Thermal power generating unit main steam temperature control method

Abstract

The invention discloses a main steam temperature control method for a thermal power generating unit, and relates to the technical field of main steam temperature control, and the method comprises the steps: introducing a long-short term memory (LSTM) network model to carry out the dynamic prediction of the main steam temperature, learning the dynamic characteristics of a boiler through a time sequence, and achieving the advanced capture of a temperature change trend; the main steam temperature monitored in real time is compared with a prediction result, a deviation signal is generated to reflect the current temperature change trend, the steam temperature after secondary temperature reduction is fed back to the main controller, the deviation is corrected in real time through the PID controller, and the control precision of the main steam temperature is improved; and adjusting a control strategy for spraying the attemperation water based on the deviation signal, namely adding a dynamic compensation algorithm to enable the action of the attemperation valve to be ahead of the actual temperature change, flexibly adjusting the injection rate and flow of the attemperation water according to the temperature change trend, and formulating a special control strategy for the initial start-stop stage of the coal mill to improve the response speed of the attemperation water.

Description

Technical Field

[0001] The invention relates to the technical field of main steam temperature control, in particular to a main steam temperature control method for a thermal power unit. Background Art

[0002] During the operation of thermal power units, the main steam temperature directly affects the unit's operating efficiency and equipment safety. The pulverizer is the core equipment for boiler fuel supply. Its fuel supply status changes significantly during start-up and shutdown, which directly affects the boiler combustion conditions and further causes disturbances to the main steam temperature. This disturbance is often manifested as drastic fluctuations in the steam temperature before secondary desuperheating and the main steam temperature, further affecting the stability of turbine operation and equipment life.

[0003] Traditional main steam temperature control methods mostly use a cascade control system, combined with feedforward and feedback links to achieve regulation. The control logic takes the main steam temperature as the core target, relies on the action of the cooling water system, and provides a feedforward signal for the cooling water system by monitoring variables such as the steam temperature before cooling, coal feed rate and combustion status, and adjusts the controller output through the main steam temperature feedback signal; in the cascade control structure, the main loop is responsible for adjusting the overall temperature deviation, and the sub-loop is responsible for finely controlling the cooling water flow.

[0004] However, during the start-up and shutdown process of the coal mill, due to the drastic changes in the fuel supply of the combustion system and the boiler heat load, the boiler combustion dynamic response inertia is large, and the feedforward signal is difficult to reflect the disturbance in time; at the same time, traditional feedback control relies on temperature deviation action, has lag, and is difficult to cope with rapidly changing temperature disturbances; in addition, the boiler combustion process has significant nonlinearity and complex coupling relationships, and simple cascade control is difficult to achieve precise control. Although the response speed and robustness can be improved by increasing the source of feedforward signals and optimizing PID parameters, these improvements cannot fundamentally solve the problem of main steam temperature fluctuations when the coal mill is started and stopped. Therefore, a main steam temperature control solution for thermal power units is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The invention provides a main steam temperature control method for a thermal power unit to solve the problems of delayed response to combustion disturbance, insufficient control accuracy and insufficient dynamic performance of the traditional method.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The embodiment of the present invention provides a method for controlling the main steam temperature of a thermal power unit, which comprises:

[0009] Step S1, real-time collection of unit data, normalization of the data as input variables, combined with historical unit data, training of the long short-term memory network LSTM model to establish a main steam temperature dynamic prediction model;

[0010] The LSTM model learns the dynamic characteristics of the boiler through time series, predicts the trend of main steam temperature changes in advance, and obtains the prediction results;

[0011] Step S2, comparing the prediction result of step S1 with the main steam temperature monitored in real time to generate a deviation signal;

[0012] Step S3, adjusting the control strategy of spraying cooling water based on the deviation signal;

[0013] Step S4, assisting combustion control based on the prediction result of step S1.

[0014] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, wherein: the unit data includes coal feed amount, main steam pressure and steam temperature data before temperature reduction;

[0015] In step S1, the changes in inlet air volume and coal powder concentration when the coal mill is started and stopped are additionally monitored and input into the main steam temperature dynamic prediction model as dynamic variables.

[0016] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the step of combining historical unit data to train a long short-term memory network LSTM model to establish a main steam temperature dynamic prediction model is as follows:

[0017] Define the input variable of the LSTM model, the input variable is X t :

[0018] X t =(a t ,b t ,c t ),

[0019] Among them, X t represents the input variable at time t, a t Indicates the amount of coal fed, b t Indicates the main steam pressure, c t Indicates the steam temperature before desuperheating.

