A boiler pressure control method based on pressure prediction model

Abstract

The present invention relates to a boiler pressure control method based on a pressure prediction model, comprising: B1, obtaining a boiler operation data set corresponding to the current moment t, a first model variable corresponding to the current moment t, and a second model variable based on boiler operation data collected at preset time intervals; B2, obtaining a boiler pressure trend feature corresponding to the current moment t based on the boiler operation data set Q corresponding to the current moment t; B3, using a pre-acquired pressure prediction model, based on the boiler pressure trend feature corresponding to the current moment t, the first model variable corresponding to the current moment t, and the second model variable, predicting and deriving the boiler pressure at k time intervals after the current moment t, and forming the boiler pressures at the k time intervals after the current moment t into a first boiler pressure sequence; B4, judging whether the first boiler pressure sequence meets a preset control target, and if so, making no adjustment to the boiler.

Description

Technical Field

[0001] The present invention relates to the technical field of energy production, and in particular to a boiler pressure control method based on a pressure prediction model. Background Art

[0002] During the production and operation of coal-fired units, boiler pressure is a key factor affecting the boiler combustion condition and one of the important control parameters. It reflects the economy and safety of the unit operation. Excessive pressure will cause damage to the unit equipment, and too low pressure cannot guarantee the normal operation of the unit. Instability of boiler pressure will not only have a great negative impact on the unit equipment and the entire production process, but also cause great economic losses to the enterprise. Therefore, it is necessary to ensure that the pressure is maintained within the allowable range, and in particular, it is required to strictly control the occurrence of excessive or insufficient pressure.

[0003] The common existing control practice is to use the PID algorithm to adjust the boiler load using pressure feedback. However, due to the dynamic characteristics of boiler operation such as large delay, large inertia and unstable load fluctuations, the feedback adjustment effect and response speed are poor, and the adjustment of control parameters is too dependent on operating experience. There are also methods based on machine learning models such as RBF neural networks to establish a functional relationship between pressure and controllable parameters such as fuel and air volume for control. However, since most of these machine learning models do not have the timing characteristics of continuous production processes, they do not reflect the energy release process during combustion well and the model complexity is high, and they cannot fundamentally solve the lag problem in pressure control. Summary of the Invention

[0004] (1) Technical issues to be resolved

[0005] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a boiler pressure control method based on a pressure prediction model, which solves the technical problem that the common existing control practice is to use the PID algorithm to adjust the boiler load using pressure feedback, but due to the dynamic characteristics of the boiler operation such as large delay, large inertia and unstable load fluctuations, the feedback adjustment effect and response speed are poor, and the setting of the control parameters is too dependent on operating experience. In addition, the control through machine learning models in the prior art does not well reflect the energy release process during combustion and the model complexity is high, which cannot fundamentally solve the lag problem in pressure control.

[0006] (2) Technical solution

[0007] In order to achieve the above objectives, the main technical solutions adopted by the present invention include:

[0008] An embodiment of the present invention provides a boiler pressure control method based on a pressure prediction model, comprising:

[0009] B1. Based on the boiler operation data collected at the preset time interval, obtain the boiler operation data set corresponding to the current time t and the first model variable X corresponding to the current time t. t and the second model variable Y t The boiler operation data set Q corresponding to the current time t = [Q t-T+1 ,...Q t-i ,...Q t-2 , Q t-1 , Q t ]; among them, Q t is the boiler operation data collected at the current time t; Q t-i The boiler operation data collected in the i time intervals before the current time t, including: coals t-i 、Air volume into furnace t-i , boiler pressure p t-i , consumption traffic steam t-i ;

[0010] Y t =Stealn t -Stealm t-l .

[0011] Among them, dc t =coal t -coal t-1 ;dw t =wind t -wind t-1 ;ds t =steam t -steam t-1 ;dc t is the change in the amount of coal fed into the furnace at time t; dw t is the change in the air volume entering the furnace at time t; ds t is the change in consumption flow at time t; 10≥T≥3; 20≥L≥10;

[0012] B2. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure trend feature trend corresponding to the current time t t ;

[0013] B3, based on the boiler pressure trend characteristics corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t, using a pre-acquired pressure prediction model, predicting and delivering the boiler pressures of k time intervals after the current moment t, and forming the boiler pressures of the k time intervals after the current moment t into a first boiler pressure sequence;

[0014] B4. Determine whether the first boiler pressure sequence meets a preset control target. If so, do not adjust the boiler.

[0015] Preferably, B2 specifically includes:

[0016] B21. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure sequence S at time t t ;

[0017] Among them, the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ];

[0018] B22, the boiler pressure sequence S at time t t As the boiler pressure sequence to be fitted, a preset strategy is adopted to obtain the boiler pressure trend characteristics corresponding to the boiler pressure sequence to be fitted;

[0019] The preset strategy is as follows: using a preset fixed index sequence [1, 2, 3...T] as an independent variable and the boiler pressure sequence to be fitted as a dependent variable, a T-order polynomial is used for fitting to obtain a polynomial function, and based on the polynomial function, the first-order derivative of the polynomial function is obtained, and the value of the first-order derivative at T is used as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted;

[0020] B23, the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted is used as the boiler pressure trend feature trend at the current time t t .

