A method and system for dynamically adjusting ammonia to coal ratio

Abstract

The present application relates to mixed combustion power generation technology field, especially to a kind of ammonia coal ratio dynamic regulation method and system, method includes: collecting including boiler load, furnace temperature, primary air volume, NO x Tail gas concentration, the signal of the amount of ammonia and the amount of coal; based on the working condition signal to establish a discrete-time state-space model, and the model parameter is vectorized; using a hybrid algorithm based on particle swarm and recursive least squares to identify the vectorized model parameters online; according to the current model parameters identified, the preset prediction step and the control step, the model predictive control objective function is constructed, and the optimal future control increment sequence is obtained by solving online through particle swarm algorithm; the optimal control increment sequence is used to dynamically regulate the ammonia coal ratio. In the present application, the model can automatically track the working condition change, and has strong self-adaptability; under the premise of meeting the emission index, the minimum ammonia injection amount is realized, the ammonia water is saved, and the cost is reduced.

Description

Technical Field

[0001] This invention relates to the field of hybrid combustion power generation technology, and in particular to a method and system for dynamic control of the ammonia-to-coal ratio. Background Technology

[0002] Although the installed capacity of renewable energy has surpassed that of coal-fired power, the availability hours of new energy sources such as solar and wind power are far lower than those of coal-fired power, resulting in a certain gap between renewable energy generation and coal-fired power generation. In the short term, energy security still requires the guarantee of coal-fired power. In the long term, coal-fired power is transforming from the main source of electricity to a basic and system-regulating power source that provides reliable capacity, peak shaving, frequency regulation, and other auxiliary functions.

[0003] Modern coal-fired power plants aim to reduce NOx x Ammonia injection denitrification technology is commonly used for emissions control, and its denitrification efficiency is highly dependent on the ratio of ammonia to coal (i.e., the ammonia-to-coal ratio). However, the traditional ammonia-to-coal ratio is generally set to a fixed value or adjusted based on experience, which makes it difficult to cope with dynamic disturbances such as load fluctuations and changes in fuel calorific value, and can easily lead to a decrease in denitrification efficiency or ammonia escape.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the general background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method and system for dynamic control of the ammonia-to-coal ratio, thereby effectively solving the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for dynamic control of the ammonia-to-coal ratio, comprising the following steps:

[0007] Data collected includes boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate;

[0008] A discrete-time state-space model is established based on the aforementioned operating condition signal, and the model parameters are vectorized.

[0009] A hybrid algorithm based on particle swarm optimization and recursive least squares is used to identify the vectorized model parameters online.

[0010] Based on the identified current model parameters, preset prediction step size and control step size, a model prediction control objective function is constructed, and the optimal future control increment sequence is obtained by solving the particle swarm algorithm online.

[0011] The optimal control increment sequence is used to dynamically adjust the ammonia-coal ratio.

[0012] Furthermore, the discrete-time state-space model is as follows:

[0013]

[0014] In the formula, x is the state vector x = [T f C NOx ] T T f C represents the furnace temperature. NOx NO x The exhaust gas concentration, u is the controlled variable, i.e., the ammonia-to-coal ratio u = a / m, where a is the ammonia input and m is the coal feed rate; d is the disturbance variable, d = [Q, F] air ] T Q is the boiler load, F air Let y represent the primary air volume, A, B, C, and E be the model parameter matrices for the state variable, control variable, output variable, and disturbance variable, respectively, and y be the predicted NO. x Exhaust gas concentration, k is the kth sampling time, and the superscript T indicates the transpose of a matrix or vector, which transforms a column vector into a row vector.

[0015] Furthermore, the vectorization of model parameters includes:

[0016] Expand matrices A, B, C, and E into one-dimensional vectors to obtain the vectorized matrix θ:

[0017] θ=[vec(A) T ,vec(B) T ,vec(C) T ,vec(E) T ] T ∈R d ;

[0018] Where d = n 2 +nm+pn+nq, where n is the dimension of the state variables, m is the dimension of the control variables, p is the dimension of the output variables, and q is the dimension of the perturbation variables.

