CN118011805A - Ultra-supercritical unit model predictive control method based on data driving and Tube optimization - Google Patents
Ultra-supercritical unit model predictive control method based on data driving and Tube optimization Download PDFInfo
- Publication number
- CN118011805A CN118011805A CN202410104106.XA CN202410104106A CN118011805A CN 118011805 A CN118011805 A CN 118011805A CN 202410104106 A CN202410104106 A CN 202410104106A CN 118011805 A CN118011805 A CN 118011805A
- Authority
- CN
- China
- Prior art keywords
- variable
- ultra
- control
- tube
- optimization
- 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.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000013528 artificial neural network Methods 0.000 claims abstract description 29
- 239000002245 particle Substances 0.000 claims abstract description 14
- 230000008569 process Effects 0.000 claims abstract description 12
- 238000005096 rolling process Methods 0.000 claims abstract description 9
- 239000011159 matrix material Substances 0.000 claims description 39
- 230000006870 function Effects 0.000 claims description 24
- 238000012545 processing Methods 0.000 claims description 12
- 230000003247 decreasing effect Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000000926 separation method Methods 0.000 claims description 6
- 239000003245 coal Substances 0.000 claims description 4
- 238000012937 correction Methods 0.000 claims description 3
- 239000013641 positive control Substances 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 9
- 230000007246 mechanism Effects 0.000 abstract description 5
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005299 abrasion Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
A method for controlling the model prediction of ultra-supercritical machine set based on data drive and Tube optimization includes such steps as evaluating the thermal states of ultra-supercritical machine set, choosing the proper dynamic model of thermal object, and building 3X 3 mechanism model of machine set. The actual operation data of the power plant is identified through a particle swarm-neural network (PSO-BP) algorithm, and a system model is obtained; then, based on Rigid Tube robust optimization model predictive control algorithm, the system state is controlled in a constraint subset called Tube through a control law, and a new cost function is obtained based on Tube pipeline set; and finally, obtaining the minimum value of the cost function by adopting a neural network algorithm, and simultaneously, adjusting three input parameters in time in the rolling optimization process to ensure the energy flow supply and demand balance of each link, so that the unit can stably output, and the efficient and stable control effect is achieved. The invention ensures that the system has better robustness and better realizes model predictive control.
Description
Technical Field
The invention belongs to the field of ultra-supercritical thermal generator set model predictive control, and particularly relates to an ultra-supercritical generator set model predictive control method based on data driving and Tube optimization.
Background
Thermal automation of thermal power generating units is an important technical measure for guaranteeing equipment safety and improving unit economy. Because the ultra-supercritical unit system has complex structure, the thermal process has the characteristics of nonlinearity, strong coupling, uncertainty and the like, and a lot of difficulties are brought to control; the complex environments such as large-scale space field, high temperature, abrasion, vibration, corrosion and the like bring great difficulty to signal measurement; the domestic unit faces complex conditions such as coal fluctuation, frequent requirement of a power grid for large-scale rapid load adjustment, multiple problems of over-temperature and over-pressure of the ultra-supercritical unit, heavy energy-saving and environment-friendly tasks and the like, and the conventional PID+feedforward controller has limited effect in the control of the thermal process. Meanwhile, in practical application, the system is inevitably disturbed by uncertainty, so that the system state deviates from the nominal track.
Therefore, by adopting a reasonable control method, under the condition of being simplified as much as possible, the input parameters can be quickly and reasonably adjusted, the response speed of the system is improved, meanwhile, the condition that related disturbance can be restrained under the condition of multiple loads is met, the regulation and stabilization are within the standard required output range, the efficiency of the power production process is further improved, and the reasonable intelligent control of the ultra-supercritical unit is realized.
