op) (30)
in the formula (30), the reaction mixture,
W(N|k-1)op＝{W(k-N|k-1)op,W(k-N+1|k-1)op,…,W(k-1|k-1)op} (31)
s4.4: iterative computation is carried out by utilizing the optimal solution of the optimization problem to obtain a system state quantity estimated value;
simultaneous type (25), formula (26) and formula (30), calculating X (k | k-1)op。
Compared with the prior art, the invention has the beneficial effects that:
(1) the dynamic influence of the primary air quantity on carbon deposition and heat accumulation of the circulating fluidized bed boiler is fully utilized, and the AGC load tracking rate of the unit is improved.
(2) The prediction model constructed based on the L PV model can improve the capability of describing the remarkable nonlinear characteristics of the circulating fluidized bed unit, can more accurately describe the nonlinear characteristics of a control object, and reduces the influence of model mismatch on the performance of a control algorithm.
Drawings
The multivariable constraint predictive control method based on the circulating fluidized bed unit is further detailed in the following by combining the embodiment and the attached drawings.
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a simulation experiment curve (200 MW-270 MW) of different control algorithms for AGC control of a CFBB unit;
FIG. 3 is a simulation experiment curve (270 MW-300 MW) of different control algorithms for AGC control of a CFBB unit;
FIG. 4 is a schematic diagram of bed temperature interval control simulation (the bed temperature changes in the control interval);
FIG. 5 is a simulation diagram of bed temperature control (the bed temperature varies outside the control range);
FIG. 6 is a graph of the disturbance rejection performance test for different control algorithms;
fig. 7 is a robust performance test curve of different control algorithms.
Detailed Description
Referring to fig. 1, a multivariable constraint predictive control method based on a circulating fluidized bed unit includes the following steps: s1: determining an AGC operation condition interval of a control object, and dividing working condition points; s2: obtaining the step response coefficients of the control object under 4 working condition points; s3: obtaining approximate expression forms of the state space of the control object under 4 working condition points; s4: acquiring a linear variable parameter model consisting of point models under different working conditions; s5: obtaining a prediction model of a prediction control algorithm; s6: performing rolling optimization of a predictive control algorithm, and calculating to obtain an optimal control quantity; s7: and carrying out feedback correction of the predictive control algorithm.
A multivariable constraint predictive control method based on a circulating fluidized bed unit specifically comprises the following steps:
s1: determining AGC operation condition interval of control object, dividing operating condition points
Determining AGC operation interval according to the characteristics of the control object of the specific circulating fluidized bed unit, and taking
Θ＝{Ne|Ne∈[40％ECR,110％ECR]} (1)
In the formula (1), Θ is an AGC operation interval, Ne is unit power (MW), ECR is a unit rated working condition, 40% ECR is a low-load operation state, and 110% ECR is an output super-power operation state.
Within the range, 4 specific operating points, namely 40% ECR, 60% ECR, 80% ECR, and 100% ECR, are selected at intervals of 20% ECR.
S2: obtaining the step response coefficients of the control object under 4 working condition points
And performing step response dynamic experiments on the actual power plant objects at the 4 working condition points. Taking the l-th working condition point as an example, if the object has m-dimensional control input and p-dimensional output, the obtained corresponding step response coefficient matrix Al,iAs shown in formula (2):
in the formula (2), m is the number of input quantities, p is the number of output quantities, N is the model length, and l is the number of operating points.
S3: obtaining approximate expression form of control object state space under 4 working condition points
Similarly, taking the ith operating point as an example, the approximate expression form of the state space of the control object is shown in formula (3):
in the formula (3), Xl(k) A state variable of a control object of the first working point, y (k) is an output variable, delta u (k) is a control quantity increment, and Sl、AlAnd ClAre respectively shown as formulas (4), (5) and (6).
In the formula (5), Al,iAs shown in formula (2).