[0019] Furthermore, the online identification of the vectorized model parameters using a hybrid algorithm based on particle swarm optimization and recursive least squares includes the following steps:

[0020] First, use the particle swarm optimization algorithm to search on historical data to obtain the initial estimated values ​​of the model parameters and the initial values ​​of the covariance matrix of the recursive least squares algorithm.

[0021] After collecting a new set of operating condition data, the model parameters and covariance matrix are updated according to the recursive least squares algorithm;

[0022] Every set number of sampling periods, using the working condition data from the most recent set number of sampling periods as input, and centering on the current model parameters, a short-time particle swarm optimization algorithm is run within a set range to correct the model parameters and reset the covariance matrix.

[0023] Furthermore, it also includes:

[0024] When a sudden change in operating conditions is detected or the model prediction error exceeds the threshold, a short-time particle swarm algorithm is immediately triggered to correct the model parameters.

[0025] Furthermore, the objective function is:

[0026]

[0027] In the formula, Np is the prediction step size, Nc is the control step size, and λ is the penalty factor for changes in the control quantity. For target NO x Exhaust gas concentration limits

[0028] The objective function also includes the following constraints:

[0029]

[0030] In the formula, Δu(k)=u(k)-u(k-1).

[0031] The present invention also includes a dynamic control system for the ammonia-to-coal ratio, using the method described above, wherein the system comprises:

[0032] The data acquisition unit is used to collect data including boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate;

[0033] The modeling unit is used to establish a discrete-time state-space model based on the operating condition signal and to vectorize the model parameters.

[0034] The parameter identification unit is used to identify the vectorized model parameters online using a hybrid algorithm based on particle swarm optimization and recursive least squares.

[0035] The solving unit is used to construct the model prediction control objective function based on the identified current model parameters, preset prediction step size and control step size, and solve it online using the particle swarm algorithm to obtain the optimal future control increment sequence;

[0036] The control unit is used to dynamically control the ammonia-coal ratio using the optimal control increment sequence.

[0037] The present invention also includes a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described above.

[0038] The present invention also includes a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described above.

[0039] The beneficial effects of this invention are as follows: by using a hybrid algorithm based on particle swarm optimization and recursive least squares to identify vectorized model parameters online, the model can automatically track changes in operating conditions and has strong adaptability; the particle swarm optimization algorithm has global search capability in nonlinear and multi-peak optimization, and can find the near-optimal ammonia-coal ratio under varying operating conditions, thereby minimizing the amount of ammonia added, saving ammonia water, and reducing costs while meeting emission standards. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart of the method of the present invention;

[0042] Figure 2 This is a schematic diagram of the system of the present invention;

[0043] Figure 3 This is a schematic diagram of the structure of the computer device of the present invention. Detailed Implementation

[0044] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0045] Example 1:

[0046] like Figure 1 As shown: A method for dynamic control of the ammonia-to-coal ratio includes the following steps:

[0047] Data collected includes boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate;

[0048] A discrete-time state-space model is established based on the operating condition signal, and the model parameters are vectorized.

[0049] A hybrid algorithm based on particle swarm optimization and recursive least squares is used to identify vectorized model parameters online.

[0050] Based on the identified current model parameters, preset prediction step size and control step size, a model prediction control objective function is constructed, and the optimal future control increment sequence is obtained by solving the particle swarm algorithm online.

[0051] The ammonia-to-coal ratio is dynamically adjusted using the optimal control increment sequence.

[0052] By employing a hybrid algorithm based on particle swarm optimization and recursive least squares to identify vectorized model parameters online, the model can automatically track changes in operating conditions and exhibits strong adaptability. The particle swarm optimization algorithm has global search capabilities in nonlinear and multi-peak optimization, and can find the near-optimal ammonia-coal ratio under varying operating conditions. Under the premise of meeting emission standards, it can minimize the amount of ammonia added, save ammonia water, and reduce costs.