Disclosure of Invention
In order to overcome the defects of the prior art and improve the performance of the power production process of the ultra-supercritical unit and achieve a more rapid and accurate control effect, the invention provides a ultra-supercritical unit model prediction control method based on data driving and Tube optimization, which comprises the steps of firstly dividing an operation working condition into a plurality of working condition points, and selecting model classes corresponding to dynamic characteristics of a system according to different working conditions corresponding to input and output so as to adapt to corresponding thermal objects; then, an improved particle swarm optimization neural network algorithm (PSO-BPNN) is adopted to identify a large number of data parameters of the power plant, and a high-precision concrete model suitable for three-input and three-output of the ultra-supercritical unit is obtained; finally, according to the related parameter requirements of the ultra-supercritical unit, reasonable pipeline (Tube) parameters are designed, so that the system track is converged to an expected value along a set reference track, and model prediction control of the unit is realized; according to the invention, under a high-precision unit three-input three-output model, the system has better robustness, model prediction control is better realized, unit input/output adjustment is more accurate and faster to reach a set standard range, and the system has better anti-interference capability.
In order to solve the technical problems, the invention provides the following technical scheme:
A method for controlling ultra-supercritical unit model prediction based on data driving and Tube optimization comprises the following steps:
S1: acquiring historical operation data of the input and output of the ultra-supercritical unit, unifying the dimensions of the input and output data of the coordination control system, and performing zero initial value processing on all modeling data; then selecting a proper dynamic model of the thermal object, and carrying out model identification on the system by adopting an improved neural network algorithm based on particle swarm optimization;
S2: the control requirement of three-input and three-output parameters of the ultra-supercritical unit is used as constraint, a Tube set X is designed, and the state X of the system is controlled in the set X through a control law, so that a new cost function J is obtained;
s3: and taking the minimum objective value function value as an optimization target, adopting a neural network algorithm to perform optimizing solution, and performing rolling optimization and correction on the system according to a set track to realize Model Predictive Control (MPC).
Further, in the step S1, the model identification is performed on the system by adopting an improved back propagation neural network algorithm based on particle swarm optimization, which comprises the following sub-steps:
S1-1, acquiring historical operation data of an input variable u (t) and an output variable y (t) of the ultra-supercritical unit, and performing zero initial value processing on the data, wherein a zero initial value calculation formula is as follows:
Wherein: e represents the number of initial points estimated by the system, u * (k) is an input variable after zero initial value processing, and y * (k) is an output variable after zero initial value processing;
The three-input three-output model of the ultra-supercritical unit is constructed as follows:
Wherein: n represents the unit load, P represents the main steam pressure, and T represents the intermediate point temperature; b represents the coal feeding amount, mu represents the opening of a main valve, and W represents the water feeding amount; g is a function matrix corresponding to input and output;
s1-2, searching the space dimension M of the neural network in the whole space by adopting a particle swarm algorithm PSO with linearly decreasing weight, wherein the weight decreasing calculation formula is as follows:
ωmin＝0.4,ωmax＝0.9 (4)
Wherein: omega is a weight value, omega min is a minimum weight value, omega max is a maximum weight value, and t is the current iteration number;
s1-3, designing a neural network topological structure, determining the number n=4 of nodes of an input layer, the number l=20 of nodes of a hidden layer and the number m=4 of nodes of an output layer according to an input-output sequence of the ultra-supercritical unit, initializing an initial weight and a threshold of the neural network by using a weight decreasing particle swarm optimization result, and iteratively updating the neural network at a learning rate c=0.3.