[0021] Preferably, B3 specifically includes:

[0022] B31, based on the boiler pressure trend characteristic trend corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t , substitute the previously acquired pressure prediction model to predict the boiler pressure p at one time interval after the current time t t+1 ;

[0023] The pre-acquired pressure prediction model is:

[0024]

[0025] in, in is the specific value of the model coefficient of the coal quantity fed into the furnace obtained in advance; is the specific value of the inlet air volume model coefficient M obtained in advance; e0 is the specific value of the flow-pressure transfer coefficient e obtained in advance;

[0026] B32, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ] and the boiler pressure p at one time interval after the current time t t+1 , using the first recursive formula, respectively obtain the boiler pressures from the second time interval to the kth time interval after the current time t, and form a first boiler pressure sequence from the boiler pressures of the k time intervals after the current time t;

[0027] The first recursive formula is:

[0028]

[0029] Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 The boiler pressure sequence to be fitted is obtained using the preset strategy;

[0030] The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 The changes in the amount of coal entering the furnace and the amount of air entering the furnace after time t are both set to 0;

[0031] When k is greater than 1, Y t+k-1 =0;

[0032] Wherein, the first boiler pressure sequence includes: t+1 、p t+2 、p t+3 ,…,p t+i ...、p t+k .

[0033] Preferably, the preset control target is the first condition or the second condition;

[0034] The first condition is:

[0035] lower≤p t+i ≤upper;i=1,2,...,k;

[0036] Among them, lower is the minimum value of the pre-set safety range;

[0037] Upper is the maximum value of the pre-set safety range;

[0038] The second condition is:

[0039]

[0040] in, is the default value;

[0041] ε is the preset tolerable pressure deviation.

[0042] Preferably, the method further comprises:

[0043] B5. If the first boiler pressure sequence does not meet the preset control target, generate a feasible solution set based on the preset single maximum coal feeding adjustment amount max and the preset coal feeding adjustment step size step;

[0044] B6. Based on the feasible solution set, obtain the corresponding pressure trend prediction sequence space;

[0045] B7. Based on the pressure trend prediction sequence space, obtain an operation value of the amount of coal entering the furnace and an operation value of the amount of air entering the furnace for controlling the operation of the boiler.

[0046] Preferably, B5 specifically includes:

[0047] If p t+i >upper or Then, within the interval [-max, 0], a feasible solution for the recommended value of the incoming coal quantity re_c is generated with a fixed step size step, and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the incoming air quantity re_w corresponding to each feasible solution for the recommended value of the incoming coal quantity re_c; where re_w = re_c*r;

[0048] The feasible solution H for generating the recommended value re_c of the amount of coal fed into the furnace with a fixed step size step in the interval [-max, 0] includes: starting from -max, the value after each fixed step size step is increased, and the recommended value re_c of the amount of coal fed into the furnace is less than 0;

[0049] If p t+i <lower or Then, within the interval [0, max], a feasible solution for the recommended value of the coal quantity re_c is generated with a fixed step size step, and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the air quantity re_w corresponding to each feasible solution for the recommended value of the coal quantity re_c; where re_w = re_c*r;

[0050] The feasible solutions for generating the recommended value re_c of the amount of coal fed into the furnace with a fixed step size step in the interval [0, max] include: starting from 0 and increasing the value after each fixed step size step, and the recommended value re_c of the amount of coal fed into the furnace is less than max;

[0051] Generate a feasible solution set based on the recommended value of the coal flow into the furnace re_c and the corresponding feasible solution re_w of the recommended value of the air flow into the furnace;

[0052] Among them, the feasible solution set is H is the total number of recommended values ​​of coal quantity re_c generated in the feasible solutions of the generated recommended values ​​of coal quantity re_c.

[0053] Preferably, the B6 specifically includes:

[0054] B61, based on the boiler pressure trend characteristic trend corresponding to the current time t t , update the first model variable X using a preset first method t The obtained X t *, and X t *Substitute into the previously acquired pressure prediction model to obtain the predicted boiler pressure value p for one time interval after the current time t t+1 *;

[0055] in,

[0056] The first preset method is: based on any j-th feasible solution [re_c j , re_w j ] in dc t Then add re_c j and dw t Then add re_w j , get the updated first model variable X t *;

[0057]

[0058] B62, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1, p t ] and the predicted boiler pressure value p at one time interval after the current time t t+1 *, using the second recursive formula, respectively obtain the predicted boiler pressure values ​​from the second time interval to the kth time interval after the current time t, and form the jth feasible solution [re_c j , re_w j ] corresponds to the second boiler pressure sequence A j ;

[0059] The second recursive formula is:

[0060]

[0061] Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 * is the boiler pressure sequence to be fitted, obtained using the preset strategy;

[0062] The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 *The changes in the amount of coal and air entering the furnace greater than time t are set to 0;

[0063] Y t+k-1 *=0;

[0064] Among them, the second boiler pressure sequence A j Includes: p t+1 * 、p t+2 * 、p t+3 * ,…,p t+i * ...、p t+k * ;

[0065] B63. Obtaining a pressure trend prediction sequence space based on the second boiler pressure sequence corresponding to each set of feasible solutions in the feasible solution set;

[0066] Among them, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ];

[0067] A j is the jth feasible solution [re_c j , re_wj ] corresponds to the second boiler pressure sequence.

[0068] Preferably, the B7 specifically includes:

[0069] B71, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ] Screening out the second boiler pressure sequence that meets the pre-set control target;

[0070] B72. Filter out the second boiler pressure sequence that meets the pre-set control target and select the recommended coal quantity with the smallest absolute value in the corresponding feasible solution, and use the recommended coal quantity with the smallest absolute value and its corresponding recommended air volume as the coal quantity operation value and air volume operation value for controlling the boiler operation.