[0020] The hidden state update formula of the LSTM model is:

[0021] h t =σ(W h ·[h t-1 ,X t ]+b h ),

[0022] Among them, h t represents the hidden state at time t, σ represents the activation function, W h Represents the weight matrix of the hidden state, [h t-1 ,X t ] means concatenating the hidden state of the previous moment with the input variable of the current moment, b h represents the bias term of the hidden state,

[0023] The output variable of the LSTM model is y t :

[0024] y t =σ(W y ·h t +b y ),

[0025] Among them, y t represents the output variable at time t, i.e. the predicted value of the main steam temperature, W y represents the weight matrix of the output variable, b y represents the bias term of the output variable,

[0026] Define the loss function of the LSTM model. The loss function formula is:

[0027]

[0028] Among them, L represents the loss function, N represents the number of samples, and y t represents the predicted value at time t, represents the actual value at time t,

[0029] Combined with historical unit data, the LSTM model is trained with the following training formula:

[0030]

[0031] Among them, θ represents the parameters of the LSTM model, θ * represents the optimal parameter, L(θ) represents the loss function,

[0032] The LSTM model predicts the changing trend of the main steam temperature. The prediction formula is:

[0033] T pred =f(X future ),

[0034] Among them, T pred represents the predicted main steam temperature, f represents the LSTM model mapping function, X future represents the future input variable,

[0035] The specific form of the nonlinear mapping function is:

[0036] f(X)=σ(W·X+b),

[0037] Among them, W and b are weight matrix and bias term respectively, σ is the activation function, and ReLU and sigmoid functions are selected.

[0038] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, wherein: the deviation signal reflects the current temperature change trend and is used for dynamic compensation lag feedback control;

[0039] In step S2, the steam temperature after the secondary temperature reduction is fed back to the main controller to correct the deviation in real time.

[0040] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the step of comparing the prediction result of step S1 with the main steam temperature monitored in real time to generate a deviation signal is as follows:

[0041] Calculate the deviation value. The deviation signal calculation formula is:

[0042]

[0043] Among them, e t represents the deviation signal at time t, represents the actual main steam temperature at time t, y t represents the predicted main steam temperature at time t,

[0044] The steam temperature feedback after secondary desuperheating is T 2nd :

[0045] T 2nd =g(e t ,T 1st ),

[0046] Among them, T 2nd represents the steam temperature after secondary desuperheating, g represents the feedback function, e t Indicates the deviation signal, T 1st Indicates the steam temperature after the first stage of desuperheating.

[0047] The PID controller is used for control, and the output of the PID controller is:

[0048]

[0049] Among them, u t represents the output of the PID controller at time t, K p , K i , K d Respectively represent the proportional, integral, and differential coefficients, ei represents the deviation signal at time i, e t represents the deviation signal at time t,

[0050] Based on the controller output, the correction formula is:

[0051] T corrected =T 2nd +u t ,

[0052] Among them, T corrected Indicates the corrected main steam temperature, T 2nd Indicates the steam temperature after secondary desuperheating, u t represents the output of the PID controller,

[0053] Define the correction error, the error formula is:

[0054] ε t =T set -T corrected ,

[0055] Among them, ε t represents the correction error at time t, T set Indicates the set main steam temperature, T corrected Indicates the corrected main steam temperature,

[0056] The corrected control parameters are updated, and the update formula is:

[0057] θ ctrl =θ ctrl-1 +α·ε t ,

[0058] Among them, θ ctrl represents the corrected control parameter, θ ctrl-1 represents the control parameter before correction, α represents the learning rate, ε t Indicates the calibration error.

[0059] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the control strategy includes:

[0060] In the cooling water control loop, a dynamic compensation algorithm is added to make the cooling valve move ahead of the actual temperature change.

[0061] Adjust the desuperheating water injection rate and flow rate according to the temperature change trend to stabilize the main steam temperature near the set value.