Still further, in the step S2, a discrete state space equation of the system is constructed according to the result of identifying the black box by the system, based on the general form of solving the optimization control problem of Tube MPC, the optimization algorithm separates the system determining part and the uncertainty part by adopting the separation control law based on Rigid Tube to obtain the determined nominal system part and the uncertainty part, and the pipeline set X k+j is determined by solving the positive robust invariant set to obtain the new cost function J, which includes the following substeps:
S2-1. MPC optimization control based on Tube is as follows:
Wherein: j is a time variable, T is a prediction step size, l s(Xk+j) is a penalty for the shape of the set X k+j, l (v) is a penalty for the decision variable v, l f(Xk+N) is a terminal penalty;
By recursion Ensuring that the system state is to be in the corresponding Tube;
S2-2, adopt separation control law Separating the system determination part from the uncertainty part to obtain a determined nominal system part:
Wherein: u is the feedback input of the system, K is the state feedback gain matrix, the current state is represented by the state x k at time K, For a part of state variables of a nominal system, A is a state cost weight matrix, and H is a decision cost weight matrix;
For a formal system, only the nominal system state needs to be ensured to be controlled, and the uncertain part is not considered;
S2-3 the Tube form used in the invention is:
Uk+j＝vk+j+KSr (8)
wherein: u k+j is a feedback input variable set, S r is a positive robust invariant set;
thus, the control optimization problem of MPC translates into:
wherein: q is a state variable weight matrix, R is a control variable weight matrix,
Furthermore, in the step S3, the objective function value is the minimum as the optimization target, and the objective function is optimized and solved based on the model predictive control algorithm to obtain the final result, which includes the following sub-steps:
the cost function of the tube model predictive control is as follows:
to simplify the solution process, the weight matrices R and Q are designed as symmetric positive definite matrices:
Rxi＝rxiI (11)
Qv＝qvI (12)
Wherein: i is an identity matrix, R xi is a weight coefficient of a state variable, Q v is a weight coefficient of a control variable, R xi is a symmetric positive control variable weight matrix, and Q v is a symmetric positive state variable weight matrix;
s3-2, combining model parameters of the ultra-supercritical unit data identification to obtain a linearization average state equation at a certain moment, wherein the linearization average state equation is as follows:
x(k+1)＝Apx(k)+Bpu(k)+Fpd(k) (13)
Wherein: a p is a system variable weight matrix, B p is a control input weight matrix, F p is an external interference input matrix, p is a prediction step length, and d (k) is external measurable interference;
The predictive equation is derived from equation (13) as:
s3-3, for solving the optimal solution, an improved neural network algorithm is adopted, and a dynamic differential equation required by the algorithm is established:
Wherein: Is a proportionality constant,/> Affecting the speed and accuracy of the solving process; z is an auxiliary variable for solving the target variable X (k); p x is an auxiliary function for satisfying system constraints; w, q is the identity matrix of the corresponding dimension;
Rolling iteration is carried out by applying a differential equation (15), and as the iteration times increase, the iteration result gradually converges to the optimal solution.
The beneficial effects of the invention are mainly shown in the following steps: according to the method, firstly, through evaluating each thermodynamic state of the ultra-supercritical unit system, a corresponding proper thermodynamic state equation is selected, and a unit 3 multiplied by 3 mechanism model is built. The actual operation data of the power plant is identified through an improved PSO-BPNN algorithm, and a system model is obtained; and the algorithm improves the model identification precision under the condition of combining the mechanism model. Then, based on Rigid Tube's robust optimization model predictive control method, the system state is controlled in a set called Tube by control law, the set is a subset of constraints, and the whole Tube set is guided to a desired position. And finally, optimizing by adopting a neural network algorithm to obtain an optimal solution, and timely feeding back and correcting in the rolling optimization process to achieve a control effect. Experimental results show that the method can obtain a higher-precision ultra-supercritical unit model, can rapidly realize predictive control on the system, and the control efficiency and the anti-interference capability of the system are obviously improved.
Drawings
FIG. 1 is a flow chart of a method for predictive control of a model of a ultra-supercritical unit based on data driving and Tube optimization.
Fig. 2 is a graph of the effect of the robust optimization model predictive control of the unit load N based on Rigid Tube.
Fig. 3 is a graph of the effect of the main steam pressure P on the basis of Rigid Tube on the predictive control of the robust optimization model.