[0053] In this embodiment, the discrete-time state-space model is as follows:

[0054]

[0055] In the formula, x is the state vector x = [T f C NOx ] T T f C represents the furnace temperature. NOx NO x The exhaust gas concentration, u is the controlled variable, i.e., the ammonia-to-coal ratio u = a / m, where a is the ammonia input and m is the coal feed rate; d is the disturbance variable, d = [Q, F] air ] T Q is the boiler load, F air Let y represent the primary air volume, A, B, C, and E be the model parameter matrices for the state variable, control variable, output variable, and disturbance variable, respectively, and y be the predicted NO. x Exhaust gas concentration, k is the kth sampling time, and the superscript T indicates the transpose of a matrix or vector, which transforms a column vector into a row vector.

[0056] Vectorizing the model parameters includes:

[0057] Expand matrices A, B, C, and E into one-dimensional vectors to obtain the vectorized matrix θ:

[0058] θ=[vec(A) T ,vec(B) T ,vec(C) T ,vec(E) T ] T ∈R d ;

[0059] Where d = n 2 +nm+pn+nq, where n is the dimension of the state variables, m is the dimension of the control variables, p is the dimension of the output variables, and q is the dimension of the perturbation variables.

[0060] The algorithm employs a hybrid approach based on particle swarm optimization and recursive least squares to identify the vectorized model parameters online, including the following steps:

[0061] First, try using the particle swarm optimization algorithm to search on historical data to obtain initial estimates of model parameters and initial values ​​of the covariance matrix of the recursive least squares algorithm.

[0062] After collecting a new set of operating condition data, the model parameters and covariance matrix are updated according to the recursive least squares algorithm;

[0063] Every set number of sampling periods, the working condition data from the most recent set number of sampling periods is used as input. Centered on the current model parameters, a short-time particle swarm optimization algorithm is run within a set range to correct the model parameters and reset the covariance matrix.

[0064] As a preferred embodiment of the above, it further includes:

[0065] When a sudden change in operating conditions is detected or the model prediction error exceeds the threshold, a short-term particle swarm algorithm is immediately triggered to correct the model parameters.

[0066] In this embodiment, the objective function is:

[0067]

[0068] In the formula, Np is the prediction step size, Nc is the control step size, and λ is the penalty factor for changes in the control quantity. For target NO x Exhaust gas concentration limits

[0069] The objective function also includes the following constraints:

[0070]

[0071] In the formula, Δu(k)=u(k)-u(k-1).

[0072] like Figure 2 As shown, this embodiment also includes a dynamic control system for the ammonia-to-coal ratio, using the method described above. The system includes:

[0073] The data acquisition unit is used to collect data including boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate;

[0074] The modeling unit is used to establish a discrete-time state-space model based on operating condition signals and to vectorize the model parameters.

[0075] The parameter identification unit is used to identify vectorized model parameters online using a hybrid algorithm based on particle swarm optimization and recursive least squares.

[0076] The solving unit is used to construct the model prediction control objective function based on the identified current model parameters, preset prediction step size and control step size, and solve it online using the particle swarm algorithm to obtain the optimal future control increment sequence;

[0077] The control unit is used to dynamically regulate the ammonia-coal ratio using the optimal control increment sequence.

[0078] Example 2:

[0079] This embodiment includes a method for dynamic control of the ammonia-to-coal ratio, comprising the following steps:

[0080] S1: Real-time data collection of boiler load, furnace temperature, primary air volume, and NO. x Operating condition signals such as exhaust gas concentration, ammonia dosage, and coal feed rate;

[0081] S2: Establish a discrete-time state-space model based on the acquired signals, and vectorize the model parameters as θ;

[0082] S3: The model parameter θ is identified online using a PSO-RLS hybrid algorithm, where:

[0083] S31: In the offline stage, the initial parameter estimate θ(0) and the initial value P(0) of the covariance matrix of the recursive least squares algorithm RLS are obtained by searching on historical data using the particle swarm optimization (PSO) algorithm.