[0071] Preferably, before B1, the method further includes:

[0072] B0-1. Obtaining an initial historical boiler operation data set based on boiler operation data collected at pre-set time intervals within a historical time period;

[0073] B0-2. Preprocess the initial historical boiler operation data set to obtain the boiler operation data set Qa corresponding to the pre-specified time a = [Q a-T+1 ,...Q a-i ,...Q a-2 , Q a-1 , Q a ];

[0074] The preprocessing is to fill the missing values ​​in the initial historical boiler operation data set using a spline interpolation method;

[0075] B0-3. Based on the pre-specified boiler operation data set corresponding to the time a, obtain the boiler pressure sequence Sa corresponding to the time a, where Sa = [p a-T+1 ,...p a-i ,...p a-2 , p a-1 , p a ];

[0076] B0-4. Take the boiler pressure sequence Sa corresponding to the moment a as the boiler pressure sequence to be fitted, and use the preset strategy to obtain the boiler pressure trend feature trend at the moment a. a ;

[0077] The preset strategy is as follows: using a preset fixed index sequence [1, 2, 3...T] as an independent variable and the boiler pressure sequence to be fitted as a dependent variable, a T-order polynomial is used for fitting to obtain a polynomial function, and based on the polynomial function, the first-order derivative of the polynomial function is obtained, and the value of the first-order derivative at T is used as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted;

[0078] B0-5. Based on the boiler operation data set Qa corresponding to the moment a, obtain characteristic change information corresponding to any two adjacent boiler operation data in the boiler operation data set Qa corresponding to the moment a;

[0079] The characteristic change information includes: the change in the amount of coal fed into the furnace, the change in the amount of air fed into the furnace, the change in the consumption flow rate, and the change in the boiler pressure trend between the two adjacent boiler operation data; the boiler air-coal ratio; the boiler air-coal ratio is the ratio of the difference in the amount of air fed into the furnace to the difference in the amount of coal fed into the furnace;

[0080] B0-6. The average value of the data within the 25% quantile to the 75% quantile of the air-to-coal ratio of all boilers corresponding to the boiler operation data set Qa corresponding to time a is taken as the optimal air-to-coal ratio coefficient.

[0081] Preferably, after step B0-6, the method further includes:

[0082] B0-7, based on the characteristic change information between the adjacent b-th boiler operation data and the b-1-th boiler operation data in the boiler operation data set Qa corresponding to time a, calculate the corresponding modeling data R according to formula (A), formula (B), and formula (C) b ;

[0083] The modeling data R b Includes: X b 、Y b 、Z b ;

[0084] Wherein, formula (A) is:

[0085]

[0086] dc b =coal b -coal b-1 ;dc b is the change in the amount of coal fed into the furnace at time b;

[0087] dw b =wind b -wind b-1 ;dw b is the change in the air volume entering the furnace at time b;

[0088] Wherein, formula (B) is:

[0089] Y b =ds b ;

[0090] ds b =steam b -steam b-1 ;ds b is the change in consumption flow at time b;

[0091] Wherein, formula (C) is:

[0092] Z b =dt b ;

[0093] dt b =trend b -trend b-1 ;dt b The boiler pressure trend characteristics at time b and the boiler pressure trend characteristics at time b-1;

[0094] B0-8. All N modeling data corresponding to the boiler operation data set Qa corresponding to time a are combined into a modeling data set;

[0095] B0-9. Based on the modeling data set, the particle swarm optimization algorithm is used to optimize the loss function to obtain the model parameters in the loss function. The specific value M* of and the specific value e* of the flow-pressure transfer coefficient e;

[0096] The loss function G is:

[0097]

[0098] In the particle swarm optimization algorithm, the inertia factor is greater than or equal to 0.1 and less than or equal to 0.9; the social factor is greater than or equal to 0.5 and less than or equal to 2.5; the individual factor is greater than or equal to 0.5 and less than or equal to 2.5; the population size is greater than or equal to 50 and less than or equal to 200; and the number of evolutions is greater than or equal to 100 and less than or equal to 5000.

[0099] (3) Beneficial effects

[0100] The beneficial effects of the present invention are: a boiler pressure control method based on a pressure prediction model of the present invention adopts model variables to simulate the influence of the amount of coal entering the furnace and the amount of air entering the furnace on the main steam flow in the boiler production process, and establishes a conduction relationship between the flow and the boiler pressure in combination with the consumption flow, and determines the pressure prediction model with the optimal parameters obtained by optimizing the boiler operation data collected according to a preset time interval in a historical time period. The prediction of future pressure trends enables the control method to have the ability to display the future dynamic behavior of the system, and then generates a feasible solution set in combination with specific future control strategies. The prediction of the feasible solution set and the determination of the optimal operating values ​​of the amount of coal entering the furnace and the operating values ​​of the air entering the furnace can fundamentally solve the problems of insufficient control accuracy and control lag in pressure control, thereby improving the overall safety and economic benefits of boiler operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 This is a flow chart of a boiler pressure control method based on a pressure prediction model of the present invention;

[0102] Figure 2 Schematic diagram of the polynomial fitting curve and first-order derivative effect in the preset strategy in an embodiment of the present invention;

[0103] Figure 3 This is a flow chart of a boiler pressure control method based on a pressure prediction model in an embodiment of the present invention. DETAILED DESCRIPTION

[0104] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation methods in conjunction with the accompanying drawings.