[0062] At the initial stage of coal mill start-up and shutdown, the action of the steam injection desuperheating valve is controlled in advance according to the change of steam temperature before the secondary desuperheating predicted by the LSTM model;

[0063] The real-time correction in step S2 corrects the deviation by feeding back the steam temperature after the secondary desuperheating, with the purpose of compensating the error between the real-time monitoring value and the predicted value;

[0064] The control strategy of step S3 adjusts the cooling water injection based on the deviation signal, which is an optimization of the cooling water action, aiming to improve the dynamic response speed and regulation accuracy of the cooling water system.

[0065] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the step of adjusting the control strategy of spraying cooling water based on the deviation signal is as follows:

[0066] The cooling water injection rate is adjusted based on the deviation signal, and the adjustment formula is:

[0067]

[0068] Among them, v water represents the cooling water injection rate, k1 and k2 represent the proportional and differential coefficients respectively, and e t Indicates the deviation signal, de t / dt represents the rate of change of the deviation signal,

[0069] The cooling water flow rate is Q water :

[0070]

[0071] Among them, Q water represents the cooling water flow rate, v water (τ) represents the cooling water injection rate at time τ,

[0072] Perform dynamic compensation, the dynamic compensation amount is Δu:

[0073]

[0074] Among them, Δu represents the dynamic compensation amount, k3 represents the integral coefficient, e τ represents the deviation signal at time τ,

[0075] Adjust the warm water injection rate. After adjustment, the cooling water injection rate is:

[0076] v water-adj =v water +Δu,

[0077] Among them, v water-adj represents the adjusted cooling water injection rate, v water represents the original cooling water injection rate, Δu represents the dynamic compensation amount;

[0078] A special control strategy is formulated for the initial start-up and shutdown of the coal mill. The control method of this strategy is:

[0079] v water-init =f init (T pred-2nd ),

[0080] Among them, v water-init represents the cooling water injection rate at the initial stage of coal mill start-up and shutdown, f init represents the special control strategy function, T pred-2nd represents the predicted steam temperature before secondary desuperheating,

[0081] The limiting conditions for the desuperheating water injection rate are:

[0082] v water-final =min(max(v water-adj ,v min ),v max ),

[0083] Among them, v water-final represents the final cooling water injection rate, v water-adj represents the adjusted cooling water injection rate, v min 、v max They represent the minimum and maximum values ​​of the cooling water injection rate respectively.

[0084] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the auxiliary combustion control includes:

[0085] According to the combustion state changes during the start and stop of the coal mill, the working state of the coal feeding system and the air supply system are adjusted in a linked manner;

[0086] And adjust the fuel and air ratio.

[0087] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the step of assisting combustion control based on the prediction result of step S1 is:

[0088] The combustion state is adjusted based on the prediction results, and the adjustment formula is:

[0089] a coal =f coal (T pred ),

[0090] Among them, a coal Indicates the adjustment value of coal feed, f coal represents the coal feed adjustment function, T pred represents the predicted main steam temperature,

[0091] Adjust the air supply volume based on the prediction results. The adjustment formula is:

[0092] aair =f air (T pred ,a coal ),

[0093] Among them, a air Indicates the adjustment value of the air supply volume, f air represents the air supply volume adjustment function, T pred represents the predicted main steam temperature, a coal Indicates the adjustment value of coal feeding amount.

[0094] The adjustment formula for the fuel and air ratio is:

[0095]

[0096] Among them, r fuel-air represents the ratio of fuel to air, a coal Indicates the adjustment value of coal feeding amount, a air Indicates the adjustment value of the air supply volume, k ratio Indicates the ratio adjustment coefficient.

[0097] As a preferred solution of the main steam temperature control method of a thermal power unit described in the present invention, the step of assisting combustion control based on the prediction result of step S1 further includes:

[0098] The optimization target of combustion efficiency is defined as η opt :

[0099] η opt =max(η(r fuel - air )),

[0100] Among them, η opt represents the optimal combustion efficiency, η(r fuel - air ) represents the combustion efficiency function under the fuel and air ratio,

[0101] The formula for calculating combustion efficiency is:

[0102]

[0103] Where η represents the combustion efficiency, Q out Indicates the effective heat output of the boiler, Q in represents the input fuel heat,

[0104] The adjustment formula of boiler thermal efficiency is:

[0105] η boiler =η base +Δη,

[0106] Among them, ηboiler represents the adjusted boiler thermal efficiency, η base represents the basic boiler thermal efficiency, and Δη represents the adjustment amount of the thermal efficiency, which is based on the optimization result of the combustion efficiency.