[0084] S32: During the online phase, for each new set of sampled data received, θ(k) and P(k) are updated according to the recursive least squares (RLS) algorithm;

[0085] S33: Every preset sampling period Tpso, using the most recent Nr samples as input, and based on the current θ(k) as the center, run a short-term PSO within the range [θ(k)±Δmax] to correct θ(k) and reset P(k);

[0086] S4: Based on the current model θ(k), the preset prediction step size Np and control step size Nc, construct the model predictive control (MPC) objective function, and use the PSO algorithm to solve it online to obtain the optimal future control increment sequence;

[0087] S5: Update the current ammonia-to-coal ratio u(k) = u(k-1) + Δu* with the optimal control increment, and apply it to the denitrification system;

[0088] S6: Repeat steps S1 to S5 to form a closed-loop dynamic control.

[0089] The specific steps are as follows:

[0090] Acquisition signals: Boiler load Q, furnace temperature T f Primary air volume F air NO x Exhaust gas concentration C NOx The amount of ammonia added (a) and the amount of coal fed (m) are shown.

[0091] Establish a discrete-time state-space model:

[0092]

[0093] In the formula, x is the state vector x = [T f C NOx ] T T f C represents the furnace temperature. NOx NO x The exhaust gas concentration, u is the controlled variable, i.e., the ammonia-to-coal ratio u = a / m, where a is the ammonia input and m is the coal feed rate; d is the disturbance variable, d = [Q, F] air ] T Q is the boiler load, F air Let y represent the primary air volume, A, B, C, and E be the model parameter matrices for the state variable, control variable, output variable, and disturbance variable, respectively, and y be the predicted NO. x Exhaust gas concentration, k is the kth sampling time, and the superscript T indicates the transpose of a matrix or vector, which transforms a column vector into a row vector.

[0094] Vectorizing the model parameters includes:

[0095] Expand matrices A, B, C, and E into one-dimensional vectors to obtain the vectorized matrix θ:

[0096] θ=[vec(A) T ,vec(B) T ,vec(C) T ,vec(E) T ] T ∈R d ;

[0097] Where d = n 2 +nm+pn+nq, where n is the dimension of the state variables, m is the dimension of the control variables, p is the dimension of the output variables, and q is the dimension of the perturbation variables.

[0098] First, try using the particle swarm optimization algorithm to search on historical data to obtain initial estimates of model parameters and initial values ​​of the covariance matrix of the recursive least squares algorithm.

[0099] After collecting a new set of operating condition data, the model parameters and covariance matrix are updated according to the recursive least squares algorithm;

[0100] Every set number of sampling periods, the working condition data from the most recent set number of sampling periods is used as input. Centered on the current model parameters, a short-time particle swarm optimization algorithm is run within a set range to correct the model parameters and reset the covariance matrix.

[0101] I. Initial offline PSO identification:

[0102] 1. Particle Swarm Optimization:

[0103] Group size: P;

[0104] Maximum number of iterations:

[0105] Inertia weight: ω∈[0.9,0.4] (linearly decreasing);

[0106] Learning factors: c1, c2;

[0107] 2. Fitness function:

[0108] Using offline historical data {u(k),d(k),y(k)}, after reconstructing the state-space model for a given θ, calculate the sum of squared prediction errors:

[0109]

[0110] 3. Iterative Process:

[0111] For each particle j = 1, ..., P, initialize its position, velocity, and individual optimality;

[0112] For t = 0, ..., T max pso -1:

[0113] Calculate the fitness of each particle Update pbest and the globally optimal gbest; update velocity and position according to the following formula:

[0114]

[0115] Constraint: If If the value exceeds the preset boundary, it will be truncated.