[0105] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. Instead, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0106] See also Figure 1 This embodiment provides a boiler pressure control method based on a pressure prediction model, including:

[0107] B1. Based on the boiler operation data collected at the preset time interval, obtain the boiler operation data set corresponding to the current time t and the first model variable X corresponding to the current time t. t and the second model variable Y t The boiler operation data set Q corresponding to the current time t = [Q t-T+1 ,...Qt-i ,...Q t-2 , Q t-1 , Q t ]; among them, Q t is the boiler operation data collected at the current time t; Q t-i The boiler operation data collected in the i time intervals before the current time t, including: coals t-i 、Air volume into furnace t-i , boiler pressure p t-i , consumption traffic steam t-i ;

[0108] Y t =steam t -steam t-1 ;

[0109] Among them, dc t =coal t -coal t-1 ;dw t =wind t -wind t-1 ;ds t =steam t -steam t-1 ;dc t is the change in the amount of coal fed into the furnace at time t; dw t is the change in the air volume entering the furnace at time t; ds t is the change in consumption flow at time t; 10≥T≥3; 20≥L≥10.

[0110] In this embodiment, the first model variable X t It is an L×2 matrix consisting of the changes in the amount of coal entering the furnace and the amount of air entering the furnace.

[0111] B2. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure trend feature trend corresponding to the current time t t .

[0112] In this embodiment, B2 specifically includes:

[0113] B21. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure sequence S at time t t ; Among them, the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ].

[0114] B22, the boiler pressure sequence S at time t t As the boiler pressure sequence to be fitted, a preset strategy is adopted to obtain the boiler pressure trend characteristics corresponding to the boiler pressure sequence to be fitted.

[0115] The preset strategy is: take the preset fixed index sequence [1, 2, 3...T] as the independent variable and the boiler pressure sequence to be fitted as the dependent variable, use the T-order polynomial to fit, obtain the polynomial function, and obtain the first-order derivative of the polynomial function based on the polynomial function, and take the value of the first-order derivative at T as the reference value. Figure 2 , as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted.

[0116] B23, the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted is used as the boiler pressure trend feature trend at the current time t t .

[0117] B3, based on the boiler pressure trend characteristics corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t , using a pre-acquired pressure prediction model, predict and present the boiler pressures of k time intervals after the current moment t, and compose the boiler pressures of k time intervals after the current moment t into a first boiler pressure sequence.

[0118] In the specific application of this embodiment, B3 specifically includes:

[0119] B31, based on the boiler pressure trend characteristic trend corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t , substitute the previously acquired pressure prediction model to predict the boiler pressure p at one time interval after the current time t t+1 .

[0120] The pre-acquired pressure prediction model is:

[0121]

[0122] in, in is the specific value of the model coefficient of the coal quantity fed into the furnace obtained in advance; is the specific value of the inlet air volume model coefficient M obtained in advance; e0 is the specific value of the flow-pressure transfer coefficient e obtained in advance.

[0123] B32, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ] and the boiler pressure p at one time interval after the current time t t+1 , using the first recursive formula, respectively obtain the boiler pressures from the second time interval to the kth time interval after the current moment t, and form a first boiler pressure sequence from the boiler pressures of the k time intervals after the current moment t.

[0124] The first recursive formula is:

[0125]

[0126] Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 The boiler pressure sequence to be fitted is obtained using the preset strategy.

[0127] The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 The changes in the amount of coal entering the furnace and the amount of air entering the furnace after time t are both set to 0.

[0128] For example, if k = 2, X t+1 for:

[0129]

[0130] If k = 3, X t+2 for:

[0131]

[0132] When k is greater than 1, Y t+k-1 =0.

[0133] Wherein, the first boiler pressure sequence includes: t+1 、p t+2 、p t+3 ,…,p t+i ...、p t+k .

[0134] B4. Determine whether the first boiler pressure sequence meets a pre-set control target. If so, do not adjust the boiler. The pre-set control target is the first condition or the second condition.

[0135] The first condition is: lower≤p t+i ≤upper; i=1, 2,...,k.

[0136] Among them, lower is the minimum value of the preset safety range; upper is the maximum value of the preset safety range.

[0137] The second condition is: in, is the preset value; ε is the preset tolerable pressure deviation.

[0138] See also Figure 3 In this embodiment, the method further includes:

[0139] B5. If the first boiler pressure sequence does not meet the preset control target, a feasible solution set is generated according to the preset single maximum coal feeding adjustment amount max and the coal feeding adjustment step step.

[0140] Wherein, the B5 specifically includes:

[0141] If p t+i >upper or Then, a feasible solution for the recommended value of the amount of coal entering the furnace re_c is generated with a fixed step size step within the interval [-max, 0], and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the amount of coal entering the furnace re_c and the corresponding feasible solution for the recommended value of the amount of air entering the furnace re_w; where re_w = re_c*r.

[0142] Among them, the feasible solution H for generating the recommended value re_c of the amount of coal entering the furnace with a fixed step size step in the interval [-max, 0] includes: the value after each increase of a fixed step size step starting from -max, and the recommended value re_c of the amount of coal entering the furnace is less than 0.

[0143] If p t+i <lower or Then, a feasible solution for the recommended value of the amount of coal entering the furnace re_c is generated with a fixed step size step within the interval [0, max], and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the amount of coal entering the furnace re_c and the corresponding feasible solution for the recommended value of the amount of air entering the furnace re_w; where re_w = re_c*r.