Fig. 4 is a graph of the effect of the robust optimization model predictive control of the intermediate point temperature T based on Rigid Tube.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1 to 4, a method for controlling model prediction of a supercritical unit based on data driving and Tube optimization includes the steps of:
S1: acquiring historical operation data of the input and output of the ultra-supercritical unit, unifying the dimensions of the input and output data of the coordination control system, and performing zero initial value processing on all modeling data; then selecting a proper dynamic model of the thermal object, and carrying out model identification on the system by adopting an improved neural network algorithm PSO-BPNN based on particle swarm optimization;
in the step S1, the model identification is performed on the system by adopting an improved back propagation neural network algorithm based on particle swarm optimization, and the method comprises the following sub-steps:
S1-1, acquiring historical operation data of an input variable u (t) and an output variable y (t) of the ultra-supercritical unit, and performing zero initial value processing on the data, wherein a zero initial value calculation formula is as follows:
Wherein: e represents the number of initial points estimated by the system, u * (k) is an input variable after zero initial value processing, and y * (k) is an output variable after zero initial value processing;
The three-input three-output model of the ultra-supercritical unit is constructed as follows:
Wherein: n represents the unit load, P represents the main steam pressure, and T represents the intermediate point temperature; b represents the coal feeding amount, mu represents the opening of a main valve, and W represents the water feeding amount; g is a function matrix corresponding to input and output;
S1-2, searching the space dimension M of the neural network in the whole space by adopting a particle swarm algorithm (PSO) with linearly decreasing weight, wherein the weight decreasing calculation formula is as follows:
ωmin＝0.4,ωmax＝0.9 (4)
Wherein: omega is a weight value, omega min is a minimum weight value, omega max is a maximum weight value, and t is the current iteration number;
S1-3, designing a neural network topological structure, determining the number n=4 of nodes of an input layer, the number l=20 of nodes of a hidden layer and the number m=4 of nodes of an output layer according to an input-output sequence of an ultra-supercritical unit, initializing an initial weight and a threshold value of the neural network by using a weight decreasing particle swarm optimization result, and iteratively updating the neural network at a learning rate c=0.3;
s2: taking the three-input three-output parameter control requirement of the ultra-supercritical unit as constraint, designing a pipeline set X, and controlling the state X of the system in the set X (Tube) through a control law to obtain a new cost function J;
In the step S2, a discrete state space equation of the system is constructed according to the result of identifying the black box by the system, based on the general form of solving the optimization control problem of Tube MPC, the optimization algorithm separates the system determining part from the uncertainty part by adopting the separation control law based on Rigid Tube to obtain the determined nominal system part, and the pipeline set X k+j is determined by solving the positive robust invariant set to obtain the new cost function J, which comprises the following substeps:
S2-1. MPC optimization control based on Tube is as follows:
Wherein: j is a time variable, T is a prediction step size, l s(Xk+j) is a penalty for the shape of the set X k+j, l (v) is a penalty for the decision variable v, l f(Xk+N) is a terminal penalty;
By recursion Ensuring that the system state is to be in the corresponding Tube;
S2-2, adopt separation control law Separating the system determination part from the uncertainty part to obtain a determined nominal system part:
Wherein: u is the feedback input of the system, K is the state feedback gain matrix, the current state is represented by the state x k at time K, For a part of state variables of a nominal system, A is a state cost weight matrix, and H is a decision cost weight matrix;
For a formal system, only the nominal system state needs to be ensured to be controlled, and the uncertain part is not considered;
S2-3. Tube used is of the form:
Uk+j＝vk+j+KSr (8)
wherein: u k+j is a feedback input variable set, S r is a positive robust invariant set;
thus, the control optimization problem of MPC translates into:
wherein: q is a state variable weight matrix, R is a control variable weight matrix,
S3: and taking the minimum objective value as an optimization target, adopting a neural network algorithm to perform optimizing solution on the minimum objective value, and performing rolling optimization and correction on the system according to a set track to realize model prediction control.