[0116] 4. Output: After the iteration, take the θ(0) corresponding to gbest and reshape it to A(0), B(0), C(0), E(0); let the initial covariance matrix of RLS be P0 = αI d (α is a relatively large constant).

[0117] II. Online RLS Update:

[0118] When a new observation (u(k),d(k),y(k)) arrives at time k:

[0119] 1. Construct the regression vector:

[0120] Based on the "expansion form" of the discrete state-space model, for example:

[0121] y(k)=Cx(k)=C(Ax(k-1)+Bu(k-1)+Ed(k-1));

[0122] This can be written in linear regression form: y(k)=φ(k) T θ(k-1)+ε(k);

[0123] Where φ(k) contains components of x(k-1), u(k-1), and d(k-1).

[0124] 2. Gain Calculation:

[0125]

[0126] Where λ∈(0,1] is the forgetting factor.

[0127] 3. Parameter update:

[0128] θ(k)=θ(k-1)+K(k)[y(k)-φ(k) T θ(k-1)];

[0129] 4. Covariance Update:

[0130]

[0131] 5. Restore the model matrix:

[0132] θ(k) is split and reshaped back into A(k), B(k), C(k), and E(k) for further prediction and control.

[0133] 3. Regular PSO correction:

[0134] To prevent RLS from getting trapped in local optima under nonlinear or abrupt conditions, PSO is run again every Tpso sampling periods (e.g., 300s) based on the most recent Nr samples {φ(i), y(i)}:

[0135] 1. Construct the search range centered on the current θ(k):

[0136] Θ min =θ(k)-△ max Θ max =θ(k)+△ max

[0137] 2. Short-term PSO optimization:

[0138] Iteration only Step, update gbest;

[0139] 3. Reset RLS:

[0140] Let θ(k) ← new gbest, P(k) ← αId.

[0141] The relevant parameters of the model are shown in Table 1:

[0142] Table 1 Key Parameter Settings

[0143] PSO group size P 20–50 A smaller scale is sufficient; for applications with high real-time requirements, 20 can be selected. PSO Maximum Iteration Tmax 50–100 100 can be taken during offline correction, and 30 can be taken during online correction. RLS Forgetting Factor λ 0.98–0.995 Historical data are given higher weight when the value approaches 1. Initial covariance coefficient α <![CDATA[10 6 –10 8 ]]> Ensure rapid convergence in the initial stage Correction period Tpso 200–500 samples Adjust according to system response characteristics Correct data window length Nr 100–300 samples Matching the correction period

[0144] For a 600MW coal-fired power unit, the specific control steps are as follows:

[0145] Data preprocessing: Sample every 5 seconds, denoise and then normalize;

[0146] Model identification: The PSO-RLS hybrid algorithm was used to obtain A, B, C, and E in the initial 50 iterations;

[0147] MPC parameter selection: prediction step size Np = 10, control step size Nc = 3, penalty factor λ = 0.5;

[0148] PSO settings: Population size P = 30, maximum number of iterations The inertia weight ω decreases linearly from 0.9 to 0.4;

[0149] Real-time operation: After the system is running stably, NO x Emissions maintained at <100 mg / Nm 3 Ammonia slip rate <2%.

[0150] Please see Figure 3 The diagram shows a structural schematic of a computer device provided in an embodiment of this application. An embodiment of this application provides a computer device 400, including a processor 410 and a memory 420. The memory 420 stores a computer program executable by the processor 410. When the computer program is executed by the processor 410, it performs the method described above.

[0151] This application embodiment also provides a storage medium 430, on which a computer program is stored, and the computer program is executed by a processor 410 to perform the above method.

[0152] The storage medium 430 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0153] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. "A plurality of" means two or more, unless otherwise explicitly specified.