[0144] Among them, the feasible solutions for generating the recommended value re_c of the amount of coal entering the furnace with a fixed step size step in the interval [0, max] include: starting from 0, the value after each increase of a fixed step size step, and the recommended value re_c of the amount of coal entering the furnace is less than max.

[0145] A feasible solution set is generated based on the recommended value re_c of the coal flow rate and the corresponding feasible solution re_w of the recommended value of the air flow rate.

[0146] Among them, the feasible solution set is H is the total number of recommended values ​​of coal quantity re_c generated in the feasible solutions of the generated recommended values ​​of coal quantity re_c.

[0147] B6. Based on the feasible solution set, obtain the corresponding pressure trend prediction sequence space;

[0148] In this embodiment, a feasible solution set for air-coal control is generated based on the predicted value, and the optimal control quantity is found after predicting each feasible solution in the feasible solution set, thereby solving the problems of insufficient control accuracy and control lag in pressure control. Specifically in this embodiment, B6 includes:

[0149] B61, based on the boiler pressure trend characteristic trend corresponding to the current time t t , update the first model variable X using a preset first method t The obtained X t *, and X t *Substitute into the previously acquired pressure prediction model to obtain the predicted boiler pressure value p for one time interval after the current time t t+1 *.

[0150] in,

[0151] The first preset method is: based on any j-th feasible solution [re_c j , re_w j ] in dc t Then add re_c j and dw t Then add re_w j , get the updated first model variable X t *.

[0152]

[0153] B62, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ] and the predicted boiler pressure value p at one time interval after the current time t t+1*, using the second recursive formula, respectively obtain the predicted boiler pressure values ​​from the second time interval to the kth time interval after the current time t, and form the jth feasible solution in the feasible solution set with the predicted boiler pressure values ​​of the k time intervals after the current time t The corresponding second boiler pressure sequence A j .

[0154] The second recursive formula is:

[0155]

[0156] Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 *The boiler pressure series to be fitted is obtained using the preset strategy.

[0157] The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 *The changes in the amount of coal and air entering the furnace greater than time t are both set to 0.

[0158] For example, if k = 2, X t+1 * for:

[0159]

[0160] If k = 3, X t+2 * for:

[0161]

[0162] Y t+k-1 *=0.

[0163] Among them, the second boiler pressure sequence A j Includes: p t+1 * 、p t+2 * 、p t+3 * ,…,p t+i * ...、

[0164] p t+k * .

[0165] B63. Based on the second boiler pressure sequence corresponding to each group of feasible solutions in the feasible solution set, obtain a pressure trend prediction sequence space.

[0166] Among them, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ].

[0167] A j is the jth feasible solution [re_c j , re_w j ] corresponds to the second boiler pressure sequence. In other words, each second boiler pressure sequence in the pressure trend prediction sequence space has a one-to-one corresponding set of feasible solutions.

[0168] B7. Based on the pressure trend prediction sequence space, obtain an operation value of the amount of coal entering the furnace and an operation value of the amount of air entering the furnace for controlling the operation of the boiler.

[0169] The B7 specifically includes:

[0170] B71, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ]Screen out the second boiler pressure sequence that meets the pre-set control target.

[0171] B72. Filter out the second boiler pressure sequence that meets the pre-set control target and select the recommended coal quantity with the smallest absolute value in the corresponding feasible solution, and use the recommended coal quantity with the smallest absolute value and its corresponding recommended air volume as the coal quantity operation value and air volume operation value for controlling the boiler operation.

[0172] In this embodiment, a boiler pressure control method based on a pressure prediction model uses a matrix composed of parameters (i.e., model variables) to simulate the influence of the amount of coal and the amount of air entering the furnace on the main steam flow during the boiler production process, and establishes a conduction relationship between the flow and the boiler pressure in combination with the consumption flow. The pressure prediction model is determined by optimizing the optimal parameters obtained based on the boiler operation data collected at pre-set time intervals within a historical time period. The prediction of future pressure trends enables the control method to have the ability to demonstrate the future dynamic behavior of the system. Combined with specific future control strategies, a feasible solution set is generated. The prediction of the feasible solution set and the determination of the optimal operating values ​​of the amount of coal entering the furnace and the amount of air entering the furnace can fundamentally solve the problems of insufficient control accuracy and control lag in pressure control, thereby improving the overall safety and economic benefits of boiler operation.

[0173] In the practical application of this embodiment, the following is also included before B1:

[0174] B0-1. Acquire an initial historical boiler operation data set based on boiler operation data collected at pre-set time intervals within a historical time period.

[0175] B0-2. Preprocess the initial historical boiler operation data set to obtain the boiler operation data set Qa corresponding to the pre-specified time a = [Q a-T+1 ,...Q a-i ,...Q a-2 , Q a-1 , Q a ].

[0176] The preprocessing is to fill the missing values ​​in the initial historical boiler operation data set by using spline interpolation. In this embodiment, the cubic spline interpolation is performed using the relevant functions in the Python mathematical calculation library Scipy.

[0177] B0-3. Based on the pre-specified boiler operation data set corresponding to the time a, obtain the boiler pressure sequence Sa corresponding to the time a, where Sa = [p a-T+1 ,...p a-i ,...p a-2 , p a-1 , p a ].

[0178] B0-4. Take the boiler pressure sequence Sa corresponding to the moment a as the boiler pressure sequence to be fitted, and use the preset strategy to obtain the boiler pressure trend feature trend at the moment a. a .