[0107] The beneficial effects of the present invention are as follows: the present invention introduces a long short-term memory network LSTM model to dynamically predict the main steam temperature, learns the dynamic characteristics of the boiler through time series, and realizes the early capture of the temperature change trend; the real-time monitored main steam temperature is compared with the prediction result, and a deviation signal is generated to reflect the current temperature change trend. The steam temperature after secondary cooling is fed back to the main controller, and the PID controller is used to correct the deviation in real time to improve the control accuracy of the main steam temperature.

[0108] The present invention adjusts the control strategy for injecting cooling water based on the deviation signal, including adding a dynamic compensation algorithm to make the cooling valve act ahead of the actual temperature change, flexibly adjusting the cooling water injection rate and flow rate according to the temperature change trend, and formulating a special control strategy for the initial start-up and shutdown of the coal mill, thereby improving the cooling water response speed, effectively reducing the fluctuation of the main steam temperature during the start-up and shutdown of the coal mill, and improving the stability and safety of the unit operation.

[0109] The present invention makes full use of the prediction results to assist combustion control, jointly adjusts the working conditions of the coal feeding system and the air supply system, optimizes the ratio of fuel to air, realizes refined control of the combustion state, and converts the optimization results of the combustion efficiency into an improvement in the thermal efficiency of the boiler, thereby improving the energy efficiency of the entire boiler system.

[0110] In summary, the main steam temperature control method of a thermal power unit provided by the present invention significantly improves the temperature control accuracy and stability of the unit during the start-up and shutdown process of the coal mill, and provides a strong guarantee for the efficient and safe operation of the thermal power unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0111] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0112] Figure 1 The present invention is a flow chart of the main steam temperature control method of a thermal power unit. DETAILED DESCRIPTION

[0113] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0114] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0115] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0116] Example 1, reference Figure 1 This embodiment provides a method for controlling the main steam temperature of a thermal power unit, comprising:

[0117] Step S1, real-time collection of unit data, normalization of the data as input variables, combined with historical unit data, training of the long short-term memory network LSTM model to establish a main steam temperature dynamic prediction model;

[0118] The LSTM model learns the dynamic characteristics of the boiler through time series, predicts the trend of main steam temperature changes in advance, and obtains the prediction results;

[0119] Unit data include coal feed rate, main steam pressure and steam temperature before desuperheating;

[0120] In step S1, the inlet air volume and coal powder concentration changes during the start and stop of the coal mill are additionally monitored and input into the main steam temperature dynamic prediction model as dynamic variables;

[0121] Combined with historical unit data, the steps of training the long short-term memory network LSTM model to establish the main steam temperature dynamic prediction model are as follows:

[0122] Define the input variable of the LSTM model, the input variable is X t :

[0123] X t =(a t ,b t ,c t ),

[0124] Among them, X t represents the input variable at time t, a t Indicates the amount of coal fed, b t Indicates the main steam pressure, c t Indicates the steam temperature before desuperheating.

[0125] The hidden state update formula of the LSTM model is:

[0126] h t =σ(W h ·[h t-1 ,X t ]+b h ),

[0127] Among them, h t represents the hidden state at time t, σ represents the activation function, W h Represents the weight matrix of the hidden state, [h t-1 ,X t ] means concatenating the hidden state of the previous moment with the input variable of the current moment, b h represents the bias term of the hidden state,

[0128] The output variable of the LSTM model is y t :

[0129] y t =σ(W y ·h t +b y ),

[0130] Among them, y t represents the output variable at time t, i.e. the predicted value of the main steam temperature, W y represents the weight matrix of the output variable, b y represents the bias term of the output variable,

[0131] Define the loss function of the LSTM model. The loss function formula is:

[0132]

[0133] Among them, L represents the loss function, N represents the number of samples, and y t represents the predicted value at time t, represents the actual value at time t,

[0134] Combined with historical unit data, the LSTM model is trained with the following training formula:

[0135]

[0136] Among them, θ represents the parameters of the LSTM model, θ * represents the optimal parameter, L(θ) represents the loss function,

[0137] The LSTM model predicts the changing trend of the main steam temperature. The prediction formula is:

[0138] T pred =f(X future ),

[0139] Among them, Tpred represents the predicted main steam temperature, f represents the LSTM model mapping function, X future represents the future input variable,

[0140] The specific form of the nonlinear mapping function is:

[0141] f(X)=σ(W·X+b),

[0142] Among them, W and b are weight matrix and bias term respectively, σ is activation function, ReLU and sigmoid functions are selected;

[0143] Specifically, step S1 defines the input variables, hidden state updates, output variables, and loss functions of the LSTM model, and trains the LSTM model in combination with historical unit data so that it can learn the dynamic characteristics of the boiler and predict the changing trend of the main steam temperature.