S4 obtaining linear variable parameter Model (L initial-Parameters-Varying Model, L PV) composed of sub-models of different working conditions
The L PV model is shown in equation (7):
in the formula (7), the reaction mixture is,
C(w)＝[λ1(w)C1λ2(w)C2…λl(w)Cl](9)
in the formula (9), w is a scheduling variable of the L PV model, taken as a power parameter Ne of the circulating fluidized bed unit, λ is a weight function, and is a potential function of the scheduling variable w.
The state variable X (k) is obtained by a rolling time domain estimation method, and comprises the following steps:
s4.1: the state space description of the discrete-time linear time-invariant system considering the multiple-input multiple-output control object is shown as equation (25):
in equation (25), v (k) is measurement noise, w (k) is system process noise, and w (k) is a system noise input matrix. The system process noise w (k) and the measurement noise v (k) are gaussian white noise sequences, and both have statistical properties shown in formula (26):
in the formula (26), Q0Is a non-negative definite matrix and represents a variance matrix of the system process noise W (k). R0Is positive definite matrix, and represents variance matrix of measurement noise V (k), and (k, j) is a Kronik symbol.
S4.2: acquiring input and output measurement data of the current time k, as shown in equation (27):
s4.3: calculating an optimal solution according to a constrained optimization problem, wherein the constrained optimization problem is represented by a formula (28):
in the formula (28), RMHE、QMHE、Π0Respectively positive definite punishment matrixes of the corresponding variables.
Estimating initial state estimators by arrival cost function
And initial penalty matrix Πk-NWherein the arrival cost function is as shown in equation (29):
and pik-NIterative calculations based on the forward rica equation,
it is obtained based on kalman estimation.
The joint vertical type (25) -formula (29) calculates to obtain an optimal solution, as shown in formula (30):
answerop＝(X(k-N|k-1)op,W(N|k-1)op) (30)
in the formula (30), the reaction mixture,
W(N|k-1)op＝{W(k-N|k-1)op,W(k-N+1|k-1)op,…,W(k-1|k-1)op} (31)
s4.4: iterative computation is carried out by utilizing the optimal solution of the optimization problem to obtain a system state quantity estimated value;
simultaneous type (25), formula (26) and formula (30), calculating X (k | k-1)op。
S5: obtaining a predictive model of a predictive control algorithm
The prediction model is shown in equation (11):
y(k)＝FX(k)+ΦΔu(k) (11)
in the formula (11), the reaction mixture is,
in the formula (12), ynThe (k +1| k) is an output prediction value at the time k +1 calculated based on the data at the time k, n is 1. In the formula (13), Δ uz(k) For the control increment, z is 1. In equation (14), P is the prediction time domain of the prediction control algorithm.
S6: rolling optimization of the predictive control algorithm is carried out, and the optimal control quantity is obtained through calculation
The control object is a 3-input 3-output structure, wherein 3 input quantities are respectively a coal supply quantity (B, Kg/s) and a main steam regulating valve (u)T%) and primary wind frequency modulation command (u)W%). The 3 output quantities are respectively set power (Ne, MW) and main steam pressure (p)tMPa) and the hearth bed temperature (Tb, DEG C). According to the specific requirements of AGC control of the circulating fluidized bed unit, the power of the unit and the pressure of main steam are set values for control, and the bed temperature of the hearth is interval control, so that the performance index established at the moment k is as shown in a formula (16):
in the formula (16), J (k) is a performance index function, w1,2(k) Is composed ofSet values of the power of the units and the pressure of the main steam, y1,2(k) The actual output values of the unit power and the main steam pressure, delta u (k) is the increment of the control quantity, q1,q2,q3Is an error weight coefficient, r1,r2,r3Is a control weight coefficient.P(k) The deviation of the bed temperature output from the set value is shown in formula (17):
in the formula (17), the compound represented by the formula (I),
in formula (18), TbminAnd TbmaxThe upper and lower limits of bed temperature interval control.
Meanwhile, constraint conditions are set for the performance indexes, including control quantity and control quantity increment, as shown in formula (19):
in the formula (19), Bmin、Bmax，uT,min、uT,max，uW,min、uW,maxRespectively, control quantity B, uT、uWMinimum and maximum values of. dBmin、dBmax，duT,min、duT,max，duW,min、duW,maxRespectively control quantity increment dB and duT、duWMinimum and maximum values of.