[0179] The preset strategy is: using a preset fixed index sequence [1, 2, 3...T] as an independent variable and the boiler pressure sequence to be fitted as a dependent variable, using a T-order polynomial to perform fitting to obtain a polynomial function, and based on the polynomial function, obtaining the first-order derivative of the polynomial function, and using the value of the first-order derivative at T as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted.

[0180] B0-5. Based on the boiler operation data set Qa corresponding to the moment a, obtain characteristic change information corresponding to any two adjacent boiler operation data in the boiler operation data set Qa corresponding to the moment a.

[0181] The characteristic change information includes: the change in the amount of coal entering the furnace, the change in the amount of air entering the furnace, the change in the consumption flow, and the change in the boiler pressure trend between the two adjacent boiler operation data; the boiler air-coal ratio; the boiler air-coal ratio is the ratio of the difference in the amount of air entering the furnace to the difference in the amount of coal entering the furnace.

[0182] B0-6. The average value of the data within the 25% quantile to the 75% quantile of the air-to-coal ratio of all boilers corresponding to the boiler operation data set Qa corresponding to time a is taken as the optimal air-to-coal ratio coefficient.

[0183] In fact, after step B0-6, the following steps are also included:

[0184] B0-7, based on the characteristic change information between the adjacent b-th boiler operation data and the b-1-th boiler operation data in the boiler operation data set Qa corresponding to time a, calculate the corresponding modeling data R according to formula (A), formula (B), and formula (C) b .

[0185] The modeling data R b Includes: X b 、Y b 、Z b .

[0186] Wherein, formula (A) is:

[0187]

[0188] dc b =coal b -coal b-1 ;dc b is the change in the amount of coal fed into the furnace at time b.

[0189] dw b =wind b -wind a-1 ;dw b is the change in the air volume entering the furnace at time b.

[0190] Wherein, formula (B) is:

[0191] Y b =ds b .

[0192] ds b =steam b -steam b-1 ;ds b is the change in consumption flow at time b.

[0193] Wherein, formula (C) is:

[0194] Z b =dt b .

[0195] dt b =trend b -trend b-1 ;dtb The boiler pressure trend characteristics at time b and the boiler pressure trend characteristics at time b-1.

[0196] B0-8. All N modeling data corresponding to the boiler operation data set Qa corresponding to time a are combined into a modeling data set.

[0197] B0-9. Based on the modeling data set, the particle swarm optimization algorithm is used to optimize the loss function to obtain the model parameters in the loss function. The specific value M* and the specific value e* of the flow-pressure transfer coefficient e.

[0198] The loss function G is:

[0199]

[0200] in: is the Hadamard product of matrix M and matrix Xb-: (multiplication of elements at corresponding positions in two matrices with the same latitude), Represents the sum of all elements in the Hadamard product matrix.

[0201] In the particle swarm optimization algorithm, the inertia factor is greater than or equal to 0.1 and less than or equal to 0.9; the social factor is greater than or equal to 0.5 and less than or equal to 2.5; the individual factor is greater than or equal to 0.5 and less than or equal to 2.5; the population size is greater than or equal to 50 and less than or equal to 200; and the number of evolutions is greater than or equal to 100 and less than or equal to 5000.

[0202] In this embodiment, a boiler pressure control method based on a pressure prediction model uses matrix operations to model the transfer function, thereby enhancing the interpretability of the prediction model and making the model closer to the actual mechanism process. Decision-making and control are performed on the predicted "known future", which can better handle processes with large delays and large inertia characteristics. The traditional feedback adjustment is replaced by a prediction-based feedforward control method, which not only eliminates the strong dependence on operators, but also improves the response speed and adjustment accuracy of boiler operation.

[0203] This embodiment also provides a prediction-based boiler pressure control system, comprising: at least one processor; and at least one memory communicatively connected to the processor, wherein the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the boiler pressure control method based on the pressure prediction model as in the embodiment.

[0204] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0205] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions.

[0206] It should be noted that, in the claims, any reference signs placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In claims enumerating several means, several of these means may be embodied by one and the same hardware. The use of the words first, second, third etc. is for convenience only and does not indicate any order. These words may be understood as part of the component name.

[0207] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0208] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments after learning the basic creative concept. Therefore, the claims should be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0209] Obviously, those skilled in the art may make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if such modifications and variations fall within the scope of the claims and their equivalents, the present invention shall also include such modifications and variations.