[0154] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0155] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0156] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0157] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0158] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0159] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0160] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for dynamic control of the ammonia-to-coal ratio, characterized in that, Includes the following steps: Data collected includes boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate; A discrete-time state-space model is established based on the aforementioned operating condition signal, and the model parameters are vectorized. A hybrid algorithm based on particle swarm optimization and recursive least squares is used to identify the vectorized model parameters online. Based on the identified current model parameters, preset prediction step size and control step size, a model prediction control objective function is constructed, and the optimal future control increment sequence is obtained by solving the particle swarm algorithm online. The ammonia-to-coal ratio is dynamically adjusted using the optimal future control increment sequence. The discrete-time state-space model is as follows: ； In the formula, x is the state vector x=[T f C NOx ] T T f C represents the furnace temperature. NOx NO x The exhaust gas concentration, u is the controlled variable, i.e., the ammonia-to-coal ratio u = a / m, where a is the ammonia injection rate and m is the coal feed rate; d is the disturbance variable, d = [Q, F] air ] T Q is the boiler load, F air Let y represent the primary air volume, A, B, C, and E be the model parameter matrices for the state variable, control variable, output variable, and disturbance variable, respectively, and y be the predicted NO. x Exhaust gas concentration, k is the kth sampling time, and the superscript T indicates the transpose of a matrix or vector, which transforms a column vector into a row vector; The process of vectorizing model parameters includes: Expand matrices A, B, C, and E into one-dimensional vectors to obtain the vectorized matrix θ: θ=[vec(A) T ,thing(B) T ,thing(C) T ,thing(E) T ] T ∈R d ; Where d=n 2 +nm+pn+nq, where n is the dimension of the state variables, m is the dimension of the control variables, p is the dimension of the output variables, and q is the dimension of the disturbance variables; The objective function is: ； In the formula, Np is the prediction step size, Nc is the control step size, and λ is the penalty factor for changes in the control quantity. For target NO x Exhaust gas concentration limits; The objective function also includes the following constraints: ， ； In the formula, Δu(k) = u(k) - u(k-1).

2. The method for dynamic control of the ammonia-to-coal ratio according to claim 1, characterized in that, The online identification of the vectorized model parameters using a hybrid algorithm based on particle swarm optimization and recursive least squares includes the following steps: First, use the particle swarm optimization algorithm to search on historical data to obtain the initial estimated values ​​of the model parameters and the initial values ​​of the covariance matrix of the recursive least squares algorithm. After collecting a new set of operating condition data, the model parameters and covariance matrix are updated according to the recursive least squares algorithm; Every set number of sampling periods, using the working condition data from the most recent set number of sampling periods as input, and centering on the current model parameters, a short-time particle swarm optimization algorithm is run within a set range to correct the model parameters and reset the covariance matrix.

3. The method for dynamic control of the ammonia-to-coal ratio according to claim 2, characterized in that, Also includes: When a sudden change in operating conditions is detected or the model prediction error exceeds the threshold, a short-time particle swarm algorithm is immediately triggered to correct the model parameters.

4. A dynamic control system for the ammonia-to-coal ratio, characterized in that, Using the method as described in any one of claims 1 to 3, the system comprises: The data acquisition unit is used to collect data including boiler load, furnace temperature, primary air volume, and NO. x Operating signals for exhaust gas concentration, ammonia dosage, and coal feed rate; The modeling unit is used to establish a discrete-time state-space model based on the operating condition signal and to vectorize the model parameters. The parameter identification unit is used to identify the vectorized model parameters online using a hybrid algorithm based on particle swarm optimization and recursive least squares. The solving unit is used to construct the model prediction control objective function based on the identified current model parameters, preset prediction step size and control step size, and solve it online using the particle swarm algorithm to obtain the optimal future control increment sequence; The control unit is used to dynamically control the ammonia-coal ratio using the optimal future control increment sequence.

5. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-3.

6. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method as described in any one of claims 1-3.

Images