Based on the performance index and the constraint condition, calculating to obtain the corresponding optimal control increment delta uM(k) The corresponding instantaneous control increment Δ u (k) is:
Δu(k)＝LΔuM(k) (20)
in the formula (20), the reaction mixture is,
in the formula (21), Δ uz,M(k) As shown in equation (22), z is 1.
In the formula (23), the compound represented by the formula,
L0＝[1 0 … 0](1×M)(24)
in the expression (19) to the expression (24), M is a control time domain of the prediction control.
Δ u (k) is thus obtained, and the value of u (k) is calculated by u (k) ═ u (k-1) + Δ u (k).
S7: performing feedback correction of predictive control algorithms
The technical scheme of the invention is further described in the following with reference to specific embodiments, and the control object of the circulating fluidized bed unit related in the example is represented by a non-linear mathematical model constructed based on a mechanism.
In the formula (32), BC(t) the carbon deposition amount in the hearth at the moment t, Kg; qi(t) total heat released by fuel combustion at time t, MJ; qfMJ/Kg is the unit calorific value of the fuel; qWPrimary wind heat capacity, MJ; p is a radical ofdIs the drum pressure, MPa.
To illustrate the role of the L PV model in constructing the predictive model of the control algorithm, the present example organizes the data related to 40%, 60%, 80% and 100% load operating points based on dynamic experimental data, identifies and obtains 4 sets of state space models, constructs L PV models, and applies the models to the predictive model description in the control algorithm.
Next, in this example, the mechanism model shown in the formula (1) is used as a control target, and simulation verification is performed based on different control algorithms. Firstly, comparing the traditional PID control algorithm with the multivariable constraint predictive control algorithm provided by the patent, taking an AGC load-up stage as an example, and verifying the performance of the algorithm; secondly, taking a main steam pressure constant pressure control stage in an AGC load-up stage as an example, verifying the effect of bed temperature interval control; finally, taking the main steam pressure constant pressure control stage in the AGC load-up stage as an example, comparing the Dynamic Matrix Control (DMC) with the multivariable constraint predictive control algorithm provided by the patent, and verifying the anti-interference performance and robustness of the algorithm.
According to the technical requirements of power grids in China, the capacity of a thermal power generating unit is 300 MW-600 MW (including 300MW), and the adjusting rate needs to reach 3MW/min during AGC joint debugging. Taking a 300MW circulating fluidized bed unit as an example, the AGC regulation rate needs to meet 1.0% Ne/min, in order to explain the effect of a predictive control algorithm and embody the advantages of the predictive control algorithm, a load-raising rate (namely 9MW/min) of 3.0% Ne/min is designed during a simulation experiment, namely a set value of the unit power is a ramp rising signal of 9MW/min, the lower limit of the AGC operation is set to 200MW, and the upper limit is set to 320 MW; the main steam pressure is controlled by two modes of constant pressure operation and sliding pressure operation, for example, a 300MW circulating fluidized bed unit, the sliding pressure operation only works when the load of the unit is 140 MW-270 MW, and the constant value of the sliding pressure is determined by a load instruction; when the unit operation load is lower than 140MW or higher than 270MW, the pressure control is switched to the constant pressure operation, and the main steam pressure set value is still linearly formed according to the unit load instruction; the variation range of the bed temperature is 850-950 ℃, namely the set value of the bed temperature is (850 ℃, 950 ℃).
Simulation experiments were performed by the simulink module of Matlab software:
simulation experiment (one): AGC simulation experiment
The simulation experiment is carried out based on two working conditions of 200 MW-270 MW and 270 MW-300 MW, wherein the pressure control in the first working condition range is sliding pressure operation, and the second working condition range is constant pressure operation. According to the AGC operation control requirement, a simulation experiment is completed, the predictive control optimization algorithm provided by the patent is compared with the traditional PID control algorithm, and an AGC control curve graph shown in FIG. 2 is obtained in a sliding pressure operation working condition interval. And obtaining an AGC control curve chart shown in figure 3 in the constant-pressure operation condition interval.