Claims

1. A boiler pressure control method based on a pressure prediction model, characterized in that: include: B1. Based on the boiler operation data collected at pre-set time intervals, obtain the boiler operation data set Q corresponding to the current time t and the first model variable X corresponding to the current time t. t and the second model variable Y t The boiler operation data set Q corresponding to the current time t = [Q t-T+1 ,...Q t-i ,...Q t-2 , Q t-1 , Q t ]; among them, Q t is the boiler operation data collected at the current time t; Q t-i The boiler operation data collected in the i time intervals before the current time t, including: coals t-i 、Air volume into furnace t-i , boiler pressure p t-i , consumption traffic steam t-i ; Y t =steam t -steam t-1 ; Among them, dc t =coal t -coal t-1 ;dw t =wind t -wind t-1 ;dc t is the change in the amount of coal fed into the furnace at time t; dw t is the change in the air volume entering the furnace at time t; 10≥T≥3; 20≥L≥10; B2. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure trend feature trend corresponding to the current time t t ; B3, based on the boiler pressure trend characteristics corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t , using a pre-acquired pressure prediction model, predicting and delivering the boiler pressures of k time intervals after the current moment t, and forming the boiler pressures of the k time intervals after the current moment t into a first boiler pressure sequence; B4. Determine whether the first boiler pressure sequence meets a preset control target. If so, do not adjust the boiler. The B2 specifically includes: B21. Based on the boiler operation data set Q corresponding to the current time t, obtain the boiler pressure sequence S at time t t ; Among them, the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ]; B22, the boiler pressure sequence S at time t t As the boiler pressure sequence to be fitted, a preset strategy is adopted to obtain the boiler pressure trend characteristics corresponding to the boiler pressure sequence to be fitted; The preset strategy is as follows: using a preset fixed index sequence [1, 2, 3...T] as an independent variable and the boiler pressure sequence to be fitted as a dependent variable, a T-order polynomial is used for fitting to obtain a polynomial function, and based on the polynomial function, the first-order derivative of the polynomial function is obtained, and the value of the first-order derivative at T is used as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted; B23, the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted is used as the boiler pressure trend feature trend at the current time t t ; The B3 specifically includes: B31, based on the boiler pressure trend characteristic trend corresponding to the current time t t , the first model variable X corresponding to the current time t t , the second model variable Y t , substitute the previously acquired pressure prediction model to predict the boiler pressure p at one time interval after the current time t t+1 ; The pre-acquired pressure prediction model is: in, in is the specific value of the model coefficient of the coal quantity fed into the furnace obtained in advance; is the specific value of the inlet air volume model coefficient obtained in advance; e0 is the specific value of the flow-pressure transfer coefficient e obtained in advance; B32, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ] and the boiler pressure p at one time interval after the current time t t+1 , using the first recursive formula, respectively obtain the boiler pressures from the second time interval to the kth time interval after the current time t, and form a first boiler pressure sequence from the boiler pressures of the k time intervals after the current time t; The first recursive formula is: Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 The boiler pressure sequence to be fitted is obtained using the preset strategy; The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 The changes in the amount of coal entering the furnace and the amount of air entering the furnace after time t are both set to 0; When k is greater than 1, Y t+k-1 =0; Wherein, the first boiler pressure sequence includes: t+1 、p t+2 、p t+3 ,…,p t+i ...、p t+k .

2. The boiler pressure control method based on the pressure prediction model according to claim 1, characterized in that: The pre-set control target is the first condition or the second condition; The first condition is: lower≤p t+i ≤upper;i=1,2,...,k; Among them, lower is the minimum value of the pre-set safety range; Upper is the maximum value of the pre-set safety range; The second condition is: in, is the default value; ε is the preset tolerable pressure deviation.

3. The boiler pressure control method based on the pressure prediction model according to claim 2, characterized in that: The method further comprises: B5. If the first boiler pressure sequence does not meet the preset control target, generate a feasible solution set based on the preset single maximum coal feeding adjustment amount max and the preset coal feeding adjustment step size step; B6. Based on the feasible solution set, obtain the corresponding pressure trend prediction sequence space; B7. Based on the pressure trend prediction sequence space, obtain an operation value of the amount of coal entering the furnace and an operation value of the amount of air entering the furnace for controlling the operation of the boiler.

4. The boiler pressure control method based on the pressure prediction model according to claim 3 is characterized in that: The B5 specifically includes: If p t+i >upper or Then, within the interval [-max, 0], a feasible solution for the recommended value of the incoming coal quantity re_c is generated with a fixed step size step, and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the incoming air quantity re_w corresponding to each feasible solution for the recommended value of the incoming coal quantity re_c; where re_w = re_c*r; The feasible solution H for generating the recommended value re_c of the amount of coal fed into the furnace with a fixed step size step in the interval [-max, 0] includes: starting from -max, the value after each fixed step size step is increased, and the recommended value re_c of the amount of coal fed into the furnace is less than 0; If p t+i <lower or Then, within the interval [0, max], a feasible solution for the recommended value of the coal quantity re_c is generated with a fixed step size step, and the optimal air-coal ratio coefficient r obtained in advance is used to calculate the feasible solution for the recommended value of the air quantity re_w corresponding to each feasible solution for the recommended value of the coal quantity re_c; where re_w = re_c*r; The feasible solutions for generating the recommended value re_c of the amount of coal fed into the furnace with a fixed step size step in the interval [0, max] include: starting from 0 and increasing the value after each fixed step size step, and the recommended value re_c of the amount of coal fed into the furnace is less than max; Generate a feasible solution set based on the recommended value of the coal flow into the furnace re_c and the corresponding feasible solution re_w of the recommended value of the air flow into the furnace; Among them, the feasible solution set is H is the total number of recommended values ​​of coal quantity re_c generated in the feasible solutions of the generated recommended values ​​of coal quantity re_c.