[0144] Step S2, comparing the prediction result of step S1 with the main steam temperature monitored in real time to generate a deviation signal;

[0145] The deviation signal reflects the current temperature change trend and is also used for dynamic compensation of hysteresis feedback control;

[0146] In step S2, the steam temperature after the secondary desuperheating is fed back to the main controller to correct the deviation in real time;

[0147] The prediction result of step S1 is compared with the main steam temperature monitored in real time, and the step of generating a deviation signal is as follows:

[0148] Calculate the deviation value. The deviation signal calculation formula is:

[0149]

[0150] Among them, e t represents the deviation signal at time t, represents the actual main steam temperature at time t, y t represents the predicted main steam temperature at time t,

[0151] The steam temperature feedback after secondary desuperheating is T 2nd :

[0152] T 2nd =g(e t ,T 1st ),

[0153] Among them, T 2nd represents the steam temperature after secondary desuperheating, g represents the feedback function, e t Indicates the deviation signal, T 1st Indicates the steam temperature after the first stage of desuperheating.

[0154] The PID controller is used for control, and the output of the PID controller is:

[0155]

[0156] Among them, u t represents the output of the PID controller at time t, K p , K i , K d Respectively represent the proportional, integral, and differential coefficients, e i represents the deviation signal at time i, e t represents the deviation signal at time t,

[0157] Based on the controller output, the correction formula is:

[0158] T corrected =T 2nd +u t ,

[0159] Among them, T corrected Indicates the corrected main steam temperature, T 2nd Indicates the steam temperature after secondary desuperheating, u t represents the output of the PID controller,

[0160] Define the correction error, the error formula is:

[0161] ε t =T set -T corrected ,

[0162] Among them, ε t represents the correction error at time t, T set Indicates the set main steam temperature, T corrected Indicates the corrected main steam temperature,

[0163] The corrected control parameters are updated, and the update formula is:

[0164] θ ctrl =θ ctrl-1 +α·ε t ,

[0165] Among them, θ ctrl represents the corrected control parameter, θ ctrl-1 represents the control parameter before correction, α represents the learning rate, ε t represents the correction error;

[0166] Specifically, step S2 calculates the deviation signal, feeds back the steam temperature after secondary cooling to the main controller, uses the PID controller to correct the deviation in real time, obtains the corrected main steam temperature, realizes dynamic adjustment of the control parameters, and improves the control accuracy of the main steam temperature.

[0167] Step S3, adjusting the control strategy of spraying cooling water based on the deviation signal;

[0168] Control strategies include:

[0169] In the cooling water control loop, a dynamic compensation algorithm is added to make the cooling valve move ahead of the actual temperature change.

[0170] Adjust the desuperheating water injection rate and flow rate according to the temperature change trend to stabilize the main steam temperature near the set value.

[0171] At the initial stage of coal mill start-up and shutdown, the action of the steam injection desuperheating valve is controlled in advance according to the change of steam temperature before the secondary desuperheating predicted by the LSTM model;

[0172] The real-time correction in step S2 corrects the deviation by feeding back the steam temperature after the secondary desuperheating, with the purpose of compensating the error between the real-time monitoring value and the predicted value;

[0173] The control strategy of step S3 adjusts the cooling water injection based on the deviation signal, which is an optimization of the cooling water action, aiming to improve the dynamic response speed and adjustment accuracy of the cooling water system;

[0174] Based on the deviation signal, the steps to adjust the control strategy of spraying cooling water are as follows:

[0175] The cooling water injection rate is adjusted based on the deviation signal, and the adjustment formula is:

[0176]

[0177] Among them, v water represents the cooling water injection rate, k1 and k2 represent the proportional and differential coefficients respectively, and e t Indicates the deviation signal, de t / dt represents the rate of change of the deviation signal,

[0178] The cooling water flow rate is Q water :

[0179]