As can be seen from the curves in fig. 2, compared with the conventional PID control, the multivariable constraint predictive control optimization algorithm has significant advantages, namely, excellent AGC tracking performance and small overshoot. The fluctuation of the control action is also small; the PID control has a poor control effect on a multivariable system, has obvious fluctuation of the control quantity, has poor processing capability on the non-minimum phase characteristics of a CFBB object, and has a very obvious reverse characteristic in the initial control stage. In terms of bed temperature interval control, the third main loop is controlled by a single PID controller, and the lower limit of the initial bed temperature set value of 850 ℃ is far away from the stable value of the current working condition, so that the PID controller inevitably acts violently.
As can be seen from the curves in fig. 3, compared with the sliding pressure operation stage, the pressure curve of the predictive control algorithm in the constant pressure operation has a certain fluctuation in the load-increasing stage, but in terms of AGC load-increasing performance, the effect of the predictive control optimization algorithm is also significant compared with the conventional PID control strategy. Firstly, the load tracking effect is achieved, the PID control is not ideal in tracking performance in a load-increasing stage, and after a load set value is stable, the control action and the corresponding control quantity still fluctuate sharply. In terms of bed temperature interval control, the advantage of a predictive control algorithm in the aspect of processing such multivariable control problems can be more highlighted here, because the bed temperature is already within a set value range under the current working condition, a single PID controller of a third main loop does not directly participate in the control action any more, the control quantity is always maintained at an initial value, and actually, due to the coupling action of a multivariable system, the primary air volume also has influence on the other two output quantities of a CFBB object, and the action of the multivariable system is fully utilized to improve the tracking performance of AGC control.
Simulation experiment (II): bed temperature interval control performance simulation experiment
To further illustrate the performance of the control algorithm in interval control of the bed temperature, a simulation experiment is performed below by taking an AGC operating interval of a 270 MW-300 MW working condition as an example, wherein the simulation experiment is divided into the following two specific cases:
(1) disturbance experiment in bed temperature control interval
Assuming that the bed temperature output of the control object is disturbed by a sinusoidal signal with an amplitude of 5 ℃ and a period of 100s at 2000s, the simulation curve is shown in fig. 4.
As can be seen from the graph of fig. 4, the multivariable constrained prediction control can effectively control the bed temperature within the predetermined interval, the disturbance on the output side rapidly causes the fluctuation of the bed temperature parameter, but the control quantity does not change obviously after 2000s, and it can be seen that as long as the output quantity changes within the allowed set interval, the control quantity will remain unchanged, so that the frequent action of the control actuator can be obviously reduced, and the service life can be prolonged.
(2) Disturbance experiment outside bed temperature control interval
Assuming that the bed temperature output of the control object is disturbed by the step signal with the amplitude of 25 ℃ at 2000s, the simulation curve is shown in fig. 5.
It can be seen from the graph of fig. 5 that when the controlled object is interfered by the step signal with the amplitude of 25 ℃, the bed temperature exceeds the upper limit of the set interval, and at this time, the multivariable constrained predictive control algorithm can quickly reconcile the bed temperature within the set range by coordinating 3 control quantities.
As can be seen from the simulation experiments shown in fig. 4 and 5, the multivariate constrained predictive control algorithm provided by the patent has an ideal control effect when dealing with the interval control problem, and can be applied to the bed temperature interval control of the CFBB unit.
Simulation experiment (c): multivariable constraint predictive control algorithm performance simulation experiment
In order to further explain the control performance of the multivariable constraint predictive control algorithm, the following simulation research is continuously carried out from two aspects by taking the working condition AGC operation interval of 270 MW-300 MW as an example:
(1) test of noise immunity
Assuming that the power loop of the power bank is subjected to step disturbance at 1000s, the disturbance rejection performance of the predictive control optimization algorithm proposed by the present patent is compared with that of a basic dynamic matrix predictive control algorithm (DMC), so as to obtain a response curve as shown in fig. 6.