In the step S3, with the minimum objective function value as an optimization target, the objective function is optimized and solved based on a model predictive control algorithm to obtain a final result, and the result is shown in fig. 2-4, and includes the following sub-steps:
the cost function of the tube model predictive control is as follows:
to simplify the solution process, the weight matrices R and Q are designed as symmetric positive definite matrices:
Rxi＝rxiI (11)
Qv＝qvI (12)
Wherein: i is an identity matrix, R xi is a weight coefficient of a state variable, Q v is a weight coefficient of a control variable, R xi is a symmetric positive control variable weight matrix, and Q v is a symmetric positive state variable weight matrix;
s3-2, combining model parameters of the ultra-supercritical unit data identification to obtain a linearization average state equation at a certain moment, wherein the linearization average state equation is as follows:
x(k+1)＝Apx(k)+Bpu(k)+Fpd(k) (13)
Wherein: a p is a system variable weight matrix, B p is a control input weight matrix, F p is an external interference input matrix, p is a prediction step length, and d (k) is external measurable interference;
The predictive equation is derived from equation (13) as:
s3-3, for solving the optimal solution, an improved neural network algorithm is adopted, and a dynamic differential equation required by the algorithm is established:
Wherein: Is a proportionality constant,/> Affecting the speed and accuracy of the solving process; z is an auxiliary variable for solving the target variable X (k); p x is an auxiliary function for satisfying system constraints; w, q is the identity matrix of the corresponding dimension.
Rolling iteration is carried out by applying a differential equation (15), and as the iteration times increase, the iteration result gradually converges to the optimal solution.
In order to enable a person skilled in the art to better understand the ultra-supercritical unit model prediction control process based on data driving and Tube optimization, the flow chart of fig. 1 is drawn, is visual and vivid, and is convenient to understand.
According to the method, firstly, through evaluation of each thermodynamic state of the ultra-supercritical unit system, a proper dynamic equation of a thermodynamic object is selected, and a unit 3 multiplied by 3 mechanism model is built. The actual operation data of the power plant is identified through an improved PSO-BPNN algorithm, and a system model is obtained; and the algorithm improves the model identification precision under the condition of combining the mechanism model. Then, by means of a Rigid Tube-based robust optimization model predictive control algorithm, the system state is controlled by means of a control law in a set of Tube, which is a subset of constraints, and the whole Tube set is guided to a desired location. And finally, obtaining the minimum value of the cost function by adopting a neural network algorithm, and simultaneously, adjusting three input parameters in time in the rolling optimization process to ensure the energy flow supply and demand balance of each link, so that the unit can stably output, and the efficient and stable control effect is achieved. The example research shows that the method can obtain a higher-precision ultra-supercritical unit model and realize the robust MPC control with better effect. Therefore, the method can realize the control of the operation of the unit, thereby further improving the efficiency and the stability of the operation of the system. In conclusion, the ultra-supercritical unit model predictive control algorithm based on data driving and Tube optimization provides a certain reference for intelligent control of the ultra-supercritical machine thermal power generation system.
The embodiments described in this specification are merely illustrative of the manner in which the inventive concepts may be implemented. The scope of the present invention should not be construed as being limited to the specific forms set forth in the embodiments, but the scope of the present invention and the equivalents thereof as would occur to one skilled in the art based on the inventive concept.
Claims (4)
1. The ultra-supercritical unit model prediction control method based on data driving and Tube optimization is characterized by comprising the following steps of:
S1: acquiring historical operation data of the input and output of the ultra-supercritical unit, unifying the dimensions of the input and output data of the coordination control system, and performing zero initial value processing on all modeling data; then selecting a proper dynamic model of the thermal object, and carrying out model identification on the system by adopting an improved neural network algorithm based on particle swarm optimization;
S2: the control requirement of three-input and three-output parameters of the ultra-supercritical unit is used as constraint, a Tube set X is designed, and the state X of the system is controlled in the set X through a control law, so that a new cost function J is obtained;
s3: and taking the minimum objective value function value as an optimization target, adopting a neural network algorithm to perform optimizing solution, and performing rolling optimization and correction on the system according to a set track to realize Model Predictive Control (MPC).