[0180] Among them, Q water represents the cooling water flow rate, v water (τ) represents the cooling water injection rate at time τ,

[0181] Perform dynamic compensation, the dynamic compensation amount is Δu:

[0182]

[0183] Among them, Δu represents the dynamic compensation amount, k3 represents the integral coefficient, e τ represents the deviation signal at time τ,

[0184] Adjust the warm water injection rate. After adjustment, the cooling water injection rate is:

[0185] v water-adj =v water +Δu,

[0186] Among them, v water-adj represents the adjusted cooling water injection rate, v water represents the original cooling water injection rate, Δu represents the dynamic compensation amount;

[0187] A special control strategy is formulated for the initial start-up and shutdown of the coal mill. The control method of this strategy is:

[0188] v water-init =f init (T pred-2nd ),

[0189] Among them, v water-init represents the cooling water injection rate at the initial stage of coal mill start-up and shutdown, f init represents the special control strategy function, T pred-2nd represents the predicted steam temperature before secondary desuperheating,

[0190] The limiting conditions for the desuperheating water injection rate are:

[0191] v water-final =min(max(v water-adj ,v min ),v max ),

[0192] Among them, v water-final represents the final cooling water injection rate, v water-adj represents the adjusted cooling water injection rate, v min 、v max Respectively represent the minimum and maximum values ​​of the cooling water injection rate,

[0193] Specifically, step S3 defines the cooling water injection rate and cooling water flow rate based on the deviation signal, performs dynamic compensation, and dynamically adjusts the cooling water injection rate; at the same time, a special control strategy is defined for the special situation in the initial start-up and shutdown of the coal mill to always keep the cooling water injection rate within a reasonable range.

[0194] Step S4, assisting combustion control based on the prediction result of step S1;

[0195] Auxiliary combustion controls include:

[0196] According to the combustion state changes during the start and stop of the coal mill, the working state of the coal feeding system and the air supply system are adjusted in a linked manner;

[0197] and adjusting the fuel and air ratio;

[0198] The steps of assisting combustion control based on the prediction result of step S1 are:

[0199] The combustion state is adjusted based on the prediction results, and the adjustment formula is:

[0200] a coal =f coal (T pred ),

[0201] Among them, a coal Indicates the adjustment value of coal feed, f coal represents the coal feed adjustment function, T pred represents the predicted main steam temperature,

[0202] Adjust the air supply volume based on the prediction results. The adjustment formula is:

[0203] a air =f air (T pred ,a coal ),

[0204] Among them, a air Indicates the adjustment value of the air supply volume, f air represents the air supply volume adjustment function, T pred represents the predicted main steam temperature, a coal Indicates the adjustment value of coal feeding amount.

[0205] The adjustment formula for the fuel and air ratio is:

[0206]

[0207] Among them, r fuel-air represents the ratio of fuel to air, a coal Indicates the adjustment value of coal feeding amount, a air Indicates the adjustment value of the air supply volume, k ratio Represents the ratio adjustment coefficient,

[0208] The step of assisting combustion control based on the prediction result of step S1 also includes:

[0209] The optimization target of combustion efficiency is defined as η opt :

[0210] ηopt =max(η(r fuel - air )),

[0211] Among them, η opt represents the optimal combustion efficiency, η(r fuel-air) represents the combustion efficiency function under the fuel and air ratio,

[0212] The formula for calculating combustion efficiency is:

[0213]

[0214] Where η represents the combustion efficiency, Q out Indicates the effective heat output of the boiler, Q in represents the input fuel heat,

[0215] The adjustment formula of boiler thermal efficiency is:

[0216] η boiler =η base +Δη,

[0217] Among them, η boiler represents the adjusted boiler thermal efficiency, η base represents the basic boiler thermal efficiency, Δη represents the adjustment of thermal efficiency, which is based on the optimization result of combustion efficiency;

[0218] Specifically, step S4 performs fine control on the combustion state of the boiler, converts the optimization result of the combustion efficiency into an improvement in the thermal efficiency of the boiler, and improves the energy efficiency of the entire boiler system.

[0219] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for controlling the main steam temperature of a thermal power unit, characterized in that: include, Step S1, real-time collection of unit data, normalization of the data as input variables, combined with historical unit data, training the long short-term memory network LSTM model to establish a main steam temperature dynamic prediction model; The LSTM model learns the dynamic characteristics of the boiler through time series, predicts the trend of main steam temperature changes in advance, and obtains the prediction results; Step S2, comparing the prediction result of step S1 with the main steam temperature monitored in real time to generate a deviation signal; Step S3, adjusting the control strategy of spraying cooling water based on the deviation signal; Step S4, assisting combustion control based on the prediction result of step S1.