As can be seen from the graph of fig. 6, when the system is not disturbed by the disturbance quantity, the performance of the basic DMC algorithm and the predictive control optimization algorithm is similar when the AGC load is tracked, but the fluctuation of the control quantity is slightly larger than the optimization algorithm; when disturbance occurs, the performance of the optimization algorithm in the aspect of stability is obviously superior to that of the DMC algorithm, the DMC algorithm has very obvious fluctuation and cannot be stabilized within a long time, the control quality tends to deteriorate, after the optimization algorithm is adjusted through necessary control quantity when the disturbance occurs, the control quantity and the controlled quantity can be stabilized quickly, and the control quality is still excellent.
(2) Robust performance testing
For the CFBB unit, the quality of coal used for combustion is generally poor, and the fluctuation of the coal quality is very frequent in the actual operation process, so that frequent model mismatch of a control algorithm is caused, and the requirement on the robust performance of a control system is higher. Assuming that the parameter K1 related to the coal feeding amount is increased by 20% at 1000s to simulate the change of coal quality and the mismatch of the prediction model, the robustness performance comparison between the predictive control optimization algorithm proposed by the present patent and the basic dynamic matrix predictive control algorithm (DMC) is performed, and then the change curve as shown in fig. 7 is obtained.
As can be seen from the curve of FIG. 7, when the coal quality changes, i.e., the prediction model is mismatched, the influence of the DMC algorithm and the optimization algorithm can be finally inhibited due to the introduction of the feedback correction link. In contrast, the fluctuation generated by the DMC algorithm is more severe and longer in duration, the overall control quality tends to deteriorate, and the optimization algorithm can be stabilized faster overall and has more excellent robust performance although a certain degree of fluctuation also occurs.
In conclusion, regardless of AGC load tracking performance, bed temperature interval control performance, or disturbance rejection and robustness, comparison of simulation curves shows that the multivariable constraint predictive control algorithm based on the MHE algorithm provided by the patent shows excellent control performance, and has important significance for practical engineering application of AGC control of the circulating fluidized bed unit.
The foregoing is merely exemplary and illustrative of the principles of the present invention and various modifications, additions and substitutions of the specific embodiments described herein may be made by those skilled in the art without departing from the principles of the present invention or exceeding the scope of the claims set forth herein.
Claims (1)
1. A multivariable constraint predictive control method based on a circulating fluidized bed unit is characterized by comprising the following steps:
s1: determining an AGC operation condition interval of a control object, and dividing working condition points;
and aiming at the characteristics of a specific circulating fluidized bed unit control object, determining an AGC operation interval, and taking:
Θ＝{Ne|Ne∈[40％ECR,110％ECR]} (1)
in the formula (1), theta is an AGC operation interval, Ne is unit power, unit is MW, ECR is unit rated working condition, 40% ECR is low-load operation state, and 110% ECR is output super-power operation state;
in the interval range, 4 specific working condition points, namely 40% ECR, 60% ECR, 80% ECR and 100% ECR, are selected at intervals of 20% ECR;
s2: obtaining the step response coefficients of the control object under 4 working condition points;
performing a step response dynamic experiment on the actual power plant object at the 4 working condition points; taking the l-th working condition point as an example, if the object has m-dimensional control input and p-dimensional output, the obtained corresponding step response coefficient matrix Al,iAs shown in formula (2):
in the formula (2), m is the number of input quantities, p is the number of output quantities, N is the length of the model, and l is the number of working condition points;
s3: obtaining approximate expression forms of the state space of the control object under 4 working condition points;
similarly, taking the ith operating point as an example, the approximate expression form of the state space of the control object is shown in formula (3):
in the formula (3), Xl(k) A state variable of a control object of the first working point, y (k) is an output variable, delta u (k) is a control quantity increment, and Sl、AlAnd ClAre respectively shown as formulas (4), (5) and (6);
in the formula (5), Al,iAs shown in formula (2);
s4, acquiring a linear variable parameter model (L inner-Parameters-Varyingmodel, & -lTtT transformation = L "& -gTt L & -lTt/T & -gTt PV) consisting of sub-models of different working conditions;