2. The method for model predictive control of a data-driven and Tube-optimized ultra-supercritical unit according to claim 1, wherein in the step S1, the model recognition of the system by using the improved neural network algorithm based on particle swarm optimization comprises the following steps:
S1-1, acquiring historical operation data of an input variable u (t) and an output variable y (t) of the ultra-supercritical unit, and performing zero initial value processing on the data, wherein a zero initial value calculation formula is as follows:
Wherein: e represents the number of initial points estimated by the system, u * (k) is an input variable after zero initial value processing, and y * (k) is an output variable after zero initial value processing;
The three-input three-output model of the ultra-supercritical unit is constructed as follows:
Wherein: n represents the unit load, P represents the main steam pressure, and T represents the intermediate point temperature; b represents the coal feeding amount, mu represents the opening of a main valve, and W represents the water feeding amount; g is a function matrix corresponding to input and output;
S1-2, searching the space dimension M of the neural network in the whole space by adopting a particle swarm algorithm (PSO) with linearly decreasing weight, wherein the weight decreasing calculation formula is as follows:
ω min＝0.4,ωmax =0.9 (4) formula: omega is a weight value, omega min is a minimum weight value, omega max is a maximum weight value, and t is the current iteration number;
s1-3, designing a neural network topological structure, determining the number n=4 of nodes of an input layer, the number l=20 of nodes of a hidden layer and the number m=4 of nodes of an output layer according to an input-output sequence of the ultra-supercritical unit, initializing an initial weight and a threshold of the neural network by using a weight decreasing particle swarm optimization result, and iteratively updating the neural network at a learning rate c=0.3.
3. The method for controlling model prediction of ultra-supercritical unit based on data driving and Tube optimization according to claim 2, wherein in step S2, a discrete state space equation of the system is constructed according to the result of system identification, a separation control law is adopted to separate a system determining part and an uncertainty part, a determined nominal system part is obtained, a pipeline set X k+j is determined by solving a positive robust invariant set, and a new cost function J is obtained, and the method comprises the following steps:
S2-1. MPC optimization control based on Tube is as follows:
Wherein: j is a time variable, T is a prediction step size, l s(Xk+j) is a penalty for the shape of the set X k+j, l (v) is a penalty for the decision variable v, l f(Xk+N) is a terminal penalty;
By recursion Ensuring that the system state is to be in the corresponding Tube;
S2-2, adopt separation control law Separating the system determination part from the uncertainty part to obtain a determined nominal system part:
Wherein: u is the feedback input of the system, K is the state feedback gain matrix, the current state is represented by the state x k at time K, For a part of state variables of a nominal system, A is a state cost weight matrix, and H is a decision cost weight matrix;
For a formal system, only the nominal system state needs to be ensured to be controlled, and the uncertain part is not considered;
S2-3. Tube used is of the form:
Uk+j＝vk+j+KSr (8)
wherein: u k+j is a feedback input variable set, S r is a positive robust invariant set;
thus, the control optimization problem of MPC translates into:
wherein: q is a state variable weight matrix, R is a control variable weight matrix,
4. The method for model predictive control of a supercritical unit based on data driving and Tube optimization according to claim 3, wherein in the step S3, the objective function value is used as an optimization target, and the objective function is optimized and solved based on a model predictive control algorithm to obtain a final result, comprising the following steps:
the cost function of the tube model predictive control is as follows:
to simplify the solution process, the weight matrices R and Q are designed as symmetric positive definite matrices:
Rxi＝rxiI (11)
Qv＝qvI (12)
Wherein: i is an identity matrix, R xi is a weight coefficient of a state variable, Q v is a weight coefficient of a control variable, R xi is a symmetric positive control variable weight matrix, and Q v is a symmetric positive state variable weight matrix;
s3-2, combining model parameters of the ultra-supercritical unit data identification to obtain a linearization average state equation at a certain moment, wherein the linearization average state equation is as follows:
x(k+1)＝Apx(k)+Bpu(k)+Fpd(k) (13)
Wherein: a p is a system variable weight matrix, B p is a control input weight matrix, F p is an external interference input matrix, p is a prediction step length, and d (k) is external measurable interference;
The predictive equation is derived from equation (13) as:
s3-3, for solving the optimal solution, an improved neural network algorithm is adopted, and a dynamic differential equation required by the algorithm is established:
Wherein: Is a proportionality constant,/> Affecting the speed and accuracy of the solving process; z is an auxiliary variable for solving the target variable X (k); p x is an auxiliary function for satisfying system constraints; w, q is a coefficient matrix of the corresponding dimension;
Rolling iteration is carried out by applying a differential equation (15), and as the iteration times increase, the iteration result gradually converges to the optimal solution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410104106.XA CN118011805A (en) | 2024-01-24 | 2024-01-24 | Ultra-supercritical unit model predictive control method based on data driving and Tube optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410104106.XA CN118011805A (en) | 2024-01-24 | 2024-01-24 | Ultra-supercritical unit model predictive control method based on data driving and Tube optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118011805A true CN118011805A (en) | 2024-05-10 |
Family
ID=90954521
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410104106.XA Pending CN118011805A (en) | 2024-01-24 | 2024-01-24 | Ultra-supercritical unit model predictive control method based on data driving and Tube optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118011805A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118211432A (en) * | 2024-05-21 | 2024-06-18 | 广能亿能(北京)核能科技有限公司 | Simulation modeling method for overload work of semi-superconducting ultrahigh energy-saving motor |
-
2024
- 2024-01-24 CN CN202410104106.XA patent/CN118011805A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118211432A (en) * | 2024-05-21 | 2024-06-18 | 广能亿能(北京)核能科技有限公司 | Simulation modeling method for overload work of semi-superconducting ultrahigh energy-saving motor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Robust distributed MPC for load frequency control of uncertain power systems | |
| CN118011805A (en) | Ultra-supercritical unit model predictive control method based on data driving and Tube optimization | |
| Sayed et al. | Gain tuning PI controllers for boiler turbine unit using a new hybrid jump PSO | |
| CN108490790A (en) | A kind of overheating steam temperature active disturbance rejection cascade control method based on multiple-objection optimization | |
| CN108167802B (en) | Multi-model intelligent optimizing and predicting control method for boiler load under low load | |
| Dong et al. | Multilayer perception based reinforcement learning supervisory control of energy systems with application to a nuclear steam supply system | |
| CN115327890B (en) | Method for optimizing main steam pressure of PID control thermal power depth peak shaving unit by improved crowd searching algorithm | |
| KR101576004B1 (en) | Boiler-Turbine coordinated control method and apparatus using Dynamic matrix control in thermal power plant | |
| CN113189871B (en) | Thermal power generating unit flexible operation-oriented coordination control strategy | |
| Dai et al. | Multi-rate layered operational optimal control for large-scale industrial processes | |
| CN114137829B (en) | Proton exchange membrane fuel cell subspace identification method based on ALMBO optimization algorithm | |
| Xu et al. | Design of type-2 fuzzy fractional-order proportional-integral-derivative controller and multi-objective parameter optimization under load reduction condition of the pumped storage unit | |
| CN115421390A (en) | Multi-working-condition self-adaptive control method for combined heat and power generation unit considering deep reinforcement learning | |
| Wang et al. | Model-based predictive control for spatially-distributed systems using dimensional reduction models | |
| Subrata et al. | Computational and intelligent optimization tuning method for Pid controller | |
| CN113110628A (en) | Water level control method of pressurized water reactor deaerator based on PSO | |
| CN113282043A (en) | Multivariable state space model-based ultra-supercritical unit coordination control method | |
| Ma et al. | ANN and PSO based intelligent model predictive optimal control for large-scale supercritical power unit | |
| Abd Mohammed et al. | Controlling water level by using modified model free adaptive controller | |
| Feng et al. | Improved Linear Active Disturbance Rejection Control Strategy Based on RBF Neural Network For Main Steam Temperature Of Thermal Power Boiler | |
| Suganya et al. | Model reference adaptive controller using MOPSO for a non-Linear boiler-turbine | |
| CN114114909B (en) | Intermittent process 2D output feedback prediction control method based on particle swarm optimization | |
| Jiang et al. | Research on optimization of hydraulic turbine governor based on PSO algorithm | |
| CN113467237B (en) | Dynamic modeling method of main steam temperature based on deep learning | |
| CN117140504B (en) | N-link mechanical arm control method based on incremental model predictive control |
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 |