2. A method for controlling the main steam temperature of a thermal power unit according to claim 1, characterized in that: The unit data include coal feed rate, main steam pressure and steam temperature data before temperature reduction; In step S1, the changes in inlet air volume and pulverized coal concentration during the start and stop of the coal mill are additionally monitored and input into the main steam temperature dynamic prediction model as dynamic variables.

3. A method for controlling the main steam temperature of a thermal power unit according to claim 2, characterized in that: The steps of combining historical unit data and training the long short-term memory network LSTM model to establish a main steam temperature dynamic prediction model are: Define the input variable of the LSTM model, the input variable is X t : X t =(a t ,b t ,c t ), Among them, X t represents the input variable at time t, a t Indicates the amount of coal fed, b t Indicates the main steam pressure, c t Indicates the steam temperature before desuperheating. The hidden state update formula of the LSTM model is: h t =σ(W h ·[h t-1 ,X t ]+b h ), Among them, h t represents the hidden state at time t, σ represents the activation function, W h The weight matrix representing the hidden state, [h t-1 ,X t ] means concatenating the hidden state of the previous moment with the input variable of the current moment, b h represents the bias term of the hidden state, The output variable of the LSTM model is y t : y t =σ(Wx h t +b y ), Among them, y t represents the output variable at time t, i.e. the predicted value of the main steam temperature, W y represents the weight matrix of the output variable, b y represents the bias term of the output variable, Define the loss function of the LSTM model. The loss function formula is: Among them, L represents the loss function, N represents the number of samples, and y t represents the predicted value at time t, represents the actual value at time t, Combined with historical unit data, the LSTM model is trained with the following training formula: Among them, θ represents the parameters of the LSTM model, θ * represents the optimal parameter, L(θ) represents the loss function, The LSTM model predicts the changing trend of the main steam temperature. The prediction formula is: T pred =f(X future ), Among them, T pred represents the predicted main steam temperature, f represents the LSTM model mapping function, X future represents the future input variable, The specific form of the nonlinear mapping function is: f(X)=σ(W·X+b), Among them, W and b are weight matrix and bias term respectively, σ is the activation function, and ReLU and sigmoid functions are selected.

4. A method for controlling the main steam temperature of a thermal power unit according to claim 3, characterized in that: The deviation signal reflects the current temperature change trend and is also used for dynamic compensation of hysteresis feedback control; In step S2, the steam temperature after the secondary temperature reduction is fed back to the main controller to correct the deviation in real time.

5. A method for controlling the main steam temperature of a thermal power unit according to claim 4, characterized in that: The step of comparing the prediction result of step S1 with the main steam temperature monitored in real time to generate a deviation signal is: Calculate the deviation value. The deviation signal calculation formula is: Among them, e t represents the deviation signal at time t, represents the actual main steam temperature at time t, y t represents the predicted main steam temperature at time t, The steam temperature feedback after secondary desuperheating is T 2nd : T 2nd =g(e t ,T 1st ), Among them, Y 2nd represents the steam temperature after secondary desuperheating, g represents the feedback function, e t Indicates the deviation signal, T 1st Indicates the steam temperature after the first stage of desuperheating. The PID controller is used for control, and the output of the PID controller is: Among them, u t represents the output of the PID controller at time t, K p , K i , K d Respectively represent the proportional, integral, and differential coefficients, e i represents the deviation signal at time i, e t represents the deviation signal at time t, Based on the controller output, the correction formula is: T corrected =T 2nd +u t , Among them, T corrected Indicates the corrected main steam temperature, T 2nd Indicates the steam temperature after secondary desuperheating, u t represents the output of the PID controller, Define the correction error, the error formula is: ε t =T set -T corrected , Among them, ε t represents the correction error at time t, T set Indicates the set main steam temperature, T corrected Indicates the corrected main steam temperature, The corrected control parameters are updated, and the update formula is: i ctrl =θ ctrl-1 +a·e t , Among them, θ ctrl represents the corrected control parameter, θ ctrl-1 represents the control parameter before correction, α represents the learning rate, ε t Indicates the calibration error.