the L PV model is shown in equation (7):
in the formula (7), the reaction mixture is,
C(w)＝[λ1(w)C1λ2(w)C2… λl(w)Cl](9)
in the formula (9), w is a scheduling variable of an L PV model, taken as a power parameter Ne of the circulating fluidized bed unit, λ is a weight function, and is a potential function of the scheduling variable w;
the state variable X (k) is obtained by a rolling time domain estimation method, and comprises the following steps:
s4.1: the state space description of the discrete-time linear time-invariant system considering the multiple-input multiple-output control object is shown as equation (25):
in the formula (25), v (k) is measurement noise, w (k) is system process noise, and is a system noise input matrix; the system process noise w (k) and the measurement noise v (k) are gaussian white noise sequences, and both have statistical properties shown in formula (26):
in the formula (26), Q0Is a non-negative definite matrix and represents a variance matrix of system process noise W (k); r0Is positive definite matrix, and represents variance matrix of measurement noise V (k), (k, j) is a Crohn's Nike symbol;
s4.2: acquiring input and output measurement data of the current time k, as shown in equation (27):
s4.3: calculating an optimal solution according to a constrained optimization problem, wherein the constrained optimization problem is represented by a formula (28):
in the formula (28), RMHE、QMHE、Π0Positive definite punishment matrixes respectively corresponding to the variables;
estimating initial state estimators by arrival cost function
And initial penalty matrix Πk-NWherein the arrival cost function is as shown in equation (29):
and pik-NIterative calculations based on the forward rica equation,
obtained based on kalman estimation;
the joint vertical type (25) -formula (29) calculates to obtain an optimal solution, as shown in formula (30):
answerop＝(X(k-N|k-1)op,W(N|k-1)op) (30)
in the formula (30), the reaction mixture,
W(N|k-1)op＝{W(k-N|k-1)op,W(k-N+1|k-1)op,…,W(k-1|k-1)op} (31)
s4.4: iterative computation is carried out by utilizing the optimal solution of the optimization problem to obtain a system state quantity estimated value;
simultaneous type (25), formula (26) and formula (30), calculating X (k | k-1)op；
S5: obtaining a prediction model of a prediction control algorithm;
the prediction model is shown in equation (11):
y(k)＝FX(k)+ΦΔu(k) (11)
in the formula (11), the reaction mixture is,
in the formula (12), yn(k +1| k) is an output predicted value at the time k +1 obtained by calculation based on data at the time k, wherein n is 1. In the formula (13), Δ uz(k) For control increment, z is 1, m is the number of input quantities; in the formula (14), P is a prediction time domain of the prediction control algorithm;
s6: performing rolling optimization of a predictive control algorithm, and calculating to obtain an optimal control quantity;
the control object is of a 3-input and 3-output structure, wherein 3 input quantities are respectively a coal feeding quantity B, and the unit is Kg/s; main steam adjusting door uTIn units of%; primary wind frequency modulation instruction uWIn units of%; the 3 output quantities are respectively the unit power Ne, and the unit is MW; main steam pressure ptIn MPa; the hearth bed temperature Tb is given in unit; according to the specific requirements of AGC control of the circulating fluidized bed unit, the power of the unit and the pressure of main steam are set values for control, and the bed temperature of the hearth is interval control, so that the performance index established at the moment k is as shown in a formula (16):
in the formula (16), J (k) is a performance index function, w1,2(k) Is set value of unit power and main steam pressure, y1,2(k) The actual output values of the unit power and the main steam pressure, delta u (k) is the increment of the control quantity, q1,q2,q3Is an error weight coefficient, r1,r2,r3To controlA weighting coefficient;P(k) the deviation of the bed temperature output from the set value is shown in formula (17):
in the formula (17), the compound represented by the formula (I),
in formula (18), TbminAnd TbmaxUpper and lower limits for bed temperature interval control;
meanwhile, constraint conditions are set for the performance indexes, including control quantity and control quantity increment, as shown in formula (19):
in the formula (19), Bmin、Bmax，uT,min、uT,max，uW,min、uW,maxRespectively, control quantity B, uT、uWMinimum and maximum values of; dBmin、dBmax，duT,min、duT,max，duW,min、duW,maxRespectively control quantity increment dB and duT、duWMinimum and maximum values of;
based on the performance index and the constraint condition, calculating to obtain the corresponding optimal control increment delta uM(k) The corresponding instantaneous control increment Δ u (k) is:
Δu(k)＝LΔuM(k) (20)
in the formula (20), the reaction mixture is,
in the formula (21), Δ uz,M(k) As shown in formula (22), z is 1.