5. The boiler pressure control method based on the pressure prediction model according to claim 4 is characterized in that: The B6 specifically includes: B61, based on the boiler pressure trend characteristic trend corresponding to the current time t t , update the first model variable X using a preset first method t The obtained X t *, and X t *Substitute into the previously acquired pressure prediction model to obtain the predicted boiler pressure value p for one time interval after the current time t t+1 * ; in, The first preset method is: based on any j-th feasible solution [re_c j ,re_w j ] in dc t Add re_c later j and dw t Then add re_w j , get the updated first model variable X t *; B62, based on the boiler pressure sequence S at time t t =[p t-T+1 ,...p t-i ,...p t-2 , p t-1 , p t ] and the predicted boiler pressure value p at one time interval after the current time t t+1 * , using the second recursive formula, respectively obtain the predicted boiler pressure values ​​from the second time interval to the kth time interval after the current time t, and form the jth feasible solution [re_c j , re_w j ] corresponds to the second boiler pressure sequence A j ; The second recursive formula is: Among them, trend t+k-1 The boiler pressure sequence S is a fixed index sequence [1, 2, 3...T] that is set in advance and is the k-1 time intervals after the current time t. t+k-1 * is the boiler pressure sequence to be fitted, obtained using the preset strategy; The new first model variable X corresponding to the k-1 time intervals after the current time t t+k-1 *The changes in the amount of coal and air entering the furnace greater than time t are set to 0; Y t+k-1 *＝0； Among them, the second boiler pressure sequence A j Includes: p t+1 * 、p t+2 * 、p t+3 * ,…,p t+i * ...、p t+k * ; B63. Obtaining a pressure trend prediction sequence space based on the second boiler pressure sequence corresponding to each set of feasible solutions in the feasible solution set; Among them, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ]; A j is the jth feasible solution [re_c j , re_w j ] corresponds to the second boiler pressure sequence.

6. The boiler pressure control method based on the pressure prediction model according to claim 5, characterized in that: The B7 specifically includes: B71, the pressure trend prediction sequence space is [A1, A2, ..., A j ,...,A H ] Screening out the second boiler pressure sequence that meets the pre-set control target; B72. Filter out the second boiler pressure sequence that meets the pre-set control target and select the recommended coal quantity with the smallest absolute value in the corresponding feasible solution, and use the recommended coal quantity with the smallest absolute value and its corresponding recommended air volume as the coal quantity operation value and air volume operation value for controlling the boiler operation.

7. The boiler pressure control method based on the pressure prediction model according to claim 6, characterized in that: Before B1, it also includes: B0-1. Obtaining an initial historical boiler operation data set based on boiler operation data collected at pre-set time intervals within a historical time period; B0-2. Preprocess the initial historical boiler operation data set to obtain the boiler operation data set Qa corresponding to the pre-specified time a = [Q a-T+1 ,...Q a-i ,...Q a-2 , Q a-1 , Q a ]; The preprocessing is to fill the missing values ​​in the initial historical boiler operation data set using a spline interpolation method; B0-3. Based on the pre-specified boiler operation data set corresponding to the time a, obtain the boiler pressure sequence Sa corresponding to the time a, where Sa = [p a-T+1 ,...p a-i ,...p a-2 , p a-1 , p a ]; B0-4. Take the boiler pressure sequence Sa corresponding to the moment a as the boiler pressure sequence to be fitted, and use the preset strategy to obtain the boiler pressure trend feature trend at the moment a. a ; The preset strategy is as follows: using a preset fixed index sequence [1, 2, 3...T] as an independent variable and the boiler pressure sequence to be fitted as a dependent variable, a T-order polynomial is used for fitting to obtain a polynomial function, and based on the polynomial function, the first-order derivative of the polynomial function is obtained, and the value of the first-order derivative at T is used as the boiler pressure trend feature corresponding to the boiler pressure sequence to be fitted; B0-5. Based on the boiler operation data set Qa corresponding to the moment a, obtain characteristic change information corresponding to any two adjacent boiler operation data in the boiler operation data set Qa corresponding to the moment a; The characteristic change information includes: the change in the amount of coal fed into the furnace, the change in the amount of air fed into the furnace, the change in the consumption flow rate, and the change in the boiler pressure trend between the two adjacent boiler operation data; the boiler air-coal ratio; the boiler air-coal ratio is the ratio of the difference in the amount of air fed into the furnace to the difference in the amount of coal fed into the furnace; B0-6. The average value of the data within the 25% quantile to the 75% quantile of the air-to-coal ratio of all boilers corresponding to the boiler operation data set Qa corresponding to time a is taken as the optimal air-to-coal ratio coefficient.

8. The boiler pressure control method based on the pressure prediction model according to claim 7, characterized in that: After step B0-6, the method further includes: B0-7, based on the characteristic change information between the adjacent b-th boiler operation data and the b-1-th boiler operation data in the boiler operation data set Qa corresponding to time a, calculate the corresponding modeling data R according to formula (A), formula (B), and formula (C) b ; The modeling data R b Includes: X b 、Y b , Z b ; Wherein, formula (A) is: dc b =coal b -coal b-1 ;dc b is the change in the amount of coal fed into the furnace at time b; dw b =wind b -wind b-1 ;dw b is the change in the air volume entering the furnace at time b; Wherein, formula (B) is: Y b =ds b ; ds b =steam b -steam b-1 ;ds b is the change in consumption flow at time b; Wherein, formula (C) is: Z b =dt b ; dt b =trend b -trend b-1 ;dt b The boiler pressure trend characteristics at time b and the boiler pressure trend characteristics at time b-1; B0-8. All N modeling data corresponding to the boiler operation data set Qa corresponding to time a are combined into a modeling data set; B0-9. Based on the modeling data set, the particle swarm optimization algorithm is used to optimize the loss function to obtain the model parameters in the loss function. The specific value M* of and the specific value e* of the flow-pressure transfer coefficient e; The loss function G is: In the particle swarm optimization algorithm, the inertia factor is greater than or equal to 0.1 and less than or equal to 0.9; the social factor is greater than or equal to 0.5 and less than or equal to 2.5; the individual factor is greater than or equal to 0.5 and less than or equal to 2.5; the population size is greater than or equal to 50 and less than or equal to 200; and the number of evolutions is greater than or equal to 100 and less than or equal to 5000.

Images