6. A method for controlling the main steam temperature of a thermal power unit according to claim 5, characterized in that: The control strategy includes: In the cooling water control loop, a dynamic compensation algorithm is added to make the cooling valve move ahead of the actual temperature change. Adjust the desuperheating water injection rate and flow rate according to the temperature change trend to stabilize the main steam temperature near the set value. At the initial stage of coal mill start-up and shutdown, the action of the steam injection cooling valve is controlled in advance according to the change of steam temperature before secondary cooling predicted by the LSTM model.

7. A method for controlling the main steam temperature of a thermal power unit according to claim 6, characterized in that: The step of adjusting the control strategy of spraying cooling water based on the deviation signal is as follows: The cooling water injection rate is adjusted based on the deviation signal, and the adjustment formula is: Among them, v water represents the cooling water injection rate, k1 and k2 represent the proportional and differential coefficients respectively, and e t Indicates the deviation signal, de t / dt represents the rate of change of the deviation signal, The cooling water flow rate is Q water : Among them, Q water represents the cooling water flow rate, v water (τ) represents the cooling water injection rate at time τ, Perform dynamic compensation, the dynamic compensation amount is Δu: Among them, Δu represents the dynamic compensation amount, k3 represents the integral coefficient, e τ represents the deviation signal at time τ, Adjust the warm water injection rate. After adjustment, the cooling water injection rate is: in water-adj =in water +Δu, Among them, v water-adj represents the adjusted cooling water injection rate, v water represents the original cooling water injection rate, Δu represents the dynamic compensation amount; A special control strategy is formulated for the initial start-up and shutdown of the coal mill. The control method of this strategy is: v water-init =f init (T pred-2nd ), Among them, v water-init represents the cooling water injection rate at the initial stage of coal mill start-up and shutdown, f init represents the special control strategy function, T pred-2nd represents the predicted steam temperature before secondary desuperheating, The limiting conditions for the desuperheating water injection rate are: v water-final =min(max(v water-adj ,v min ),v max ), Among them, v water-final represents the final cooling water injection rate, v water-adj represents the adjusted cooling water injection rate, v min 、v max They represent the minimum and maximum values ​​of the cooling water injection rate respectively.

8. A method for controlling the main steam temperature of a thermal power unit according to claim 7, characterized in that: The auxiliary combustion control includes: According to the combustion state changes during the start and stop of the coal mill, the working state of the coal feeding system and the air supply system are adjusted in a linked manner; And adjust the fuel and air ratio.

9. A method for controlling the main steam temperature of a thermal power unit according to claim 8, characterized in that: The step of assisting combustion control based on the prediction result of step S1 is: The combustion state is adjusted based on the prediction results, and the adjustment formula is: a coal =f coal (T pred ), Among them, a coal Indicates the adjustment value of coal feed, f coal represents the coal feed adjustment function, T pred represents the predicted main steam temperature, Adjust the air supply volume based on the prediction results. The adjustment formula is: a air =f air (T pred ,a coal ), Among them, a air Indicates the adjustment value of the air supply volume, f air represents the air supply volume adjustment function, T pred represents the predicted main steam temperature, a coal Indicates the adjustment value of coal feeding amount. The adjustment formula for the fuel and air ratio is: Among them, r fiel-air represents the ratio of fuel to air, a coal Indicates the adjustment value of coal feeding amount, a air Indicates the adjustment value of the air supply volume, k ratio Indicates the ratio adjustment coefficient.

10. A method for controlling the main steam temperature of a thermal power unit according to claim 9, characterized in that: The step of assisting combustion control based on the prediction result of step S1 further includes: The optimization target of combustion efficiency is defined as η opt : or opt =max(η(r fuel-air )), Among them, η opt represents the optimal combustion efficiency, η(r fuel-air ) represents the combustion efficiency function under the fuel and air ratio, The formula for calculating combustion efficiency is: Where η represents the combustion efficiency, Q out Indicates the effective heat output of the boiler, Q in represents the input fuel heat, The adjustment formula of boiler thermal efficiency is: or boiler =the base +D, Among them, η boiler represents the adjusted boiler thermal efficiency, η base represents the basic boiler thermal efficiency, and Δη represents the adjustment amount of the thermal efficiency, which is based on the optimization result of the combustion efficiency.

Images