In the formula (23), the compound represented by the formula,
L0＝[1 0 … 0](1×M)(24)
in the formula (19) -formula (24), M is a control time domain of the predictive control;
thereby obtaining Δ u (k), and calculating the value of u (k) by u (k) ═ u (k-1) + Δ u (k);
s7: performing feedback correction of a predictive control algorithm;
the control object of the circulating fluidized bed unit is represented by a nonlinear mathematical model constructed based on a mechanism;
in the formula (32), BC(t) the carbon deposition amount in the hearth at the moment t, Kg; qi(t) total heat released by fuel combustion at time t, MJ; qfMJ/Kg is the unit calorific value of the fuel; qWPrimary wind heat capacity, MJ; p is a radical ofdIs the drum pressure, MPa.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710579652.9A CN107270283B (en) | 2017-07-17 | 2017-07-17 | Multivariable constraint predictive control method based on circulating fluidized bed unit |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710579652.9A CN107270283B (en) | 2017-07-17 | 2017-07-17 | Multivariable constraint predictive control method based on circulating fluidized bed unit |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107270283A CN107270283A (en) | 2017-10-20 |
| CN107270283B true CN107270283B (en) | 2020-08-04 |
Family
ID=60072214
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710579652.9A Active CN107270283B (en) | 2017-07-17 | 2017-07-17 | Multivariable constraint predictive control method based on circulating fluidized bed unit |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107270283B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108021027B (en) * | 2017-11-21 | 2021-02-02 | 华北电力大学 | Output power prediction system and method for supercritical circulating fluidized bed unit |
| CN110094724A (en) * | 2019-05-27 | 2019-08-06 | 淮南矿业(集团)有限责任公司 | A kind of burning in circulating fluid bed boiler control method and device |
| CN112097253B (en) * | 2020-08-27 | 2022-06-03 | 中节能(肥城)生物质能热电有限公司 | Method for improving combustion efficiency of circulating fluidized bed boiler |
| CN113267994B (en) * | 2021-04-23 | 2023-05-05 | 湖南省湘电试验研究院有限公司 | Main steam pressure control method and system of thermal power generating unit based on three-stage control series connection |
| CN113835342B (en) * | 2021-09-18 | 2024-04-16 | 国网河北能源技术服务有限公司 | Disturbance rejection predictive control method for overheat steam temperature system |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH11218303A (en) * | 1998-02-03 | 1999-08-10 | Ebara Corp | Pressurized fluidized bed boiler |
| CN101551103B (en) * | 2009-04-30 | 2010-11-10 | 东莞德永佳纺织制衣有限公司 | Automatic boiler combustion control system of circulating fluid bed |
| CN101788809B (en) * | 2009-08-17 | 2013-03-06 | 杭州和利时自动化有限公司 | Coordinated control system (CCS) of large-size circulating fluidized bed boiler (CFBB) unit |
| CN102183015B (en) * | 2011-04-01 | 2012-10-31 | 北京三博中自科技有限公司 | Combustion optimizing control system of circulating fluidized bed boiler in wide-range change of load |
| CN102401371B (en) * | 2011-12-06 | 2014-07-09 | 东南大学 | Reheated gas temperature optimization control method based on multi-variable predictive control |
| CN104214772B (en) * | 2014-07-16 | 2016-06-22 | 山西大学 | A kind of control method of Properties of CFB AGC load instruction response |
| CN104296131B (en) * | 2014-10-23 | 2015-09-30 | 东南大学 | A kind of multivariable cooperative control method of twin furnace Properties of CFB |
| CN104656448B (en) * | 2015-01-16 | 2017-09-26 | 东南大学 | A kind of supercritical unit forecast Control Algorithm based on decoupling and disturbance observation |
| CN104698842B (en) * | 2015-02-06 | 2017-11-24 | 浙江大学 | A kind of LPV model nonlinear forecast Control Algorithms based on interior point method |
| CN105889910B (en) * | 2016-05-04 | 2017-11-03 | 东南大学 | A kind of new A GC control methods of CFBB |
| CN106705034B (en) * | 2016-08-24 | 2019-03-01 | 华北电力大学 | A kind of supercritical circulating fluidized bed boiler unit quick load change control method |
| CN106524142B (en) * | 2016-12-22 | 2019-06-04 | 河北省电力建设调整试验所 | A method of improving Properties of CFB AGC responding ability |
-
2017
- 2017-07-17 CN CN201710579652.9A patent/CN107270283B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN107270283A (en) | 2017-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107270283B (en) | Multivariable constraint predictive control method based on circulating fluidized bed unit | |
| CN111443681B (en) | Multi-model predictive control design method for supercritical thermal power generating unit coordinated control system | |
| CN107479389B (en) | Fuzzy self-adaptive PID control method for thermal power generating unit superheated steam temperature prediction | |
| CN102494336B (en) | Combustion process multivariable control method for CFBB (circulating fluidized bed boiler) | |
| CN102401371B (en) | Reheated gas temperature optimization control method based on multi-variable predictive control | |
| CN102841539B (en) | Based on the subcritical control method for coordinating of multi-model PREDICTIVE CONTROL | |
| CN102841540A (en) | MMPC-based supercritical unit coordination and control method | |
| CN110824926B (en) | Thermal power generating unit deep peak shaving primary frequency modulation control method | |
| CN102998974A (en) | Multi-model generalized predictive control system and performance evaluation method thereof | |
| CN106707756B (en) | The supercritical thermal power unit boiler-turbine coordinated control method of fusion expansion observer | |
| Zhang et al. | Zone economic model predictive control of a coal-fired boiler-turbine generating system | |
| Prasad et al. | Plant-wide predictive control for a thermal power plant based on a physical plant model | |
| CN103322553A (en) | Multi-model disturbance estimation predictive-control method for superheated steam temperature of thermal power generating unit | |
| CN105180136A (en) | Thermal-power-plant boiler main steam temperature control method based on fractional order proportional-integral (PI) dynamic matrix | |
| CN105700357B (en) | Method of Boiler Combustion Control System based on multivariable PID-PFC | |
| CN110579968A (en) | Prediction control strategy for ultra-supercritical unit depth peak regulation coordination system | |
| CN106855691B (en) | The double-deck control system for supercritical thermal power unit machine furnace system Steam Generator in Load Follow | |
| Darwish et al. | Advanced boiler control system for steam power plants using modern control techniques | |
| CN108762086B (en) | Secondary reheat steam temperature control device and control system based on model predictive control | |
| CN113835342A (en) | Disturbance rejection prediction control method of superheated steam temperature system | |
| Esmaeili et al. | Robust & nonlinear control of an ultra-supercritical coal fired once-through boiler-turbine unit in order to optimize the uncertain problem | |
| CN107783941B (en) | Solar energy compensation type electric boiler heat supply control method based on valley electricity price change | |
| CN116047897A (en) | Gas turbine predictive control method based on parameter self-adaptive disturbance rejection controller | |
| CN113282043A (en) | Multivariable state space model-based ultra-supercritical unit coordination control method | |
| CN107045287A (en) | Coordinated Control Systems and control method based on Prediction and Control Technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: No. 56 Hefei River avenue of 230051 cities in Anhui Province Applicant after: Anhui Electrical Engineering Professional Technique College Applicant after: National network Anhui Electric Power Co., Ltd Training Center Applicant after: State Grid Corporation of China Address before: No. 56 Hefei River avenue of 230051 cities in Anhui Province Applicant before: Anhui Electrical Engineering Professional Technique College Applicant before: State Grid Anhui Electric Power Company Training Center Applicant before: State Grid Corporation of China |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |